â—† AI Article Pricer Compare AI article generation costs
How-to guide · generated 2026-08-20 · unedited

What a 1,500-Word AI Article Actually Looks Like

The same brief went to 5 models, from $0.00321 a draft to $0.1173 — a 37× price difference at current list rates applied to recorded usage. Read blind, or jump straight to the comparison.

Every other page here tells you what an article costs. None of them tell you what that money buys. So: 5 drafts of a how-to guide for a non-technical reader, one commission, nothing edited, the model and the price hidden behind a click on each one.

The original drafts are unedited No typo fixed, no preamble trimmed, no generation re-rolled. Raw output as it came back from each API on .
What this experiment shows

The findings before the full drafts

The same 1,500-word how-to guide brief produced 5 drafts ranging from 1,506 to 2,908 words. This is one published run, not a general model ranking. Model identities stay hidden here.

  • Only one draft hit the commissioned length. The brief asked for 1,500 words. Draft C delivered 1,506; Draft E nearly doubled the commission at 2,908. The overshoot is billed at the same rate as the words you wanted, and an editor is then paid to cut it — you pay for the excess twice.
  • One draft opens with exactly the banned throat-clearing. The brief prohibited opening on the importance of content. Draft E opens with "your website is the cornerstone of your B2B marketing efforts" — the sentence the brief exists to prevent. The other four start from the reader’s problem, as asked.

Compare model identities, usage and costs · Read all editorial observations · Understand estimate accuracy

Compare the opening paragraph of each draft (blind)

Draft A

You have 40–80 published pages, limited time, and probably a mix of product pages, blog posts, comparison pages, gated asset landing pages, and a few old announcements nobody has looked at in years. A content audit is a structured review of those pages so you can decide what to keep, improve, combine, or remove. For a small B2B software site, the goal is not to create a perfect content database. The goal is to make better decisions about pages that affect pipeline, search visibility, buyer trust, and your team’s limited attention.

Draft B

You have somewhere between 40 and 80 pages live, and you cannot confidently say what most of them do. Some were written by a founder in 2021. Some were commissioned from a freelancer whose name you have forgotten. Two of them rank for something, though you are not sure which two. When someone asks whether you should write more, you have no defensible answer, because you have no inventory of what already exists.

Draft C

You have somewhere between 40 and 80 pages live on your website, and you genuinely don't know what half of them are doing for you. Some are probably ranking for keywords you forgot you targeted. Some are getting traffic but no conversions. Some haven't been touched since a product launch two years ago and reference a feature you've since renamed. You have two days and a spreadsheet. That's enough.

Draft D

You have 40–80 published pages, you’ve never done this before, you’re working with a spreadsheet and free analytics, and you’ve got two days to make sense of it all. Here’s a practical, no-frills approach that fits your constraints and helps you decide what to keep, update, merge, or delete, then track progress three months down the line.

Draft E

You know your website is the cornerstone of your B2B marketing efforts, but with 40 to 80 pages of content, it's easy for some pieces to get lost, become outdated, or simply not pull their weight. You need to ensure every page contributes to your goals, engages your audience, and justifies its place on your site. This guide will walk you through a practical, no-cost content audit you can complete in about two working days, using just a spreadsheet and your free analytics.

The commission

How-to / procedural

A sample without its brief is unreadable as evidence — you cannot judge a draft without knowing what was asked for. This is the exact prompt every model received, with no other instruction beyond a system line telling it to return the article and nothing else.

Read the full brief every model was given

Write a 1,500-word how-to article titled "How to Run a Content Audit on a Small B2B Website".

The reader runs marketing at a B2B software company with 40–80 published pages. They have never done a content audit, have no budget for tooling beyond a spreadsheet and free-tier analytics, and have about two working days to spend on this.

Cover, in this order:

  1. What a content audit is and what it is not, in two paragraphs
  2. What to collect for each page, as a concrete list of spreadsheet columns
  3. How to decide what to keep, update, merge or delete — give the reader a rule for each of the four outcomes, not just a description of them
  4. What to do first once the decisions are made
  5. How to tell three months later whether it worked

Requirements:

  • Second person, addressed to the reader directly
  • Every recommendation must be actionable with a spreadsheet and free-tier analytics
  • Include at least one worked example with specific numbers
  • No tool recommendations by brand name
  • Do not open with a definition of content marketing or any throat-clearing about the importance of content — start with the reader's problem

System prompt: “You are a professional writer producing a draft for an editor. Follow the brief exactly, including its structure and length. Return only the article itself in Markdown — no preamble, no commentary on the brief, no notes to the editor.”

Blind first

5 drafts, in a random order

Read at least two before you reveal anything. Knowing the price changes what you see in the prose — that is exactly the bias this page exists to remove. The letters were assigned from the run's own seed before generation started, so the order is not ours to arrange.
DraftWordsvs the 1,500-word commissionReading timeState
Draft A 2,709 +81% ~12 min Complete
Draft B 1,665 +11% ~7 min Complete
Draft C 1,506 +0% ~7 min Complete
Draft D 2,369 +58% ~10 min Complete
Draft E 2,908 +94% ~13 min Complete

The at-a-glance row carries no model and no price — that stays behind each draft's reveal. Overshoot is billed at the same rate as the words that were asked for.

Draft A

2,709 words

Reveal the model and what it cost

GPT-5.5 OpenAI

Recorded usage at current rates
$0.1173
Output tokens
3,856
Tokens per word
1.42
Calculator quoted
$0.1076
Price this model for your own volume →

How to Run a Content Audit on a Small B2B Website

You have 40–80 published pages, limited time, and probably a mix of product pages, blog posts, comparison pages, gated asset landing pages, and a few old announcements nobody has looked at in years. A content audit is a structured review of those pages so you can decide what to keep, improve, combine, or remove. For a small B2B software site, the goal is not to create a perfect content database. The goal is to make better decisions about pages that affect pipeline, search visibility, buyer trust, and your team’s limited attention.

A content audit is not a full rebrand, a writing project, or a forensic investigation into every historical marketing decision. You are not trying to rewrite the whole site in two days. You are creating a practical inventory, adding enough performance and business context to each URL, and assigning a next action. If a page is clearly doing its job, leave it alone. If it is weak but strategically important, improve it. If it overlaps with another page, merge it. If it is outdated, irrelevant, and unused, delete it or redirect it.

Read the rest of Draft A

What to Collect for Each Page

Start with one spreadsheet. Each row is one URL. If your site has 40–80 pages, you can build the inventory manually from your CMS page list, XML sitemap, navigation, blog index, and analytics landing page reports.

Use the last 90 days of data where possible. If your sales cycle is long or traffic is low, also look at the last 12 months for context, but make decisions from the most recent 90 days unless seasonality is obvious.

Create these columns:

ColumnWhat to enter
URLThe full page address
Page titleThe visible or browser title
Page typeProduct, solution, blog, case study, comparison, landing page, pricing, support, company, etc.
Funnel stageAwareness, consideration, decision, customer, or internal/other
Target audienceBuyer, user, technical evaluator, partner, investor, candidate, customer
Primary topicThe main subject of the page
Target keyword or queryThe main search phrase, if any
Search intentInformational, commercial, transactional, navigational, or not search-focused
Date publishedIf known
Date last updatedIf known
OwnerPerson or team responsible
90-day organic clicksFrom free search performance data
90-day impressionsFrom free search performance data
Average search positionFor the main queries, if available
90-day sessions or visitsFrom free web analytics
90-day entrancesHow often the page was the first page visited
Engagement signalUse average engagement time, bounce rate, or another available metric
90-day conversionsForm fills, demo requests, trial starts, newsletter signups, or relevant events
Assisted value notesAny known sales use, internal linking value, or executive importance
Backlinks or referring domainsOnly if available in free data; otherwise mark “unknown”
Internal links inCount obvious links from navigation, related posts, product pages, or manually note “low/medium/high”
Internal links outKey pages this page links to
Content quality score1–5, based on accuracy, depth, clarity, and usefulness
Brand/message fit score1–5, based on whether it reflects your current positioning
Business importance score1–5, based on relevance to revenue, product strategy, or buyer needs
Duplicate/overlap URLAny page that covers the same topic
Recommended actionKeep, update, merge, or delete
PriorityHigh, medium, low
NotesAnything that explains the decision

Do not overcomplicate the scoring. Use this simple scale:

  • 5 = strong, current, useful, clearly aligned
  • 3 = acceptable but could be better
  • 1 = outdated, thin, inaccurate, off-message, or unhelpful

For conversions, choose the events that matter for your site. On a B2B software site, this might include demo requests, contact form submissions, free trial starts, pricing page clicks, webinar registrations, or content download forms. If your analytics setup is incomplete, add a note. Do not stop the audit because the data is imperfect.

To save time, collect in this order:

  1. Export or copy your URL list.
  2. Add page type, funnel stage, topic, and audience manually.
  3. Add traffic and conversion data from free analytics.
  4. Add search clicks, impressions, and average position from free search performance reporting.
  5. Score content quality, brand fit, and business importance.
  6. Assign an action.

For 40–80 pages, this should fit into two working days: half a day to build the inventory, half a day to add performance data, half a day to score and decide, and half a day to review priorities with whoever owns sales, product marketing, or demand generation.

How to Decide What to Keep, Update, Merge, or Delete

Every page should receive one of four actions. Do not create vague labels like “review later” unless you also assign a real action. The point of the audit is to reduce ambiguity.

Keep

Rule: Keep a page if it is accurate, aligned with your current positioning, and either produces measurable value or serves a necessary business purpose.

A page qualifies for “keep” if it meets at least two of these conditions:

  • It generated at least one meaningful conversion in the last 90 days.
  • It receives steady qualified traffic for its role, even if the number is modest.
  • It supports a key sales or product narrative.
  • It ranks or appears for relevant search queries.
  • It is required for trust, such as pricing, security, integrations, case studies, or company information.
  • Its content quality score and brand fit score are both 4 or 5.

A keep decision does not mean the page is perfect. It means you should not spend your limited two-day audit budget fixing it now. Add minor issues to the notes column, but do not let small edits distract you from pages with bigger upside.

Example: Your “Security” page had 180 sessions in the last 90 days, only one form submission, and a high average engagement time. Sales says prospects often ask for it during procurement. Its content quality score is 4 and business importance is 5. Keep it, even if it is not a major traffic driver.

Update

Rule: Update a page if it is strategically important but underperforming, outdated, incomplete, or misaligned.

A page should be marked “update” if it meets one or more of these conditions:

  • Business importance is 4 or 5, but content quality or brand fit is 3 or lower.
  • It gets search impressions but has weak clicks, suggesting the title, angle, or relevance may be poor.
  • It gets traffic but no conversions, and the page should be conversion-oriented.
  • It covers a current product, use case, or buyer problem but includes outdated messaging, old screenshots, weak proof, or missing calls to action.
  • It ranks near the bottom of page one or on page two for a relevant query and could plausibly improve with better content.

Use the spreadsheet to identify update candidates with high upside. Sort by business importance first, then impressions, then conversions or entrances. A page with 2,000 impressions and 20 clicks may have more upside than a page with 30 impressions and 2 clicks, assuming the topic matters.

Worked example:

You have a blog post titled “How to Automate Vendor Onboarding.” In the last 90 days it had:

  • 1,200 organic impressions
  • 48 organic clicks
  • Average search position of 9.8
  • 70 total sessions
  • 0 demo requests
  • Content quality score of 3
  • Brand fit score of 2
  • Business importance score of 5

This is an update, not a delete. The topic matches your product, searchers are seeing it, and it sits close enough to better positions that improvement is realistic. Your update should include current product messaging, a clearer explanation of the buyer pain, examples, internal links to the relevant solution page, and a call to action such as “See how this works in a demo.”

Merge

Rule: Merge pages when two or more URLs target the same audience, intent, and topic, and none of them is clearly strong enough to stand alone.

Mark pages for merge if:

  • They answer the same question with similar content.
  • They target the same keyword or search intent.
  • They split traffic, links, or conversions across multiple weak pages.
  • One page is a thin older version of a newer, better page.
  • Sales or buyers would not understand why both pages exist.

For each merge group, choose one destination URL. Usually, keep the URL with the best combination of traffic, search visibility, backlinks if known, and strategic fit. Move the best sections from the weaker pages into the destination page. Then redirect the old URLs to the chosen destination.

Example: You have three posts:

  • /blog/vendor-onboarding-checklist with 90 sessions and 0 conversions
  • /blog/vendor-onboarding-process with 75 sessions and 1 conversion
  • /resources/vendor-onboarding-template with 40 sessions and 0 conversions

All three serve the same audience and intent: people trying to improve vendor onboarding. The process post has the only conversion and the clearest fit with your product. Merge the checklist and template content into the process page, make it the strongest page, and redirect the other two URLs to it.

Delete

Rule: Delete a page only if it has no meaningful traffic, no conversions, no strategic value, and no useful content worth merging.

Use deletion carefully. In most cases, you should redirect rather than simply remove. Mark a page “delete” if it meets all of these conditions:

  • It had zero or near-zero sessions in the last 90 days.
  • It had zero conversions.
  • It has no relevant search impressions.
  • It is outdated, inaccurate, off-brand, or no longer related to your product.
  • It has no obvious internal or external value.
  • There is no better page where the content should be merged.

Good delete candidates include old hiring announcements, obsolete event pages, outdated feature pages for retired products, thin news posts, and landing pages for campaigns that ended years ago.

If a deleted page has a close replacement, redirect it. If it has no replacement and no traffic, returning a proper “not found” status is acceptable. Keep a record of every deleted URL and what happened to it.

What to Do First Once the Decisions Are Made

Once every row has an action, do not start editing randomly. Sort your spreadsheet by recommended action and priority. Your first job is to protect what is already working, then create the biggest gains with the least effort.

Start with these steps:

  1. Review all delete and merge decisions before touching the site. Check them with one other person if possible, especially someone in sales, product marketing, or customer success. You are looking for hidden value: a page sales uses, a partner page, a legal requirement, or a campaign URL still in use.
  1. Create a redirect map. Add two columns: “Old URL” and “Redirect target.” For every merged or deleted page with a relevant replacement, list where it should go. The redirect target should be the closest useful page, not always the homepage.
  1. Handle merges before deletions. If you delete a page before moving its useful content, you may lose examples, explanations, or search relevance. For each merge group, copy the best material into the destination page, improve the structure, add internal links, then redirect the weaker URLs.
  1. Update high-priority pages next. Prioritize pages with high business importance and evidence of demand. A page with 1,500 impressions, mediocre rankings, and a strong connection to your product should come before a low-traffic opinion post.
  1. Add or improve calls to action. For every updated or merged page, decide the next step you want the reader to take. Link to a demo, pricing page, product page, case study, comparison page, or relevant resource. Do not rely on the main navigation alone.
  1. Strengthen internal links. Link from high-traffic informational pages to decision-stage pages. Link from product and solution pages to relevant proof, such as case studies or technical pages. Keep anchor text descriptive.
  1. Record completion dates. Add a “Completed date” column and a “Change summary” column. Three months later, you will need to know what changed and when.

If you only have time to act on five pages after the audit, choose the five with the highest combination of business importance and opportunity. Use a simple formula: add business importance score, content quality gap, and demand signal.

For example:

  • Business importance: 5
  • Content quality gap: 2 points below ideal
  • Demand signal: 3, because it has more than 1,000 impressions

Total priority score: 10

You do not need a perfect model. You need a consistent way to avoid spending your first effort on low-impact pages.

How to Tell Three Months Later Whether It Worked

Before you make changes, save a copy of the audit spreadsheet as your baseline. Three months later, compare the same metrics over the same length of time: the 90 days before implementation versus the 90 days after implementation. Do not compare a full quarter to two weeks of data.

Track results at three levels: site, page group, and individual page.

At the site level, look at:

  • Organic clicks
  • Organic impressions
  • Total sessions
  • Entrances from organic search
  • Demo requests, trial starts, or other primary conversions
  • Conversion rate from content pages
  • Number of indexed or discoverable pages, if available from free search reporting

At the page group level, compare:

  • Updated pages versus their previous performance
  • Merged pages as a group, combining old and new URL performance
  • Deleted pages to confirm they did not remove valuable traffic
  • Kept pages to make sure they remained stable

At the individual page level, focus on whether the intended job improved. For an awareness article, organic clicks and engaged visits may matter most. For a solution page, conversions and clicks to demo or pricing matter more. For a security page, engagement and assisted sales value may be more meaningful than raw traffic.

Use a simple before-and-after table:

Metric90 days before90 days afterChange
Organic clicks to updated pages620810+31%
Demo requests from updated pages611+83%
Organic impressions for merged topic3,4004,100+21%
Sessions to deleted URLs140No issue
Total content conversions1825+39%

Be careful with interpretation. A content audit can improve performance, but it can also make reporting cleaner by removing noise. If traffic drops slightly but conversions rise, that may be a win for a B2B software site. If impressions rise but conversions do not, your updates may have improved visibility without improving the offer or next step. If merged pages lose traffic, check whether redirects are working and whether the new destination fully covers the old search intent.

Your three-month review should end with another action list, not just a report. Mark each changed page as:

  • Working: leave it alone for now.
  • Needs another update: improve title, introduction, depth, proof, internal links, or call to action.
  • Needs promotion: add internal links from stronger pages or include it in campaigns.
  • Needs reconsideration: if a page still has no value after an update, merge or delete it.

A small B2B content audit is successful if you can say, with evidence, that your site is cleaner, your best pages are easier to find, outdated material is gone, and your highest-intent pages have a better chance of turning visitors into pipeline. You do not need expensive tooling to get there. You need one spreadsheet, consistent rules, and the discipline to make decisions page by page.

Draft B

1,665 words

Reveal the model and what it cost

Claude Opus 5 Anthropic

Recorded usage at current rates
$0.0828
Output tokens
3,222
Tokens per word
1.94
Calculator quoted
$0.0561
Price this model for your own volume →

How to Run a Content Audit on a Small B2B Website

You have somewhere between 40 and 80 pages live, and you cannot confidently say what most of them do. Some were written by a founder in 2021. Some were commissioned from a freelancer whose name you have forgotten. Two of them rank for something, though you are not sure which two. When someone asks whether you should write more, you have no defensible answer, because you have no inventory of what already exists.

Two days and a spreadsheet will fix that. Here is how to spend them.

Read the rest of Draft B

What a content audit is, and what it is not

A content audit is an inventory with a decision attached to every row. You list every URL you own, pull a small set of facts about each one, and then commit — in writing, in a cell — to keeping it, updating it, merging it into something else, or deleting it. The output is not a report. The output is a to-do list ordered by impact, and a smaller, sharper site three weeks later. If you finish the audit and nothing on the site has changed, you did not run an audit; you built a spreadsheet.

It is not a quality review, a rewrite, or an SEO project. You are not reading every page closely and grading its prose — at 60 pages that would take a week and produce opinions rather than decisions. You are not diagnosing technical crawl issues, chasing keyword gaps, or planning next quarter's editorial calendar. Those are separate jobs that become much easier after the audit, because you will know what you have. Resist the urge to fix things as you go. Auditing and fixing at the same time is how a two-day project becomes a two-month project that stalls at page 19.

What to collect for each page

Open one spreadsheet. One row per URL. Build these columns, left to right, and fill them in the same order for every page — batching by column is far faster than working page by page.

  1. URL — the full address, including any trailing slash.
  2. Page title — the H1 or the title tag, whichever is more descriptive.
  3. Page type — blog post, product page, solution page, case study, gated asset landing page, docs, legal, careers, other. Use a dropdown so you can filter later.
  4. Published date — approximate is fine.
  5. Last updated date — if you cannot tell, mark "unknown" and treat it as the published date.
  6. Word count — rough buckets are enough: under 400, 400–1,000, 1,000–2,000, 2,000+.
  7. Sessions, last 12 months — from your analytics landing-page report.
  8. Sessions, last 3 months — the same report with a shorter window. You need both to see trend.
  9. Conversions, last 12 months — demo requests, trial signups, or whatever your single primary conversion is, attributed to the page as a landing page. If you cannot attribute conversions to a page, use a proxy: clicks on the page's primary call to action, or sessions that continued to your pricing or demo page.
  10. Traffic source mix — one word: organic, direct, referral, social, paid, or mixed. Take whichever channel supplied the most sessions.
  11. Top query or entry context — if you have search console data on the free tier, the query bringing the most impressions. If not, write how people arrive: "linked from nav," "shared in newsletter," "no idea."
  12. Internal links in — how many other pages on your site link to this one. Do not do this by hand for all 60 pages; do it for the 20 pages you are considering deleting or merging, which is where it matters.
  13. Product accuracy — a three-value dropdown: accurate, minor drift, wrong. Minor drift means old screenshots or an outdated plan name. Wrong means it describes a feature you no longer ship or a positioning you have abandoned.
  14. Strategic fit — does this page speak to a buyer you still want? Yes / partly / no.
  15. Decision — keep, update, merge, delete.
  16. Merge target — the URL it should fold into, if applicable.
  17. Effort — hours you estimate: 0, 1, 3, or 8.
  18. Notes — one line, maximum.

Columns 1–6 take about two hours for 60 pages. Columns 7–11 take another two if your analytics is set up; export the landing-page report to CSV and use a lookup formula rather than copying numbers by hand. Columns 12–14 are judgment and take the longest — budget half a day. Do not start on decisions until every other column is filled, or your first twenty decisions will use different standards than your last twenty.

How to decide: four rules

Sort by sessions, descending, and work top to bottom. Apply these rules in order — the first one that matches wins.

Delete when the page has fewer than 30 sessions in twelve months, zero conversions, fewer than two internal links in, and is either strategically off-target or factually wrong. All four conditions must hold. A page meeting all four is costing you: it dilutes your crawl budget marginally, but more importantly it is a page a prospect might land on and read something false about your product. Delete does not mean "unpublish and hope" — it means remove the page and 301-redirect the URL to the closest relevant page, or to the blog index if nothing is close. Expect this to catch 15–25% of a neglected B2B site: event recaps, "we've raised a seed round" posts, partner announcements for partners you no longer have.

Merge when two or more pages target the same buyer question and their combined twelve-month traffic exceeds the best individual page's traffic by less than 40%. That gap tells you they are cannibalising each other rather than serving distinct intents. Pick the page with the most internal links in — not the most traffic — as the survivor, because inbound links are harder to rebuild than rankings. Move the genuinely unique material from the others into it, then redirect the others to it. Small B2B sites accumulate three or four near-duplicate posts on their core topic because each new hire wrote their own version.

Update when the page has meaningful traffic or conversions but product accuracy is "minor drift" or "wrong," or when the last update was more than 18 months ago and the page still gets steady organic traffic. Meaningful means above the median session count for your site. These are your highest-return hours in the whole project: the demand already exists and you are removing friction, not creating demand from scratch.

Keep — and change nothing — when the page is accurate, strategically on-target, and either performing above median or serving a function that traffic does not measure, such as your security page, your careers page, or a case study your sales team sends manually. Write "keep" in the cell and move on. The discipline here is refusing to improve things that are fine.

A worked example

Say you have four pages about integrations. Page A: 2,400 sessions, 11 conversions, 9 internal links, published 2022, updated never, describes a connector UI you redesigned last spring. Page B: 620 sessions, 1 conversion, 2 internal links, same topic, published 2023. Page C: 140 sessions, 0 conversions, 0 internal links, an announcement of a partnership that ended. Page D: 3,100 sessions, 4 conversions, accurate, updated four months ago.

C is deleted: under 30 sessions? No — 140. So the delete rule fails, and you check merge instead. C shares no intent with the others and is factually dead, so treat it as an exception and delete it anyway with a redirect to A, noting the override. A and B: combined 3,020 against A's 2,400 — a 26% lift, under 40%, so merge B into A. A survives on internal links. Then A becomes an update: strong traffic, wrong screenshots. Estimate 3 hours. D is a keep.

Four pages, roughly four hours of work, and you have consolidated 3,020 sessions onto one accurate page.

What to do first

Filter to decision = delete and execute all of them in one sitting. Deletions are the fastest changes, they need no writing, and shipping them the same week proves to yourself and to anyone watching that this is a project with output. Set up the redirects as you go, and keep a tab recording old URL, new URL, and date — you will need it when someone asks why a link in a two-year-old email is broken.

Second, execute the merges, hardest-hitting first. Third, sort your update rows by sessions divided by estimated hours, and work down that list. Do not attempt more than five updates in the first month; you will find that each one reveals a product detail you need someone else to confirm.

How to know, three months later, whether it worked

Before you change anything, freeze a baseline in a new tab: total sessions and total conversions for the trailing 90 days, the same two figures for the pages you are keeping and updating, your page count, and the median sessions per page.

At the 90-day mark, pull the same five numbers. Judge them like this: total sessions may dip slightly — you removed pages — and that is acceptable if conversions held or rose. Median sessions per page should rise noticeably, often 30% or more, because you removed the long tail of dead weight. The pages you updated should each beat their own prior 90-day session count; if fewer than half of them did, your update rule was too generous and you were polishing rather than fixing. Conversions are the deciding number: if total conversions fell more than 10% and no other channel changed, revisit your redirects first, because a mis-targeted redirect is the most common cause.

Then re-run the whole thing in twelve months. It will take four hours.

Draft C

1,506 words

Reveal the model and what it cost

Claude Sonnet 5 Anthropic

Recorded usage at current rates
$0.0302
Output tokens
2,929
Tokens per word
1.94
Calculator quoted
$0.0204
Price this model for your own volume →

How to Run a Content Audit on a Small B2B Website

You have somewhere between 40 and 80 pages live on your website, and you genuinely don't know what half of them are doing for you. Some are probably ranking for keywords you forgot you targeted. Some are getting traffic but no conversions. Some haven't been touched since a product launch two years ago and reference a feature you've since renamed. You have two days and a spreadsheet. That's enough.

What a Content Audit Actually Is

Read the rest of Draft C

A content audit is an inventory of everything you've published, paired with a judgment call on each piece: keep it, fix it, combine it with something else, or kill it. It's a decision-making exercise, not a data-collection exercise. The spreadsheet is just the mechanism that lets you see 60 pages at once instead of clicking through them one at a time in a CMS.

It is not a rewrite project, a keyword research project, or an SEO audit in the technical sense — you're not checking site speed or crawlability here. It's also not a popularity contest where the pages with the most traffic automatically win. A page with low traffic that supports a high-intent buyer keyword can be more valuable than a blog post that gets 2,000 visits a month from people who will never buy software from you. The audit's job is to separate pages that are doing work from pages that are just sitting there, and to give you a short, prioritized list of actions instead of a vague sense that "the blog needs help."

What to Collect for Each Page

Open a spreadsheet, one row per URL, and pull these columns. All of this is available from your CMS export and free-tier analytics — you don't need to pay for anything.

  • URL
  • Page title / H1
  • Page type (blog post, product page, landing page, guide, case study, comparison page, etc.)
  • Publish date
  • Last updated date
  • Word count
  • Sessions, last 12 months (from analytics)
  • Sessions, last 3 months (to catch recent trend, not just annual average)
  • Organic entrances / organic sessions (isolate search traffic specifically)
  • Top landing page keyword, if your analytics or search console shows one with meaningful impressions
  • Average position for that keyword (from your search console equivalent)
  • Conversions attributed to the page — this can be as blunt as "form submits on this URL" or "this URL appears in the conversion path" depending on what your free analytics tier gives you
  • Internal links pointing to this page (a rough count; you can get this from a site search or crawl tool with a free tier)
  • CTA present? (yes/no, and what kind — demo request, newsletter signup, none)
  • Owner (who on your team would need to touch this if it's updated)
  • Notes (anything qualitative you notice while reviewing — outdated screenshots, broken links, wrong pricing, tone mismatch)

That's 15 columns. Filling them in for 60 pages is mechanical work — budget four to five hours for it, most of which is exporting analytics data and pasting it in, not typing. Do this in one sitting if you can; context-switching in and out of the audit is where these projects die.

How to Decide: Keep, Update, Merge, Delete

This is where most audits fall apart, because "review each page and decide" isn't a decision rule — it's an invitation to relitigate every page for ten minutes. Use these four rules instead, in this order, and apply them mechanically.

Delete if: the page has had fewer than 10 organic sessions in the last 12 months, has zero internal links pointing to it, and isn't a page you're legally or contractually required to keep (like a pricing page or a terms of service page). If a page meets all three conditions, it is not being found, not being linked to, and not contributing anything — deleting it (with a redirect to a relevant page, not a 404) removes clutter and can consolidate authority elsewhere. Don't delete based on traffic alone; a page with 8 sessions a year that's linked from your homepage nav is a different animal than an orphaned blog post nobody sees.

Merge if: two or more pages target the same keyword or answer the same buyer question, and neither one is clearly winning. A practical trigger: if two pages both rank between position 8 and 20 for overlapping terms, and both have similar or mediocre conversion rates, you're splitting your own relevance. Combine them into the stronger of the two URLs, redirect the weaker one, and consolidate the internal links that pointed to the loser onto the winner.

Update if: the page gets meaningful organic traffic (say, more than 50 sessions in the last 3 months) but has a stale publish date, references outdated product details, or has a conversion rate visibly lower than similar pages on your site. This is your highest-leverage bucket — you're not creating anything from scratch, you're fixing something that's already proven it can attract visitors.

Keep as-is if: it's converting at or above your site average, was updated within the last 6 months, and isn't cannibalizing another page's keyword. Leave these alone. Resist the urge to "improve" pages that are already working — that's how audits turn into open-ended redesign projects that never ship.

Worked Example

Say you find these three pages in your spreadsheet:

  • /blog/what-is-api-integration — published 3 years ago, 1,400 words, 340 organic sessions last 12 months, 6 sessions last 3 months, ranks position 34 for "what is api integration," 2 internal links, no CTA, 0 conversions.
  • /blog/api-integration-guide — published 8 months ago, 2,100 words, 1,200 organic sessions last 12 months, 410 sessions last 3 months, ranks position 6 for "api integration guide," 14 internal links, has a demo CTA, 9 conversions.
  • /product/integrations — published 2 years ago, unchanged since, 600 words, 890 organic sessions last 12 months, 240 sessions last 3 months, ranks position 4 for "software integrations platform," 22 internal links, has a demo CTA, 31 conversions.

Applying the rules: the first page is a merge candidate — it overlaps heavily with the guide, is losing badly (position 34 vs. position 6), and is trending toward irrelevance (6 sessions in the last quarter). Fold its unique content into the guide, redirect the URL, and move on. The second page is a clear keep — strong recent performance, healthy conversion rate relative to traffic, recently published. The third page is an update — it converts extremely well (31 conversions off 890 sessions is a strong rate) but hasn't been touched in two years, which is a risk given how fast product feature sets change; refresh the copy and screenshots without touching the structure that's clearly working.

Three pages, three different actions, decided in about two minutes once the columns are filled in. That's the whole method — do this for every row.

What to Do First

Once your spreadsheet has a decision in every row, don't start with the deletes and don't start with the biggest project. Start with the update pages that have the highest current traffic, because they're the fastest path to measurable lift — you're improving something Google already trusts, not waiting for a new or merged page to get re-crawled and re-ranked. In the example above, the /product/integrations refresh would be first in line, not the merge.

Batch the deletes and redirects together and do them in one pass, ideally the same afternoon, so you're not leaving a trail of half-finished 404s. Merges come next, since they involve writing and redirect setup. Save "keep" pages for a calendar reminder to recheck in 6 months — they need nothing from you right now.

Write the priority order directly into your spreadsheet as a numbered column so this doesn't live only in your head. You have limited time; the point of the audit is that you don't have to re-decide priority later.

How to Tell in Three Months Whether It Worked

Pick three metrics before you start executing, and check them again at the 90-day mark using the same free analytics tier:

  1. Organic sessions on updated pages, compared to their own pre-update 3-month baseline — not compared to site average, since you want to isolate the effect of your changes.
  2. Conversion rate on updated and merged pages, same before/after comparison.
  3. Total indexed pages, to confirm your deletes and merges actually reduced clutter rather than just adding notes to a spreadsheet nobody acted on.

If updated pages show flat or declining traffic after three months, that's not necessarily a failure of the audit — it may mean the update itself was too shallow, or the page was declining for reasons unrelated to content quality, like a competitor outranking you structurally. Either way, you'll know, and you'll know specifically which pages to look at next, which is the entire point of doing this exercise instead of guessing.

Draft D

2,369 words

Reveal the model and what it cost

GPT-5 nano OpenAI

Recorded usage at current rates
$0.00321
Output tokens
7,988
Tokens per word
3.37
Calculator quoted
$0.00125
Price this model for your own volume →

How to Run a Content Audit on a Small B2B Website

You have 40–80 published pages, you’ve never done this before, you’re working with a spreadsheet and free analytics, and you’ve got two days to make sense of it all. Here’s a practical, no-frills approach that fits your constraints and helps you decide what to keep, update, merge, or delete, then track progress three months down the line.

1) What a content audit is and what it is not

Read the rest of Draft D

A content audit is a two-part exercise you can complete with a single, clean sheet of data:

  • First, you inventory every page, capture what exists, and pull basic performance signals from your free analytics. You end up with one row per page and a consistent set of fields you’ll populate for every page.
  • Second, you apply measurable rules to decide what action to take on each page (keep as-is, update, merge with another page, or delete). The goal is to surface value, fix gaps, remove duplications, and improve overall site performance in a way you can track.

A content audit is not a redesign sprint, a long-term editorial plan, or a marketing manifesto. It is not a vanity exercise, and it does not require expensive tooling. It is not a stockpile of ideas for new topics or a commitment to rewrite every page at once. It is a practical, repeatable process you can run on a constrained budget to surface the truth about what’s actually working on your site today—and what isn’t.

2) What to collect for each page, as a concrete list of spreadsheet columns

Create a single spreadsheet with one row per page and a consistent column set. Here is a concrete, bite-sized list you can implement now. Where you report numbers, pull them from your free analytics export and from your site’s content.

  • Page URL (the exact path)
  • Page Title (the on-page title)
  • H1 Text (the main heading on the page)
  • Meta Title (what shows in search results)
  • Meta Description (search snippet)
  • Canonical URL (should be the same as Page URL, unless you know you have canonical issues)
  • Page Type (choose one: Product Page, Pricing, Case Study, Blog Post, How-To, Guide, Resource, FAQ, Support)
  • Section/Category (the site’s top-level grouping, e.g., “Product,” “Solutions,” “Resources”)
  • Publish Date (or first publish date)
  • Last Updated Date (most recent content refresh)
  • Word Count (rough word count on the page)
  • Images on Page (count)
  • Internal Links (count of internal links on the page)
  • External Links (count of external links)
  • Monthly Unique Visitors (last 30 days)
  • Sessions Last 90 Days (total sessions to the page in the last 90 days)
  • Avg Time on Page (seconds)
  • Bounce Rate (percent)
  • Entrances (sessions where this page was the first page)
  • Conversions (any form submissions, demo requests, contact clicks attributed to this page in the last 90 days)
  • Conversion Rate (Conversions Ă· Entrances, expressed as a percentage)
  • Traffic Trend (percent change in Sessions Last 90 Days vs. previous 90 days)
  • Ranking Signals / Keywords (optional: top keyword impressions or rankings for this page, if you can export from a free search data tool)
  • Internal Link Anchors Targeting This Page (brief note on key internal links pointing to this page)
  • Status (to be filled after you audit: Keep, Update, Merge, Delete)
  • Notes/Recommendations (for quick rationale you can reference later)

Worked example (one row, illustrative numbers)

  • Page URL: /product/demo
  • Page Title: "Demo of Our Product"
  • H1 Text: "See the Product in Action"
  • Meta Title: "Product Demo | Free Trial"
  • Meta Description: "Watch a quick demo and start your free trial today."
  • Canonical URL: /product/demo
  • Page Type: Product Page
  • Section/Category: Product
  • Publish Date: 2023-11-10
  • Last Updated Date: 2025-02-01
  • Word Count: 842
  • Images on Page: 8
  • Internal Links: 12
  • External Links: 3
  • Monthly Unique Visitors: 30
  • Sessions Last 90 Days: 120
  • Avg Time on Page: 210
  • Bounce Rate: 42%
  • Entrances: 60
  • Conversions: 4
  • Conversion Rate: 6.7%
  • Traffic Trend: +5% (last 90 days)
  • Ranking Signals / Keywords: Top keyword “product demo” rank #6
  • Internal Link Anchors Targeting This Page: “Book a demo,” “Pricing”
  • Status: Keep
  • Notes/Recommendations: "Stale pricing in the CTA; refresh with current pricing and add a stronger value prop."

Tips for getting this done quickly

  • Use your content inventory to map pages to a simple key: URL, type, and current performance. You don’t need perfect data—just consistent data you can compare across pages.
  • For fields you can’t fill yet, leave blank and fill later, but keep a placeholder so you don’t forget.
  • Favor clarity over perfection. A clean, complete sheet at 70–90% is far more usable than a perfect sheet that’s never finished.
  • If you have pages with identical or near-identical content, you can pre-flag them for a potential merge by marking their Page Type as “Duplicate” or “Similar Topic.”

Where to pull numbers from your free analytics

  • Sessions Last 90 Days, Entrances, Conversions, Conversion Rate, Avg Time on Page, Bounce Rate, Monthly Unique Visitors: export from your site analytics within the free tier. You’ll typically get a CSV or Excel export you can paste into the sheet.
  • Traffic Trend: compute as (Sessions Last 90 Days - Sessions in prior 90 days) Ă· Sessions in prior 90 days, then format as a percentage.
  • Ranking Signals / Keywords: if you can export impressions, clicks, and average position from a free search-performance feature, include top keywords and their current rank. If not, leave this column blank or use a qualitative note such as “not tracked” for now.

Three quick notes on a worked example

  • Keep only if the page is a primary conversion path or shows material engagement (Conversions > 0 OR Sessions Last 90 Days > 20). If a page has 0 conversions and 5 sessions in 90 days, that’s a candidate for deletion or major updating unless it serves a unique purpose (reference or evergreen resource with long-tail reach).
  • If a page has 180 seconds average time but 0 conversions and the content is older than 18 months, consider updating rather than deleting.

3) How to decide what to keep, update, merge or delete — a simple rule for each outcome

Use one rule per outcome so your decisions are consistent and auditable. Apply these rules to each row in your sheet.

  • Keep (rule)
  • Keep if the page has had at least one conversion in the last 90 days OR its 90-day sessions are 20 or more and it continues to support a core funnel (product demo, pricing, trial request). In practice: Conversions_Last_90d > 0 OR Sessions_Last_90d >= 20 and Page_Type is a core funnel piece. Why: this page still earns measurable value and contributes to revenue-related goals.
  • Update (rule)
  • Update if the page shows traffic but no conversions, or if the content is older than 12–18 months and still relevant. In practice: Sessions_Last_90d >= 20 AND Conversions_Last_90d = 0 OR Last_Updated_Date < 18 months ago. Why: you can improve the page’s clarity and relevance without a full rewrite; this often yields quick wins.
  • Merge (rule)
  • Merge if two or more pages cover the same intent and the combined page would outperform the individual ones. In practice: Page A and Page B share the same topic with overlapping keywords and one page has notably higher Sessions_Last_90d and Conversions_Last_90d than the other (e.g., A has 60 sessions, 2 conversions; B has 28 sessions, 0 conversions; merge into A). Then set a 301 redirect for the dead pages to the primary page. Why: reduce duplication and concentrate ranking signals on a stronger page.
  • Delete (rule)
  • Delete if the page is non-essential, has minimal traffic, and adds no unique value. In practice: Conversions_Last_90d = 0 AND Sessions_Last_90d < 10 AND InternalLinks < 2 AND Page_Type not in {Product Page, Pricing, Help Center} (or other clearly critical types). Why: frees up crawl budget and reduces maintenance while still safeguarding critical assets.

How to apply these rules without drama

  • Filter your sheet into four groups by Status: Keep, Update, Merge, Delete.
  • For each group, write a one-line justification you can reference in quarterly reports (e.g., “Merged pages A and B into C; created 301 redirects; updated meta titles”).
  • Keep a short “what changed” summary per page so you can explain decisions to teammates.

4) What to do first once the decisions are made

With decisions in hand, you’ll execute in a tight loop. Here’s a practical, two-day plan you can follow, plus the concrete actions you’ll take for each outcome.

Two-day plan (your constraint)

  • Day 1 (inventory, flags, and quick wins): Build your inventory spreadsheet, pull 90-day metrics, tag pages for Keep/Update/Merge/Delete, and identify high-potential update opportunities (e.g., pages with 20+ sessions but 0 conversions).
  • Day 2 (perform actions and document): Implement updates, plan merges and redirects, delete pages that meet the delete rule, and fix internal links and metadata. Run a quick QA pass and capture notes for month 3 measurement.

Concrete actions by outcome

  • Keep: Update internal notes in the spreadsheet, lock in a lightweight refresh plan (e.g., update meta description to include a current benefit statement within 15 minutes, adjust a heading to align with a value proposition, refresh 2–3 sentences in the copy).
  • Update: Prioritize metadata and CTAs. For example, swap in a clearer CTA such as “Start a free trial” or “Book a demo,” ensure pricing is up to date, and refresh the featured benefit in the opening paragraph.
  • Merge: Create a single, stronger page as the master (e.g., Page A). Move essential content from Page B into Page A (without duplicating conflicting claims). Set up a 301 redirect from Page B to Page A. Update internal links to point to Page A.
  • Delete: Remove the page from navigation if necessary, then 301-redirect to the most relevant parent page (for example, a closely related guide or a product page). Remove internal links pointing to the deleted page to avoid broken paths.

Documentation you’ll produce during execution

  • A succinct change log: page URL, action, why, and the date.
  • A short QA checklist: redirects in place, meta data updated, internal links corrected, sitemap updated, and any 404s to monitor.

Tight constraints to stay on track

  • Don’t try to rewrite everything at once. Focus on quick wins first (keeps with conversions or strong traffic) and reserve deeper rewrites for later sprints.
  • Keep the changes minimal but visible in impact. If you can’t measure a change in three weeks, you can still document it and monitor the trend in the longer term.

5) How to tell three months later whether it worked

Three months out, you want evidence that your audit improved performance without introducing new problems. Gather the same metrics you used in the audit for the audited pages and compare to the baseline you captured before you started.

What to measure three months after your actions

  • Total organic sessions to audited pages: has it climbed compared with the baseline 90 days before the audit?
  • Conversions generated by audited pages: have demo requests, trials, or contact form completions increased?
  • Engagement signals on updated pages: has average time on page improved? Has bounce rate declined?
  • Page-level performance: for pages you kept or updated, do you see a positive delta in conversions or time on page? For merged pages, does the consolidated page outperform the sum of the parts in terms of conversions per session?

Simple three-month checklist

  • Compare total Sessions_Last_90d for all audited pages to the baseline. Target: at least a 15% increase.
  • Compare total Conversions_Last_90d for audited pages to baseline. Target: at least a 20% increase.
  • Compare Avg Time on Page for audited pages. Target: +10–15% improvement where content was updated.
  • Trace 3–month changes by status: do Keeps show stability or improvement? Do Updates show uplift after the change? Do Merges drive improved consolidations? Do Deletes reduce crawl depth without harming conversions?

Worked example Assume your audit covered 60 pages. Baseline (the 90 days before the audit) showed:

  • Total Sessions to audited pages: 6,000
  • Total Conversions (demo requests, trials): 120
  • Average Time on Page (audited pages): 210 seconds

Three months after implementing the audit:

  • Sessions to audited pages: 6,900 (up 15%)
  • Conversions: 150 (up 25%)
  • Average Time on Page: 235 seconds (up 12%)

This would indicate a positive outcome: more people are visiting the audited pages, more of them are converting, and the pages themselves are more engaging. If you saw mixed results (e.g., traffic up but conversions flat), you’d zoom into which pages drove the lift and adjust the update/merge strategy accordingly.

Tying it back to your constraints

  • You set up a compact, repeatable framework that works with a sheet and free analytics. You’re not shipping a multi-tool, you’re running a disciplined, data-driven cleanup. The three-month check gives you a clear, measurable verdict on whether the audit moved the needle or if you need another, tighter iteration.

Final tips to keep you moving

  • Keep the dataset lean. If you’re overwhelmed, pare columns to the essentials: URL, Page Type, 90-day Sessions, Conversions, Last Updated, and Status. You can always add fields later as you scale.
  • Treat the two-day window as a hard constraint. Don’t chase perfection in the first pass. If you finish early, you can deepen the updates on the highest-potential pages.
  • Use the decision rules as a fairness standard. If a page isn’t clearly a Keep/Update/Merge/Delete by rule, tag it for a quick review later rather than guessing.

Your content audit will yield a practical action plan you can execute without extra budget. It will give you a defensible set of changes, a clear path for updating or consolidating pages, and a way to quantify success three months down the line—all with a single spreadsheet and the free analytics you already have access to.

Draft E

2,908 words

Reveal the model and what it cost

Gemini 2.5 Flash Google

Recorded usage at current rates
$0.0190
Output tokens
7,575
Tokens per word
2.60
Calculator quoted
$0.00957
Price this model for your own volume →

How to Run a Content Audit on a Small B2B Website

You know your website is the cornerstone of your B2B marketing efforts, but with 40 to 80 pages of content, it's easy for some pieces to get lost, become outdated, or simply not pull their weight. You need to ensure every page contributes to your goals, engages your audience, and justifies its place on your site. This guide will walk you through a practical, no-cost content audit you can complete in about two working days, using just a spreadsheet and your free analytics.

A content audit is a systematic process of inventorying and evaluating your website's existing content. Its primary purpose is to assess the performance, relevance, and alignment of each page with your current business objectives and target audience. By gathering data on key metrics for every piece of content, you can make informed decisions about what to keep, update, consolidate, or remove, ultimately optimizing your website for better engagement, search visibility, and lead generation. This isn't just about cleaning house; it's about strategically leveraging your existing assets to maximize their impact.

Read the rest of Draft E

What a content audit is not is equally important. It's not a deep dive into technical SEO issues like server response times or broken schema markup. It's also not a comprehensive keyword research initiative from the ground up, nor is it a detailed competitive analysis of your rivals' content strategies. Instead, this specific audit is a focused, practical exercise designed to give you immediate, actionable insights into your own content's effectiveness without requiring specialized tools or extensive time commitments. Your goal here is to optimize what you already have, not to embark on a complete website overhaul or a large-scale content strategy redesign.

What to Collect for Each Page

To conduct an effective audit with the resources you have, you'll start by creating a central spreadsheet. This will be your primary tool for collecting data and making decisions. For each of your 40-80 pages, you'll gather specific pieces of information. This process might seem tedious initially, but it’s critical for making data-driven decisions.

Here are the columns you’ll need in your spreadsheet, along with an explanation of how to fill them using your existing website and free analytics:

  1. URL: The complete web address of the page. You can easily compile a list of all your site’s URLs by manually navigating your site or by using your analytics platform's "All Pages" report.
  2. Page Title: The text that appears in the browser tab. You can find this by visiting each page and looking at the browser tab or by viewing the page source (<title> tag).
  3. Meta Description: The short summary text for the page that often appears in search engine results. View the page source (<meta name="description"> tag) to find this.
  4. Content Type: Categorize the page (e.g., Blog Post, Solution Page, Product Feature, Case Study, About Us, Contact Page, Landing Page). This helps you understand the content mix.
  5. Publish Date: When the content first went live. This is often visible on the page itself or accessible via your content management system (CMS).
  6. Last Updated Date: When the page was last significantly revised. If your CMS tracks this, record it. If not, note "N/A" or manually estimate based on content freshness.
  7. Primary Keyword/Topic: The main keyword or topic you intended for this page to rank for or address. Even without advanced tools, you likely have a sense of each page’s core focus.
  8. Target Audience: Which specific segment of your B2B audience is this page designed for? (e.g., IT Managers, Sales Directors, Developers, HR Professionals).
  9. Business Goal: What strategic purpose does this page serve? (e.g., generate leads, educate prospects, support existing customers, establish thought leadership, drive product sign-ups). Be specific.
  10. Unique Pageviews (Last 12 Months): This is a crucial performance metric. Go into your free analytics platform, select a 12-month period (e.g., last calendar year), navigate to your "All Pages" or "Page Performance" report, and record the unique pageviews for each URL. Using a consistent 12-month window ensures fair comparison.
  11. Bounce Rate (Last 12 Months): Also from your analytics. A high bounce rate often indicates that users aren't finding what they expect or that the content isn't engaging.
  12. Average Time on Page (Last 12 Months): Another key engagement metric from your analytics. Longer times typically mean more engaged readers.
  13. Conversion Rate (Last 12 Months): If you track specific conversions (e.g., form submissions, demo requests, whitepaper downloads) tied to individual page goals, record this. If not, mark "N/A" or note the number of conversions.
  14. Internal Links In: How many other pages on your site link to this specific page? For 40-80 pages, a manual check by searching your site (e.g., site:yourdomain.com "page title") or reviewing related content can give you a rough count. This indicates the page's importance within your site's structure.
  15. Internal Links Out: How many links from this page go to other internal pages on your site? Manually count these. Good internal linking keeps users on your site and helps distribute "link equity."
  16. External Links Out: How many links from this page go to external websites? Manually count and quickly check if they are still working and relevant.
  17. Notes: Use this column for any initial observations, questions, or ideas that come to mind as you collect the data. This could include "looks outdated," "poorly written," or "could be combined with X."
  18. Decision: (This column will remain blank initially, filled in the next stage with Keep, Update, Merge, or Delete).
  19. New URL (if applicable): If you decide to merge or delete a page and need to implement a redirect, note the target URL here.
  20. Action Items: (This column will also be filled later with specific tasks for each decision).

Remember to maintain a consistent time frame (e.g., the last 12 full months) for all your analytics data. This ensures you're comparing apples to apples across all your content. For a small site, gathering this data manually might take a significant portion of your first day, but it's an indispensable foundation for your decisions.

How to Decide What to Keep, Update, Merge, or Delete

With your spreadsheet populated, you're ready for the most critical phase: making decisions. Your goal is to optimize your content inventory, ensuring every page earns its place. For a site of your size, a practical approach is to consider your average site-wide performance as a benchmark. For instance, if your average blog post gets 50 unique pageviews per month, anything significantly above that is "high performing," and anything significantly below is "low performing."

Here are the rules to guide your decisions for each outcome:

Keep

Rule: If a page consistently attracts a high volume of unique pageviews (e.g., in the top 25% of your site's content), exhibits strong engagement (e.g., a bounce rate below your site's average and above-average time on page), and directly supports a core business goal or product offering without needing significant revision, then keep it as is. These are your high-performing assets; ensure they are well-linked internally and prominently featured.

Worked Example: Let's say your site's average unique pageviews per page are 60/month, your average bounce rate is 75%, and average time on page is 1:45. You have a blog post titled "Understanding API Integrations for SaaS Businesses."

  • Data: Unique Pageviews: 280/month. Bounce Rate: 42%. Average Time on Page: 4:10.
  • Analysis: This page significantly outperforms your site averages. It targets IT Managers interested in your core product offering and likely drives education and potential leads. The content is evergreen and still relevant.
  • Decision: KEEP. You might add a note to occasionally check for internal linking opportunities, but no immediate major changes are needed.
Update

Rule: If a page addresses a relevant topic, aligns with a business goal, and shows moderate but not top-tier performance (e.g., unique pageviews between 25-75% of your site's average) combined with signs of underperformance (e.g., a bounce rate above your site's average, average time on page below 2 minutes), or contains information that is clearly outdated (e.g., statistics older than 2-3 years, references to old product versions), then update it. Focus on improving clarity, depth, keyword targeting, adding fresh data, and enhancing internal linking to make it more valuable and engaging.

Worked Example: You have a solution page titled "Optimizing Your B2B Sales Funnel with [Your Software Name]."

  • Data: Unique Pageviews: 75/month. Bounce Rate: 88%. Average Time on Page: 1:05. Publish Date: 3 years ago.
  • Analysis: This page gets average traffic but users are leaving quickly, suggesting the content isn't resonating or is missing something. It's a core topic for your target audience (Sales Directors) and critical to your business. The age indicates it's likely outdated.
  • Decision: UPDATE.
  • Action Items: Refresh statistics, add a new section detailing recently launched features that support sales funnel optimization, improve calls-to-action (CTAs), embed a relevant demo video, and add internal links to case studies.
Merge

Rule: If you identify two or more pages that cover very similar topics, target the same audience, and individually perform poorly (e.g., each receiving unique pageviews below 25% of your site's average, high bounce rates for all involved pages), then merge them into one comprehensive, authoritative page. This consolidates "link equity," improves depth for search engines, and creates a better user experience by centralizing information. Ensure you implement 301 redirects from the old URLs to the new consolidated page.

Worked Example: You have three blog posts: "The Basics of Cloud Security" (25 unique pageviews/month, 87% bounce rate), "Securing Your Data in the Cloud" (18 unique pageviews/month, 90% bounce rate), and "Top Cloud Security Best Practices" (30 unique pageviews/month, 85% bounce rate). All were published over two years ago and target IT Managers.

  • Analysis: Individually, these pages are underperforming significantly. They are closely related in topic and audience. Consolidating them into one strong resource would eliminate redundancy and create a more valuable piece of content.
  • Decision: MERGE.
  • Action Items: Create a new comprehensive page titled "The Ultimate Guide to B2B Cloud Security," combining the best elements of the three existing posts. Ensure the new page is well-optimized with updated information and a clear CTA. Set up 301 redirects from the three old URLs to the new guide's URL.
Delete

Rule: If a page is entirely irrelevant to your current business goals, targets an audience you no longer serve, contains severely outdated or incorrect information that cannot be easily updated, or has consistently negligible traffic over a prolonged period (e.g., fewer than 5 unique pageviews/month for 12+ months) with no conversions and minimal or no internal links, then delete it. If the page has any internal links pointing to it, or if it has ever received traffic from external sources (even if historical), implement a 301 redirect to a relevant, high-performing page or your homepage. If it has truly zero historical value or links, simply remove it.

Worked Example: You discover an old company announcement blog post titled "Our Annual Team Retreat 2017."

  • Data: Unique Pageviews (Last 12 Months): 1. Bounce Rate: 100%. Average Time on Page: 0:05. No conversions. Only internally linked from a distant "Archives" page. Not relevant to your software offerings.
  • Analysis: This page provides no value to your current target audience or business goals, receives virtually no traffic, and has no strategic importance.
  • Decision: DELETE.
  • Action Items: Remove the page from your website. Since it has negligible traffic and no strategic internal links, a 301 redirect is likely unnecessary, simplifying the process. If, however, you found that it had, for example, 50 unique pageviews/month in 2017 (even if none now) or an external link, you would then set up a 301 redirect to your main blog page or homepage to preserve any lingering SEO value.

By applying these rules consistently across your spreadsheet, you'll systematically categorize every page. This step is where your audit truly becomes actionable, laying the groundwork for your content optimization strategy.

What to Do First Once the Decisions Are Made

Once you’ve assigned a decision (Keep, Update, Merge, Delete) to every page in your spreadsheet, you’ll have a clear roadmap. Now it’s time to act, but don’t try to do everything at once. Focus on impact and efficiency.

  1. Prioritize Your Actions:
  • Quick Wins (Deletions): Start by identifying pages marked "Delete" that do not require a 301 redirect. These are typically outdated, irrelevant, and low-traffic pages with no significant internal or external links. Removing them instantly cleans up your site with minimal effort.
  • Strategic Consolidations (Merges): Next, focus on your "Merge" decisions. Consolidating weak, related pages into one strong, authoritative piece can yield significant SEO benefits and improve user experience quickly. Implement the redirects immediately after the new page is live.
  • High-Potential Improvements (Updates): Then, tackle "Update" tasks, prioritizing pages that have moderate traffic but high bounce rates or low time on page, especially if they are relevant to your core offerings. These pages have existing visibility that you can leverage with targeted improvements.
  • Maintenance (Keeps): Finally, for pages marked "Keep," your initial action might just be to review their internal links to ensure they're well-connected to other relevant content on your site.
  1. Create a Detailed Action Plan: For every "Update" and "Merge" page, expand on the "Action Items" column in your spreadsheet. Break down the work into specific, manageable tasks:
  • For Updates: "Rewrite intro," "add new statistics from [source]," "embed [specific video]," "improve CTA wording to [new text]," "add internal links to [page X] and [page Y]."
  • For Merges: "Draft new page combining content from [old URL 1], [old URL 2], [old URL 3]," "design new header image," "implement 301 redirects from [old URL 1] to [new URL]."
  1. Implement Deletions and Redirects: This is a crucial technical step. For all "Delete" pages needing a 301 redirect (any page with any historical traffic or inbound links) and all old URLs from your "Merge" operations, work with your web developer or CMS administrator to implement these redirects. A 301 redirect signals to search engines that a page has permanently moved, passing on any existing SEO value to the new destination. Incorrect redirects can harm your site's SEO, so ensure this is done correctly.
  1. Begin Content Refinement: With the foundational deletions and redirects in place, you can now begin the actual writing and editing work for your "Update" and "Merge" pages. Tackle these systematically, focusing on improving clarity, value, and aligning them even more closely with your business goals and audience needs.
How to Tell Three Months Later Whether It Worked

A content audit isn't a one-and-done activity; it's the beginning of an optimization cycle. To determine if your efforts paid off, you’ll revisit your free analytics three months after completing your initial actions. Compare the data from the three months post-audit to the 12 months of data you collected before the audit. Look for positive shifts in these key metrics:

  1. Overall Site Traffic:
  • What to look for: An increase in total unique pageviews and organic search traffic across your site.
  • Why it matters: A more focused, higher-quality content library should attract more visitors and perform better in search engine results.
  1. Bounce Rate (Site-wide and Per Page):
  • What to look for: A lower overall site bounce rate, and significant drops in bounce rates specifically on the pages you updated or merged.
  • Why it matters: Lower bounce rates indicate that users are finding your content more relevant and engaging, staying longer on your site, and exploring more pages.
  1. Average Time on Page (Site-wide and Per Page):
  • What to look for: An increase in average time on page, especially for the content you updated or merged.
  • Why it matters: More time spent on a page suggests greater user engagement and that your content is more effectively delivering the information your audience seeks.
  1. Conversion Rate:
  • What to look for: An uplift in your overall site conversion rate, and ideally, an increase in conversions directly attributed to pages you optimized to drive specific actions (e.g., demo requests, whitepaper downloads).
  • Why it matters: Ultimately, your B2B content should contribute to lead generation and business growth. Improved conversions are a direct indicator of content effectiveness.
  1. Organic Search Visibility (Qualitative Check):
  • What to look for: While free analytics won't show direct keyword rankings, you can manually check Google for the primary keywords associated with the pages you updated or merged. Did they move up in search results? Are new "merged" pages appearing for relevant queries?
  • Why it matters: Better content that is more focused and comprehensive tends to rank higher, increasing your organic visibility and attracting more qualified traffic.
  1. Internal Link Clicks:
  • What to look for: If your analytics platform allows, observe increased clicks on internal links from the pages you updated.
  • Why it matters: Improved internal linking creates a more logical and engaging user journey, helping users discover more of your valuable content and signaling content relationships to search engines.

By reviewing these metrics, you’ll gain a clear understanding of the impact of your content audit. This data will not only validate your efforts but also inform your ongoing content strategy, allowing you to continually refine your approach and ensure your small B2B website consistently drives maximum value.

After the reveal

Compare models, recorded usage and current-rate cost

These are the providers' own token counts for the generations above, priced with the same current rate data the calculator uses. Usage is recorded; these dollar amounts are current-rate repricing, not preserved historical invoices. Rates last verified 2026-09-24.

Model Usage at current rates Calculator estimate Words Output tokens per word
GPT-5 nano Read Draft D $0.00321 $0.00125 2,369 3.37
Gemini 2.5 Flash Read Draft E $0.0190 $0.00957 2,908 2.60
Claude Sonnet 5 Read Draft C $0.0302 $0.0204 1,506 1.94
Claude Opus 5 Read Draft B $0.0828 $0.0561 1,665 1.94
GPT-5.5 Read Draft A $0.1173 $0.1076 2,709 1.42

“Calculator estimate” is what this site would have quoted for an article of that length at 1.3 tokens per word. Compare estimated usage with recorded usage on the same current-rate basis before you budget. Each model name links its provider's calculator page; every rate behind these figures is on the full pricing table.

Every model here used more than the 1.3 output tokens per word the calculator assumes — between 1.42 and 3.37. The base estimate understated recorded usage in this run; this observed range is not a universal lower bound or statistical error bar.

The brief asked for 1,500 words. Claude Sonnet 5 came closest at 1,506; Gemini 2.5 Flash wrote 2,908 — 94% over. You are billed for the overshoot, and an editor pays for it twice, because cutting is work.

Draft D, Draft E used more than two output tokens per written word. Aggregate usage does not tell us how much of that comes from reasoning versus tokenizer differences. Do not infer a reasoning-token breakdown from this ratio alone.

After the blind read

Editorial observations on this run

Written 2026-08-24, after reading all 5 drafts of the 2026-08-20 run in full.

Read the drafts blind first — these notes are spoilers of judgement, if not of identity. Drafts are named by letter only, so the reveal stays yours. The notes are opinion; usage counts are recorded and dollar amounts use current list rates.
Note 01

Only one draft hit the commissioned length

The brief asked for 1,500 words. Draft C delivered 1,506; Draft E nearly doubled the commission at 2,908. The overshoot is billed at the same rate as the words you wanted, and an editor is then paid to cut it — you pay for the excess twice.

Note 02

One draft opens with exactly the banned throat-clearing

The brief prohibited opening on the importance of content. Draft E opens with "your website is the cornerstone of your B2B marketing efforts" — the sentence the brief exists to prevent. The other four start from the reader’s problem, as asked.

Note 03

Rules versus descriptions — the brief’s hardest ask

The brief demanded a testable rule for each keep/update/merge/delete outcome, not a description. Drafts B and C deliver numeric thresholds an editor can apply mechanically. Draft D chains AND and OR without brackets, so two readers can apply the same rule and reach different decisions. Draft A offers "meets two of these" condition lists — workable, softer.

Note 04

The best moment on the page is an override

Draft B is the only one whose worked example runs into a case its own rules cannot handle — and says so, deletes the page anyway, and records the override. That habit of noticing when the framework fails is the judgment an editor is actually buying.

Note 05

Template-speak is the tell

Draft D’s "worked example" is a filled-in spreadsheet row with invented numbers, and its three-month section sets improvement targets (+15%, +20%) with no basis. It reads plausibly and survives a skim; it does not survive an editor asking "why these numbers?"

Method

How this was made, and how it could mislead you

Nothing was cherry-picked

Each model wrote the brief 3 times. Run 1 is published — a number fixed in the config before the first API call. The other 10 generations are committed to the repository beside these ones, so you can check that the published run is not the flattering one.

One brief is not a verdict

This is a single How-to guide commission. A model that handles it well may handle a technical explainer or a comparison piece differently. Treat this as a sample of house style and competence, not a ranking.

No sampling parameters were sent

No temperature, no top-p, to anyone. Current Claude models reject them outright, and pinning them for some providers while Claude ran on defaults would be an unfair comparison dressed up as a controlled one. Every model ran on its own defaults.

A model is not a fixed thing

Providers change what sits behind a model name. These drafts are a claim about 2026-08-20 and nothing else. When they get old enough to mislead, they get regenerated or taken down.

On the record

What failed while this was generated

  • gemini-3.7-flash — 6 failed attempts: “This model is currently experiencing high demand. Spikes in demand are usually temporary. Please try again later.”. It produced no draft at all, so it is not in the gallery above.

gemini-2.5-flash was generated separately to fill the empty slot. Only a model that returned nothing was replaced — a draft that came back and that we did not like would still be up there.

The part this page cannot show

None of these are publishable as they stand

Every draft here is a first draft from a machine that cannot check a fact, has no opinion worth printing and has never met your reader. The gap between the best of these and something you would put your name on is editing — and that cost is not on any pricing page, including ours. What the drafts do tell you is how much editing each one implies, which is the number that actually decides which model is cheapest for you.

Common questions

What people ask about these drafts

Answered from this run's own numbers, not in general.

Can AI write a good how-to article?

Judge it above — that is what this page is for. What this run shows: five current models turned the same 1,500-word brief into drafts currently repriced at $0.00321 to $0.1173 each. All five are competent enough to edit rather than rewrite, but only one hit the commissioned length, one opened with the exact phrasing the brief banned, and the drafts differ more in how they turned instructions into usable rules than in what they cost.

How much does an AI-written article cost?

Recorded token usage reprices to $0.00321 to $0.1173 each at current list rates, a 37Ă— spread for the same commission. These are not preserved invoice amounts. The calculator estimates your own length, drafts and monthly volume.

Which model wrote the best draft?

This page deliberately does not rank them. The drafts are published blind so you form a view before the model name or the price can bias it, and one commission is not a benchmark — the same models placed differently on the other briefs. The editorial observations name concrete differences by draft letter; the model guide explains how we weigh cost against quality when a choice does have to be made.

Now put a number on it

You have seen what the drafts look like. The calculator turns that into a monthly bill for your own volume, and the model guide explains where each one earns its price.

Price your own workflow Which model should I use? →