Full-Site Content Audit: 150-Page Review, Cleanup, and Keep Decisions

Bottom line first: A 150-page content site runs a full-site audit. The output isn’t a report but three tables: a keep list, a merge map, and a takedown redirect table. After applying four criteria, typically 15%–25% of pages get merged or taken down. The whole process takes 6–8 weeks at 8–10 hours per week. Batch actions cap at 30 pages per week. After each batch, check data at day 7 and day 30. One big overhaul is the surest way to trigger site-wide volatility.

1. Full-site audits and single-page refreshes are two different things

Draw the boundary before you start. How to refresh an old article, when to rewrite it — the site already has single-page refresh decisions covered in a dedicated article. How to merge two pages competing for the same keyword —keyword cannibalization that article gives the full process. What to do on a fixed monthly basis —six monthly maintenance tasks the rhythm is already set. All three cover “ongoing actions on individual pages.” This article covers something else:once a site passes a hundred pages, review all content in one pass and make a concentrated round of keep, merge, and takedown decisions.

Why do this: once a content site grows past 150 pages, three types of sediment almost inevitably appear — duplicate pages written for the same intent, outdated pages where the product line changed but the page didn’t, and dead pages that never got traffic after publishing. Individually, each page has a “fine to keep” excuse. Only when measured together against the same ruler can you see which ones are diluting the whole site’s crawl budget and topical density.

Audit cadence: sites over a hundred pages should run a full audit every 12–18 months; sites over 150 pages, every 12 months. Between full audits, maintain with quarterly rolling reviews — covered in section 7.

2. Export the list: pull data from three sources, lock the fields first

The first step isn’t reading content page by page — it’s getting the full account. Pull the page list from three sources, merge, and deduplicate:

Data source	What you get	What you don’t get	Purpose
CMS post list	All URLs, publish and update dates, authors, status	Traffic and rankings	Serves as the baseline set
Search Console Pages report	Last 90 days of impressions, clicks, average position	Pages with no impressions aren’t fully captured	Attach data to each page
Sitemap	All URLs you “think exist”	Whether they’re actually indexed	Cross-check the baseline set, find ghost pages

After merging the three sources, your baseline set will usually run 5%–15% larger than you remember — the extra pages are forgotten drafts, test pages, and leftovers from old versions. Flag those separately.

The inventory needs at least seven columns: URL, publish date, last updated, impressions (last 90 days), clicks (last 90 days), number of ranking keywords, and number of internal inbound links. The last two columns set how strict your decision lines will be: a page with 0 ranking keywords but 12 inbound links is handled very differently from one with 0 inbound links too. If data sources are missing, keep the first three columns in priority order — an inventory with no traffic data is a vote with your eyes closed.

3. Tag each page: five switches decide its fate

Go through every page and set five switches. Each switch is either on or off — no scoring, no “good/average” ratings:

Switch	Turns on when	Where to verify	Risk of misjudging
Has traffic	≥10 organic clicks or ≥500 impressions in the last 90 days	Search Console	Branded clicks don’t count as product content value
Has rankings	At least 1 keyword in the top 30	Search Console Queries report	Long-tail keywords ranking on other pages get misattributed
Has internal links	At least 3 other pages on the site link to it	Crawler tool or manual check	Nav links don’t count — only in-body links
Has conversions	Produced an inquiry, form submission, or download in the last 180 days	Form source field	When attribution is unreliable, mark it on
Informational value only	No traffic, no rankings, but the content is still must-read basics for customers	Manual judgment	Subjective — needs a second person to review

The combination of the five switches determines which decision line the page follows. The biggest mistake at the tagging stage is fixing things as you go — spotting a page and giving it a “quick tune-up” is the worst trap in an inventory, because your edits contaminate the data being exported that day. Tag everything first, touch nothing; all decisions are executed together after Section 4.

At 3–5 minutes per page, 150 pages takes about 2 working days for two people. For the subjective “informational value only” switch, have the person who knows the customer best make the call.

4. Four decision categories and their thresholds

Once the five switches are set, every page falls into one of four categories. The thresholds are:

How to handle it	Condition (any one qualifies)	Typical share	Action
Keep as is	Impressions in the last 90 days, or a conversion in the last 180 days	40%–55%	No action; add to the quarterly rolling pool
Update	Ranked but outdated: cited data older than 24 months, outdated standard version, stale screenshots	10%–15%	Add to the update queue and run the refresh process
Merge	Two or more pages compete for the same intent, each ranking for ≤2 keywords and all outside the top 20	8%–12%	Pick the main page; 301 the rest into it
Take down	Zero impressions, zero internal links, zero conversions, and either duplicate content or a discontinued product	8%–12%	301 or 410; add to the redirect table

Of the four decision lines, the one that sparks the most debate is pages that rank but get no traffic: sitting at positions 15–30 with a few dozen impressions a month, they hold a URL without producing anything. My take is that these pages should default to an update, not a takedown. They’re one step in content quality away from gaining traction, and taking them down throws away evaluation cost you’ve already paid. What actually deserves a takedown is pages with three zeros (zero impressions, zero internal links, zero conversions) and a duplicate topic. On a 150-page site, that usually clears out 10–20 pages.

5. How to close out a takedown: 301, 410, and the never-take-down list

A takedown isn’t just deleting the page. How you close it out decides whether the signals it built up are kept or wiped:

Close-out method	When it applies	Signal retention	Risk
301 to a page on the same topic	You have a replacement page on the site with a similar intent	High, authority passes through the redirect	Pointing to a page too far off gets flagged as a soft 404
410 permanent removal	No replacement page, content has no value worth keeping	Wiped, but the engine is clearly told the page is gone	Deindexing takes 2–6 weeks
Keep it with noindex	The content is useful to returning visitors but you don’t want to spend crawl budget on it	Partially retained	Long-term noindex pages gradually lose weight
Straight 404	Only for temporary takedowns restored within a week	None	Anything down longer than two weeks is treated as a 410

The “takedown redirect table” in the three spreadsheets needs four columns: original URL, action taken, target URL, and date executed. Keep this table for at least three years. External links pointing at old pages, customer bookmarks, and old references in AI engine caches will all hit these URLs over the following months, and having the table on hand lets you quickly trace unusual traffic.

Never-take-down list Write this out before you touch anything: landing pages currently generating inquiries, pages cited by outside media or in customer emails, About and Services pages that carry brand-term conversions, and pages with external backlinks where no other page can take the link. Even with three zeros, these four categories don’t come down. At most, put them on noindex or do a single round of trimming and rewriting. In takedown decisions, the cost of killing the wrong page far outweighs the cost of keeping one too many. Keeping one too many costs crawl budget; killing the wrong one costs a real backlink and a history of inquiries.

6. Internal link surgery and the batch execution rhythm

Once you’ve merged and taken pages down, the site is left with a batch of dead internal links pointing at deleted pages and a batch of “orphan pages”—pages worth keeping that nothing links to. So the fourth move in the audit is internal link surgery, in two steps:

Steps	Action	Quantity standard	Time
Repoint inbound links	Repoint every internal link pointing at a taken-down page to the 301 target or a related page you’re keeping	Every inbound link, no exceptions	10–15 minutes per page
Add links to orphans	For kept pages that rank but have ≤1 inbound link, add in-body links from 2–3 high-authority related posts	Add 2–3 per page	15 minutes per page

Bulk execution is where audits go wrong most often. If you push every change live within the same two weeks, search engines read it as a site-wide overhaul, and 2–4 weeks of ranking volatility is the usual price. A safe pace looks like this:

Pace item	Criteria	Description
Pages handled per week	≤30 pages	150 pages spread over 5–6 weeks
Checkpoint after each batch	Day 7, Day 30	Day 7: check indexing and index errors. Day 30: check clicks and rankings
Batch composition	Keep merges and removals in separate batches	Never delete and merge on the same day
Stop line	Organic clicks drop more than 20% week over week after a batch	Pause the next batch and check your redirect setup

On day 7 after each batch, open the Index Coverage report in Search Console and watch for two things: new “soft 404s” and “redirect errors.” More than 5 of either usually means your redirect map is misconfigured. For a full walkthrough of how to troubleshoot indexing after a bulk overhaul, see Troubleshooting indexing issues on the site.

7. After the audit: quarterly rolling reviews and three cases where you skip the audit

A full audit isn’t something you redo every year. The sustainable approach is a quarterly rolling review: each quarter, pull 30 pages from the site — prioritizing pages last updated more than 18 months ago and pages published more than 90 days ago — and run them through the same five-switch process. On a 150-page site, four rounds a year covers a round and a half, turning a full audit every 12–18 months into routine maintenance.

Mechanism	Cycle	Owner	Output
Quarterly rolling review	30 pages per quarter	Content owner	Update the rolling decision table
Tag new pages on entry	90 days after publishing	Publisher	Five-switch tags
Annual redirect map check	Once a year	Site maintainer	Clean up dead redirects
Full audit	12–18 months	Whole team	Rebuild all three tables from scratch

In three cases, don’t start a full-site audit. Here’s how to tell, and what to do instead:

When to skip the audit	Basis for determination	Replacement move
Site has fewer than 30 pages	You know every page by heart; the process costs more than it returns	Edit page by page and skip the three-table process
Migrated or redesigned less than 3 months ago	Migration noise distorts the data, so every threshold is unreliable	Let data settle for a quarter before starting. For protecting rankings during a migration, see Site migration and 301 mapping
Traffic is rising with no anomalies	Restructuring during an upswing: the upside is uncertain, the risk is not	Wait until you plateau; in the meantime, only do quarterly rolling reviews

Restructuring during an upswing has an uncertain payoff and a certain risk. Wait for the plateau.

FAQ

Can I use a crawler to auto-tag pages during the audit?

Crawlers can handle the structural fields — internal link counts, index status, duplicate titles — which cover two and a half of the five switches. “Information value only” and “has conversion potential” must be judged manually, because they depend on knowing the customer and the business. A workable approach: run the crawler first to generate a base table, then have a human fill in only the subjective items. For 150 pages, that saves about one working day. But note that crawlers can’t pull Search Console data, so traffic and ranking columns still need to be exported and merged separately.

When merging, how do you decide which page is the primary?

Check three conditions in order: the page with the most historically ranking keywords becomes primary; the URL whose semantics best match the current intent becomes primary; the earlier-published page becomes primary. If the first two conflict, go by ranking keyword count. Once the primary is set, 301 the secondary pages to it, fold the higher-quality passages from the secondary pages into the primary, and have the title and H2s cover the combined keyword set of both pages. All external and internal links pointing to the secondary pages must be redirected to the primary — skip this step and the merge’s impact is cut in half.

What if a page being taken down still has external backlinks?

First check where the backlinks come from and how many there are. If they come from high-authority sites or client email signatures, don’t 410 — 301 to the closest surviving page to preserve the link’s passing value. If they come from low-quality directories or spam sites, a 410 cleans things up nicely. When in doubt, always 301 — the worst a 301 does is discount the value passed; the worst a 410 does is zero it out.

How long do the three tables need to be maintained?

The takedown redirect table should be kept for at least three years: AI engines and old caches will keep hitting old URLs for months, and it’s your only tool for diagnosing abnormal traffic. The merge mapping table should be kept for one year, so you can trace back any redirect anomalies during that period. The keep list is itself the input for rolling audits, so maintain it permanently. Put all three tables in the same online document, update them once per quarterly rolling audit, and write the owner’s name in the header.

Should we keep publishing new articles during an audit?

Yes, but tag new articles with the five switches as soon as they’re added — don’t let new pages become the “untagged inventory” of the next audit round. The only thing to adjust is timing: stagger your publishing cadence from your batch disposal batches. If you delete a large batch of pages and publish several new articles in the same week, attribution gets messy. Keep new articles and disposal batches at least 3 days apart, and pause publishing during the week of a quarterly rolling audit.

What if rankings drop overall after the audit?

First separate fluctuation from a real drop: an overall shift of 5–10 positions within 2–4 weeks after batch disposal is a normal re-evaluation process — leave it alone. If it’s still more than 15% below baseline on day 30, check three things: whether the redirect table has chained redirects or soft 404s, whether the 301’d secondary pages are stealing the primary’s keywords, and whether the never-take-down list was hit by mistake. If you pinpoint specific pages, roll back those pages — don’t roll back the whole batch, because a full rollback triggers another site-level re-evaluation.

Final thoughts

The real value of a full-site audit isn’t in how many pages you delete — it’s in turning “which content is actually working for you” from a gut feeling into a ledger. Five switch states, four decision lines that trigger action, three tables that leave a paper trail, and a batch cadence that keeps risk in check — run this process once and every quarterly rollover after that is smooth sailing.

Where to start: this month, do just the three data pulls in section two and assemble your baseline set. Often, simply laying a 150-page list out in one spreadsheet is enough to make the duplicate and ghost pages surface on their own.

One line worth remembering: an audit isn’t clearing inventory — it’s taking stock of your site’s content as assets. Every page you keep should be able to say what work it does for you.