Content Audit Template: Keep, Update, Merge or Remove

Build the sheet with one row per URL, and key each row on the canonical URL. Search Console attributes clicks, impressions and position to the canonical URL, not to the duplicate.

After a merge and a 301 redirect, the old URL's rows fade and the target's rows rise. Track the target URL in the sheet, and keep the old URL until its rows stop appearing.

The seven groups are inventory, Search Console, GA4, links, judgment, decision and process. The judgment and decision columns are yours to fill. The rest come from a crawler, Search Console, GA4 or your CMS.

For the crawl and technical columns, start with the SEO audit tool, then add the crawler export.

Google recommends one query per day of data, so pull each day separately from the API. The interface export stops at 1,000 rows. The API allows up to 50,000 rows per day per site per search type.

The API default is 1,000 rows. Set rowLimit anywhere from 1 to 25,000, then use startRow to pull the next block, such as rows 25,001 to 50,000.

Page with startRow until a response comes back empty, so the last block is not missed.

Anonymized queries stay out of the tables, but chart totals still count them. Requests that combine page and query can drop rows, so treat the query column as a partial view.

Search Console keeps 16 months of data. Two full 12-month windows need 24 months, so a 12-month against prior-12-month comparison cannot come from Search Console alone. Compare the last 90 days with the same 90 days a year earlier, which fits inside the window.

A daily BigQuery bulk export keeps your own copy from the day you turn it on. It holds all performance data except anonymized queries.

GA4 landing pages are paths, so add the host before you join them to Search Console URLs. A GA4 Data API report returns 10,000 rows by default, and a limit of 250,000 raises it.

A standard GA4 property gets 200,000 core tokens per day, 40,000 per hour and 10 concurrent requests. Token cost rises with rows, columns, filter complexity and date range.

The default sampling setting for standard GA4 properties is 10m events. Results can group the least common values under an (other) row.

GA4 data thresholds are system defined, and you cannot change them. Data retention settings affect explorations and funnel reports, not standard reports.

The Search Console Links report is a sample, and its tables stop at 1,000 rows. Ahrefs Free is free for websites you add to it, and it shows which websites and pages link to yours.

Distribb says referring domains per URL are the one thing you cannot get free at any scale. Treat that as vendor advice, and plan for the limits of each free source.

The Page indexing report gives totals but may list only up to 1,000 URLs. URL Inspection allows 2,000 queries per day and 600 per minute per site, so a 10,000-URL sheet needs at least 5 days of inspection.

The free Screaming Frog SEO Spider crawls up to 500 URLs per crawl and cannot save crawls. Its licence costs £199 a year.

Search Console and GA4 totals never match exactly. Google says the trends follow the same pattern, and clicks and sessions are the most comparable pair.

Set every threshold before you sort the sheet. None of these numbers comes from Google. Each one is a vendor rule of thumb or a template default, so change it for your site.

Distribb built its thresholds from its own blog audit and calls them rules of thumb, not a validated model.

Distribb reads 90 days of Search Console data, and it uses 12 months when traffic is strongly seasonal.

Sites with high domain authority should raise every impression threshold. A new site should wait longer before it deletes anything.

Shila calls its thresholds starting points, not laws. On a small site, 30 clicks a year can be a meaningful page.

Apply the rules in this order, and stop at the first rule that matches. A page with backlinks and no traffic is a redirect candidate, never a delete.

1. Wait: the page is younger than min_age_months. Give it time before you make any other call.
2. Merge: another of your URLs ranks for the same top query and earns more clicks in the same window. Move the unique content into that page first, then redirect with a 301.
3. Noindex: customers need the page, and its impressions are at or below dead_max_impressions. Search should not show it.
4. Update: the average position falls between position_band_low and position_band_high, and impressions are at or above update_min_impressions.
5. Update: clicks fell by drop_pct or more, and the average position is 20 or better.
6. Redirect or remove: impressions are at or below dead_max_impressions, and the page is at least dead_min_age_months old. If the page has referring domains at or above link_min_ref_domains, or any key event, and a live page answers the same need, redirect it to that page. Otherwise, remove it with a 404 or 410.
7. Keep: no rule above matched.

A page answers the same need when a live page covers the same question for the same reader. A page on a related topic does not count. For a whole-site view of cannibalization, the complete SEO checklist has a one-line task, and this ladder goes deeper.

Google's content questions help decide update or merge. Ask whether the page adds substantial value compared with other pages in search results.

Score quality_score_1_5 from 1 to 5 with three of Google's questions. Ask whether the content gives original information, reporting, research or analysis, and whether it gives a substantial, complete description of the topic. Ask whether it avoids copying or rewriting other sources.

Pruning does not raise the rest of the site by itself. Google says adding or removing a lot of content to make a site seem fresh will not help rankings.

Keep the visible date honest. Google asks whether page dates were changed to seem fresh when the content has not substantially changed.

Scaled content abuse covers many pages made mainly to manipulate rankings, however they are made. Send those pages to removal or merge, not to update.

Google lists producing lots of content on many topics in hope that some of it performs as a warning sign. So does summarizing what others say without adding much value.

Word count is not a threshold. Google says it has no preferred word count, so do not cut pages by length alone.

Start the review from the drops. Google suggests auditing the pages most affected by a drop, and the search types they lost.

Google says the helpful content system became part of its core ranking systems.

For template pages, the programmatic SEO checklist has a sunset workflow for pages that get no traffic. Use it before you remove a template set.

Merge when two pages answer the same question. Semrush uses this label for pages that cover very similar topics and rank for similar keywords. Plan topic clusters on the content strategy page before you merge pages inside a cluster.

Google ranks its consolidation signals. Redirects are a strong signal, rel=canonical is a strong signal, and sitemap inclusion is a weak one. The methods stack.

Do not use noindex to choose a canonical within one site. Google says noindex blocks the page from Search completely, so use rel=canonical.

Do not use robots.txt for canonicalization. Google may still index URLs that robots.txt disallows, without their content.

Redirect several old URLs to one page only when that page holds the consolidated content. Google warns that redirecting many old URLs to one irrelevant page may be treated as a soft 404.

Distribb tells readers to redirect a page that has referring domains, even to a loosely related page. Google's guidance points the other way, so this checklist follows Google. Redirect only to a page that answers the same need, and return a 404 or 410 when none does.

Keep each redirect in place for as long as possible. When you chain a second redirect, point the first URL straight at the final target, as Distribb advises.

Whether a 410 leaves the index faster is not settled. Seer Interactive quotes a Mueller comment from 2018 that a 410 can fall out a little faster, on the order of a couple of days.

Google's own documentation lists both codes under the same treatment, so do not plan a timeline around the difference.

A 404 is not necessarily a problem when the page was removed with no replacement. If the page moved, use a 301 redirect to the new location. Googlebot keeps crawling a removed URL less and less often, and you cannot tell it to forget a URL.

Permanent redirects show the new target in results, so use one only when the redirect will not be reverted.

Download the content audit template CSV. Three invented example rows show the ladder.

The three example rows are invented. Their numbers are not data, and each row has owner set to example. Replace them before you use the sheet.

Keep the placeholder thresholds on a second tab, and point the action column at those cells. Apply the ladder in order, then check the reason column for each row.

Re-pull Search Console after the lag, and use the same window before and after the change. A pull on the day of the change shows little new data.

Compare clicks and sessions, the two most comparable metrics. Totals will not match exactly, but the trends should follow the same pattern.

Measure each action by the pages it changed, not by the whole site. One page's result should not hide the rest of the group.

Annotate each change with its date in the sheet. A later reader can then match each trend to the change that caused it.