An SEO content audit is a decision exercise wearing a spreadsheet costume. You are not cataloguing pages. You are deciding, for every URL you own, whether it earns its place — and then acting on that decision, which is the part most audits never reach.
Sites accumulate content the way garages accumulate boxes. A 2021 campaign left thirty posts. A previous agency published weekly for a year. Someone spun up 400 location pages that never ranked. None of it was deleted, because deleting feels risky, so the site now asks Google to crawl and evaluate four times more pages than it has anything useful to say about.
Pull Three Data Sets Before You Judge Anything
You need a complete URL list, performance data, and competitive context. Skipping any one of them produces confident bad decisions.
- A full crawl for every indexable URL, its title, H1, word count, canonical, and internal link count. A deep site crawl matters here — quick scans of the first 25 pages will not surface the 400 orphaned location pages, which are exactly what you are hunting.
- Search Console for 12 to 16 months of clicks, impressions, and average position per URL. This is measured data about your own site and it outranks every estimate.
- GA4 for sessions, engagement, and conversions or assisted conversions per landing page.
Join them on URL. Then add one column that most audits omit: the date the page was last meaningfully updated. A page published six weeks ago with no traffic is not a failure yet. A page from 2022 with no traffic is a decision.
Sort Every URL Into One of Four Buckets
Resist the temptation to invent a scoring model with nine weighted inputs. Four buckets, applied firmly, beat a sophisticated model nobody acts on.
- Keep. Traffic, conversions, or strategic importance. Leave it alone. Roughly 10–20% of pages on most sites, and they usually carry 70–90% of the traffic.
- Improve. Ranking positions 5–20 for terms with real volume, or impressions with a poor click-through rate. This is the highest-return bucket by a wide margin, because you are pushing pages that already have traction rather than starting from nothing.
- Merge. Multiple pages targeting the same intent. Pick the strongest URL, fold the useful content in, 301 the others.
- Delete or noindex. No traffic, no links, no conversions, no strategic reason, and no realistic path to any of those.
Do the sort in one pass and do not revisit. Second-guessing individual URLs is how content audit SEO projects turn into three-month archaeology exercises that end with nothing shipped.
The Improve Bucket Deserves Most of Your Time
Pages sitting in positions five through fifteen are the cheapest wins available anywhere in SEO. They have already proven Google considers them relevant. They need a reason to move up, not a rewrite from scratch.
Open the SERP for the target term and compare your page to the weakest result on page one — not the best one, the weakest. Position nine is the page you have to displace. Usually the gap is specific and fixable: it covers three subtopics you skipped, it answers the question in the first paragraph while yours meanders for 200 words, it has a comparison table and yours has prose, or it was updated this year and yours says 2023 in the intro.
Fix that specific gap, resubmit in Search Console, then wait. Eight weeks is a fair evaluation window. Movement of two or three positions in either direction inside the first fortnight is noise, and rewriting a page again because it dropped from 9 to 11 in week two is a good way to waste a quarter.
Merging Is Usually Better Than Deleting
When four posts all address the same intent, they split their own signals — internal links, external links, and topical relevance spread across four mediocre URLs instead of concentrating in one strong one. Google picks whichever it likes best, which is rarely the one you would choose.
Merge properly: pick the URL with the strongest link profile and ranking history, move genuinely unique sections into it, 301 the rest, then update internal links to point at the survivor directly rather than through the redirect. Leaving the old internal links in place is the most common mistake and it wastes most of the benefit.
Expect a dip. Merged pages commonly lose ground for two to six weeks while Google reprocesses, then recover above the previous combined level. If you have not warned your stakeholder about that dip in advance, you will spend the fourth week defending a decision that is working.
Deleting Safely
Before removing anything, check three things: does it have external backlinks, does it get any conversions, and does it serve a non-search purpose like a support page customers reach from email. Any yes means redirect rather than delete.
For the rest, 410 or 404 is fine and is not a penalty. Redirecting a deleted page to the homepage is worse than deleting it — irrelevant redirects get treated as soft 404s anyway, and you have added a hop for nothing. Redirect to the closest genuinely relevant page or let it go.
Move in batches of 50 to 100 URLs so you can attribute effects. Deleting 800 pages in one afternoon means you will never know which portion of the subsequent movement came from which decision.
Making the Audit Repeatable
A one-time audit fixes a backlog. A quarterly habit prevents one. Every three months, re-pull the same three data sets and check only what changed: new pages that failed to gain traction after a fair window, old pages that decayed, and clusters that have started to overlap again.
Consolidating the inputs helps more than any template. SEO Rocket runs deep full-site crawls with concrete evidence per issue — actual titles, URLs, and H1 text rather than issue counts — alongside keyword data, content gap analysis across up to five competitors, and rank tracking with Search Console and GA4 connected as ground truth, at a flat $50/month. Having crawl, performance, and competitive context in one place removes the VLOOKUP stage that makes most audits a two-week project.
Measure Whether the Pruning Worked
Track four numbers before and 90 days after: total indexed pages, pages receiving organic traffic, non-brand impressions, and organic conversions. The healthy pattern is indexed pages down, pages receiving traffic flat or up, impressions up, conversions up.
Be honest about attribution. If a core update landed in your measurement window, you cannot cleanly separate its effect from yours, and pretending otherwise will eventually embarrass you. Annotate the date, report both, and give it another quarter.
The uncomfortable truth most sites discover is that the audit’s biggest win is not any single rewritten page. It is that after removing 300 pages of noise, the remaining content finally reads like a site about one thing — and both Google and your readers respond to that faster than to anything else you could have done that quarter.
