Content Marketing · September 15, 2026 · 8 min read
Thin Pages: How to Tell Which Ones to Fix, Merge, or Delete
Sorting a crawl export by word count finds the wrong pages - here are the reports and thresholds that put real thin-page candidates in the fix, merge, or delete pile.
By FluxWriter Team
Thin pages are a value problem rather than a length problem, and that distinction decides which ones you fix, merge, or delete. Most audits get it backwards — they sort a crawl export by word count, flag everything under 500 words, and spend the next quarter rewriting pages nobody searches for. This covers the reports that surface real candidates, the three-way triage that follows, and the thresholds behind each call.
Why Word Count Is the Wrong Test
A 250-word page that answers its query completely is not thin. A 1,900-word page that repeats what the first ten results already say is. Length is a proxy, and it fails in both directions — which is why word-count sorts produce audit lists full of pages that were never the problem.
The useful definition is about addition, not volume. A page is thin when it adds nothing a searcher could not get from a result they already saw, or from another page on your own site. Google's guidance on helpful content has pointed at value rather than length for years, and the pages that get quietly demoted tend to be the ones carrying nothing of their own — no first-hand detail, no number that is not already in the other nine results.
That reframing changes what you look for. Instead of asking how long a page is, ask what it uniquely knows.
Scale matters too. On a 60-page site, thin pages cost you a handful of rankings and nothing else. On a site of tens of thousands of pages they also become a crawl problem — enough low-value template pages will compete for crawl attention with the pages you actually care about.
The Reports That Find Thin Pages
Thin pages hide from you, because nobody visits them and nothing ever breaks. Start with the report that lists every URL rather than the popular ones.
Open the Search Console Performance report, switch to the Pages tab, and set the range to the full 16 months. Two thresholds do most of the sorting. Any URL under 10 clicks in 12 months goes on the candidate list. Any URL with 200 or more impressions and single-digit clicks goes on a separate list — that page has demand and a weak answer, which is a different problem entirely.
The Page indexing report gives you the closest thing to a direct verdict. Anything sitting in "Crawled — currently not indexed" was fetched, assessed, and passed over. That bucket is where most genuinely thin pages end up, and reading it takes about ten minutes.
Then add two more passes. Any crawler — Screaming Frog, Sitebulb, Ahrefs Site Audit — reports word count, near-duplicate similarity and internal link counts in one export. Flag anything above roughly 85% similarity to a sibling, and anything with no inbound internal links at all. In analytics, average engagement time under 15 seconds on an informational page is a dependable third signal.
Sort Candidates Into Three Piles
A page's word count tells you nothing about which pile it belongs in. Demand decides it first — a page nobody searches for cannot be rewritten into one that earns, no matter how good the rewrite is. Overlap and inbound links settle the rest.
Score each candidate on those three, then read the row that matches:
| Signal on the page | What it means | Pile |
|---|---|---|
| 200+ impressions, under 10 clicks in 12 months | Real demand, weak answer | Fix |
| Ranks 8–30 for queries another page also ranks for | Two pages splitting one intent | Merge |
| Under 50 impressions in 12 months, no referring domains | No demand, no equity | Delete |
| Zero impressions but a real business job | Never a search page | Keep, noindex |
| Near-duplicate of a stronger sibling | Google already picked one | Merge |
The fourth row is the one that saves you from an expensive mistake. Contact forms, policy pages, thank-you screens and filtered listings have no organic job at all — keeping them live and out of the index is correct, and deleting them because a report called them thin breaks something a customer uses.
Work one page at a time. A library-wide sweep tempts you to apply a single rule to 300 URLs at once, and the rule that is right for 240 of them will quietly wreck the other 60.
The Fix Pile: Demand Without a Good Answer
These are the pages worth your hours, and there are fewer of them than you expect. A page holding 400 impressions and 6 clicks is already being shown for queries you can win — the position is too low, or the snippet is not earning the click.
Fixing means adding what is missing, not adding words. Pull the actual queries for that URL in Search Console and check whether the page answers each one directly. Most thin pages fail on specificity. They describe a category where the searcher wanted a number, a price, a comparison, or a sequence of steps.
Three additions carry most of the gain, and the first is the one nobody can fake. Original detail — your own pricing, your own process, a photo of your own work — is the part no competitor can copy off you. Then put a direct answer inside the first 100 words, where the snippet can reach it. Last, link to the page from your three strongest related URLs.
Then wait. Give a fixed page 8–12 weeks before you judge it, and set a hard floor: if it has not moved after two full quarters, it belongs in one of the other two piles.
The Merge Pile: Two Pages, One Job
Merging pays more than the other two moves and gets skipped the most, because a crawl export can flag a pair but cannot say which URL deserves to live.
For a thin page the decision is narrower, because the thin page is almost never the survivor. It is the donor. Ask one thing of it: is there anything here worth carrying across — a section, a table of your own numbers — or is the URL a weaker restatement of a page you already have?
Pick the recipient on links and age, not on which one reads better — a three-year-old URL with two editorial links beats a cleaner page published last spring. Move what survives into it. Then retire the thin URL and point a permanent redirect at the winner. If nothing survives, it was a delete.
Fix: update every internal link that pointed at the retired URL. Redirects pass the value through intact, but every internal link still aimed at a URL you retired sends readers and crawlers on a pointless extra hop — repoint them while you still have the list open.
One caution before you pair anything up. Two pages on the same subject are not automatically in conflict — a definition page and a troubleshooting page can both rank, and merging them costs you one of the two. Compare the query lists before you commit.
The Delete Pile: No Demand, No Equity
Deleting feels dangerous and rarely is. A page with no impressions and no referring domains has never done anything for you, and the ten minutes it takes to prove that is the entire safeguard.
Run two checks first. Look up referring domains for the URL in whatever backlink tool you own — one editorial link is enough to make redirecting the better call. Then check assisted conversions over 12 months, because some pages get read on the way to a sale and never get credited for it.
Every deleted page still needs a decision about its old address. Redirect when a genuinely close match exists, and let the URL go when nothing is close — an unrelated page pointed at your homepage is read as a soft error and helps nobody.
Volume is a judgement call, not a rule. Clearing 5–10% of a small library in one pass is routine. Clearing 40% of a 5,000-page site in a single week makes attribution impossible if traffic moves, so work in batches of a few hundred and leave 4–6 weeks between them.
FAQ
Is there a word count below which my page is automatically thin?
No, and picking one is how these audits go wrong. A 300-word page that answers a specific question with a number and a date can outrank a 2,000-word page circling the same ground. Use word count to build the candidate list, then judge each page on what it adds.
How long until Google reflects the changes?
Removals register as pages get recrawled, typically 2–6 weeks on a mid-sized site and longer on a large one. Merges take longer, because the redirect has to be seen and the target reassessed. Expect 8–12 weeks before ranking movement means anything, and avoid re-editing inside that window.
Will removing thin pages lift my remaining rankings?
Often, though the mechanism is less direct than it is usually sold. Sites that prune well tend to gain, but they also repaired internal links and consolidated overlapping pages in the same pass. That is correlation, not a lever — the dependable gain comes from merges, where one strong page replaces two weak ones.
The Practical Takeaway
Pull a 16-month Search Console export at page level, flag every URL under 10 clicks, and cross-check that list against "Crawled — currently not indexed". Sort what survives by demand and overlap rather than word count: fix the pages with impressions and no clicks, fold the ones a stronger page already covers into that stronger page, delete the rest once you have confirmed no links and no assisted conversions. Work in batches — 20 to 50 URLs on a small library, a few hundred on a large one — and leave 4 to 6 weeks between them. Start with the 20 lowest-click pages on your site this week.
If you are rebuilding the pages that survive an audit like this, tools like FluxWriter can help produce replacements written to a defined brief at volume — but deciding which URLs deserve one, and which should have been deleted instead, is a call the reports can only inform.