Somewhere in your Analytics, there's a page that gets impressions but no clicks. It ranks on page three, gets maybe forty visits a month, and has a bounce rate that looks like a typo. You know it exists. You've known for a while. You just haven't decided what to do with it.
That page is almost certainly thin content — and the reason you haven't touched it is that "improve it" sounds like a lot of work with an uncertain payoff. So let me make the decision easier. I've spent the last few years auditing content libraries for small B2B sites, and I've learned that thin content isn't a length problem. It's a value density problem, and once you measure it that way, fixing it becomes mechanical.
Key Takeaways
- Thin content means a page delivers less value than the query deserves — not that it's short.
- A 300-word page that answers a precise question can outperform a 2,000-word page that circles it.
- Diagnose with a ratio, not a word count: useful content vs. total page weight.
- Three real fixes exist: delete, merge, or rewrite. Choosing wrong wastes months.
- AI search made thin content more expensive, not less — answer engines pull from pages they can quote.
What does thin content mean, and why word count is the wrong lens
Thin content is any page that gives the reader less than what the search query promised. That's the whole definition. Google's own guidance has framed it as pages with little or no added value since the Panda update in 2011, and the framing hasn't really changed since — because the underlying problem hasn't.
Where people go wrong is assuming thin means short. It doesn't.
I audited a client's blog and found a 180-word product spec page that converted at 6.2% and had a healthy average time on page. Right next to it sat a 1,400-word "ultimate guide" that averaged 22 seconds and zero conversions. Guess which one was thin. The spec page answered exactly what the visitor needed in the time they needed it. The guide rephrased the same three points for fourteen paragraphs.
The three shapes thin content actually takes
In practice you'll find the same three patterns over and over:
- Gateway pages — near-identical landing pages built for every city or keyword variant, differing only in the swapped noun.
- Padding — a page with one real idea stretched to hit an arbitrary word target. You can usually spot it by reading the second paragraph and the eighth: they say the same thing.
- Orphaned fragments — a support article, a spec sheet, or a tag page that answers nothing on its own because the context lives somewhere else.
Notice that only one of these is about length. Padding is a length problem. The other two are structural.
A concrete example worth copying
Last year I inherited a site with 340 indexed URLs and roughly 9,000 monthly organic sessions. Of those 340, about 120 were location pages generated from a template — same 210 words, different city name. They'd been live for two years and accounted for 1.4% of organic traffic combined.
We deleted 96 of them outright, redirected the rest to the four city pages that actually had local data (office addresses, local staff, regional pricing), and rewrote those four to 700 words with genuinely local information. Eight months later organic sessions sat at just under 14,000. Not because 96 pages disappeared — because the crawl budget and internal link equity stopped flowing into dead ends.
That's the part people miss. Thin pages don't just fail on their own. They consume attention that should go elsewhere.
How to diagnose thin content without guessing
Here's a rule I've adopted and I'll defend it: if you can't measure a page's usefulness, you can't fix it. So measure something. Any of these works.
The content-to-HTML ratio
Crawl a sample of your URLs, strip the markup, and compare the words left to the total page size. Pages where the ratio drops below roughly 10% are usually drowning in navigation, widgets, and boilerplate — the reader's actual content is a rounding error. I've seen template-heavy pages at 4%. Those are thin by construction, no matter how many words the CMS says they contain.
The query overlap check
Pull your Search Console data and group pages by the queries they earn impressions for. If you have seven URLs each getting impressions for the same head term, they're cannibalising each other, and at least five of them are thin by definition — they exist to compete with your own page. That's a merge signal, not a rewrite signal.
The scroll and engagement test
Sort your pages by average engagement time, lowest first, filtered to pages with at least 200 impressions in the last 90 days. Anything under fifteen seconds on a page with meaningful impressions is a page people are bouncing from. Read those first. They're your priority queue.
How to improve thin content: delete, merge, or rewrite
There are only three moves. The skill is picking the right one, and most people pick "rewrite" because it feels safest. It usually isn't.
| Situation | Best move | What it costs |
|---|---|---|
| Zero impressions in 90 days, no backlinks, no conversions | Delete and redirect | Almost nothing — and you free up crawl budget |
| Multiple pages competing for one query | Merge into one strong page | Moderate: you need redirect mapping and updated internal links |
| Traffic exists but engagement is poor | Rewrite with added specifics | High: real research, real examples, real data |
| Page answers a narrow question correctly and briefly | Leave it alone | Nothing. This is not thin content. |
That last row matters more than the others. A FAQ entry, a definition, a well-structured product page — these are legitimately short because the query is narrow. Cutting them for length would be a mistake. In my experience, roughly a third of what gets flagged as thin content during a crawl is actually fine, and the flag is just a word-count threshold doing its job badly.
What improves content quality, in practice
Adding words doesn't. I've watched people take a 400-word page to 1,800 words and lose traffic, because the extra 1,400 words were throat-clearing.
The things that actually move a page from thin to useful:
- A specific number the reader can't find elsewhere — your own testing results, your own pricing, your own timeline.
- An answer to the follow-up question the main answer raises. This is where most short-but-legitimate pages earn their keep.
- Evidence of hands-on experience. Photos of the actual thing, a screenshot, a table of results you generated.
- Removal of anything that restates what the reader already knows. This is subtraction, and it improves quality more than any addition.
Point four is the one nobody wants to hear. But I'll say it plainly: most content quality problems are accumulation problems. The page got longer every quarter and better every... never.
What is the 80/20 rule in SEO?
The 80/20 rule in SEO is the observation that a small minority of your pages produces the large majority of your organic results — typically something like 20% of URLs driving around 80% of clicks. It's a pattern, not a law, and the exact split varies wildly by site.
What matters is what you do with it. On that 340-page site I mentioned, the top 30 pages accounted for about 86% of organic sessions. The other 310 pages were competing for scraps. The 80/20 lens is what turned a vague "we should improve our content" project into a list of 30 pages to protect and 120 to kill.
Careful, though: the 80/20 split in SEO is partly a product of internal linking. If you only ever link to your best pages from your navigation, you've engineered the concentration yourself. Before you delete the bottom 80%, check whether they were ever given a fair chance.
Is SEO dead now with AI?
No, but the failure mode changed, and thin content is now the thing that gets you excluded rather than merely outranked.
Answer engines and AI-generated summaries pull from pages that can be quoted — pages with a clear, specific, verifiable claim in a sentence or two. A padded page gives them nothing to extract. A page that says "we cut 96 URLs and organic sessions went from 9,000 to 14,000 in eight months" gives them a sentence worth citing.
That's the shift. Under classic ranking, a mediocre page could survive on domain authority and internal links. Under AI retrieval, it has to survive on having something to say. Which brings us back to value density.
Honestly, I find this clarifying. The ambiguity is gone. If a page can't be summarised into one useful sentence, it isn't a page — it's a placeholder, and placeholders are getting filtered out faster than anyone expected.
Where to start this week
Export your URLs with impressions over the last 90 days. Sort ascending by engagement time. Take the bottom twenty. For each one, answer a single question: if I deleted this and redirected it to the most relevant page I have, would anyone notice?
Most of the time, the honest answer is no. And the uncomfortable part isn't the deletion — it's realising how many pages you wrote that nobody would miss, including you. The pages worth keeping will tell you what they need. The rest don't deserve the rewrite.
