You have 200 articles. You submitted to AdSense. It came back rejected, and the email or the Sites tab says some variant of “low value content” (Google’s own wording is “we believe that there isn’t enough original, rich content that would be of value to users”). Your gut says: “But I have 10x more articles than the threshold people talk about.”
Fastest fix: stop counting pages. AdSense’s reviewer samples your URLs and judges the average quality, so the bottom 40% of weak pages sinks the strong ones. Cut or merge the thin pages, deepen the strong ones with genuine first-hand value, wait ~2-3 weeks for a re-crawl, then request review again. 20-40 pages with original insight beat 200 templated / AI-paste / paraphrase-of-Wikipedia pages every time.
This article shows how to find which pages are dragging the site down and the surgical fix path — usually delete or merge, not “add more.”
What Google actually checks (June 2026)
Two policies cause almost all of these rejections, and they have separate official names:
- Insufficient content — “Your site was found to have too little text, and/or your site was deemed to be ‘under construction.’” Page count and depth.
- Content quality / low value content — “There isn’t enough original, rich content that would be of value to users.” Originality and usefulness.
Since the March 2026 core update, Google’s crawlers weigh two things heavily and AdSense review mirrors them:
- E-E-A-T — Experience, Expertise, Authoritativeness, Trustworthiness. First-hand experience (“Experience”) is the part AI-paste sites fail.
- Information Gain — does this page add anything the first page of Google results doesn’t already say? If 100,000 sites have the identical “How to start a blog” steps, your version is low value by default.
There is no official minimum word count, but as of June 2026 pages under ~400 words are routinely flagged, and “quality, not length” is the real bar.
Common causes
Ordered by hit rate, highest first.
1. Programmatic SEO pages with light substance
Each page covers one variant of a template (“Best AI tools for X,” “Best AI tools for Y,” “Best AI tools for Z”) with only the keyword changing. Reviewers see this immediately, and it reads as “no Information Gain.”
How to spot it: Open 5 pages from the same category at random. If they share the same paragraph structure with mostly different nouns, you have programmatic-light pages.
2. AI-generated articles with no unique angle
ChatGPT-default cadence: an “In today’s world” opener, three subheadings of ~200 words each, a generic conclusion. Even if grammatically perfect, reviewers detect the pattern, and it fails the “Experience” leg of E-E-A-T because nobody actually did the thing.
How to spot it: Read your articles out loud. If they all sound like the same author — and that author is ChatGPT — your content reads as AI-paste.
3. Near-duplicate articles within the site
You wrote “How to use ChatGPT for writing” and “Best ways to write with ChatGPT” — they cover 80% the same ground. Reviewers count these as one and a half pages, not two, and Search Console often files the weaker one under “Duplicate without user-selected canonical.”
How to spot it:
# pairwise compare article opening paragraphs to surface repeated structures
for f in src/content/articles/en/*.mdx; do
head -c 500 "$f"
echo "---"
done | sort | uniq -c | sort -rn | headSpot repeated opening structures.
4. Summaries of freely available info, no interpretation
“What is X? X is…” with the answer pulled from Wikipedia or product docs. No personal experience, no opinion, no original framing — zero Information Gain.
How to spot it: For each article, ask “what would a reader get here that they wouldn’t get from the first Google result?” If the honest answer is “nothing,” it’s a thin page.
5. Tag and category archive pages have no content
Your /tag/ai-tools/ page is just a list of links with no intro paragraph. AdSense’s crawler sees this as a thin auto-generated index page, and many tags create many such pages.
How to spot it: curl -s https://yoursite.com/tag/ai-tools/ | wc -c. Under ~5000 bytes of HTML usually means a thin archive.
6. Auto-generated localized variants
You used machine translation to produce 200 pages in 5 languages. Each language version is reviewed and judged on its own, and raw MT with no editing counts as thin in every locale.
How to spot it: Compare your /en/ vs /zh/ vs /es/ content. If the non-English versions are direct MT with no human editing, those count as thin.
Which bucket am I in?
| Symptom you can see | Likely cause | Primary fix |
|---|---|---|
| Many URLs share the same skeleton, only nouns change | Programmatic-light (#1) | Add 1 original element each, or cut |
| Every article has the same AI cadence/voice | AI-paste (#2) | Rewrite from real experience |
| Two titles, same topic | Near-duplicate (#3) | Merge into one + 301 |
| ”What is X” with no opinion | Summary-only (#4) | Add interpretation or delete |
/tag/ and /category/ pages are bare lists | Thin archives (#5) | Add intro + curate, or noindex |
| Non-English pages are raw MT | Localized thin (#6) | Human-edit or noindex that locale |
Shortest path to fix
Step 1: Score every page
Create a spreadsheet with columns:
| URL | Word count | Original elements (0-3) | Inbound links | GSC status | Action |
|---|
Original element = your data, your screenshot, your opinion, your worked example, a quote from an interview you did. Generic numbered lists don’t count. Pull GSC status from Search Console -> Indexing -> Pages: any URL sitting under “Crawled - currently not indexed” is one Google itself has already judged not worth indexing, so treat those as thin-by-default.
| Original elements | Word count | Action |
|---|---|---|
| 0 | any | Delete or noindex |
| 1 | < 500 | Expand or merge |
| 1 | 500-800 | Improve before keeping |
| 2-3 | 800+ | Keep |
Step 2: Remove or noindex the bottom 30-50%
Don’t just unpublish — actually decide:
- Delete if there’s no way to make it useful (and return
410 Goneor404for the old URL). - Merge if it overlaps with a stronger article on the same topic, then
301the weaker URL to the survivor. noindexif you want the URL to stay live but not be in AdSense’s content sample.
For Astro, add draft: true to the frontmatter to drop the page from the build, or emit <meta name="robots" content="noindex"> in the page head if you need the URL to stay reachable but out of the index. Do not block the URL in robots.txt if you also want noindex to be seen — a disallowed URL can’t be crawled, so Google never reads the noindex.
Step 3: Deepen the top 20%
For your strongest articles, add original elements:
- Run an actual experiment, screenshot the result.
- Interview one practitioner, quote them.
- Build a comparison table with your own numbers.
- Write a contrarian section: “Why I disagree with the common advice.”
Aim for at least 2 original elements per article and enough depth to actually answer the query — usually 1,500+ words for a how-to, but depth beats raw length.
Step 4: Fix tag and category archive pages
Each archive page should have:
- A 100-200 word intro paragraph explaining what this collection is.
- A clear curation rationale.
- Hand-picked “featured” articles, not just a chronological list.
For tags with < 3 articles, merge them into broader tags or delete them. Search Console will otherwise reflag thin tags as low-value URLs.
Step 5: Wait, then request review
After cleanup, wait at least 2-3 weeks so AdSense and Google can re-crawl. Rapid resubmissions look like manipulation and don’t give the crawler time to see the new state. Newer domains (under ~3 months old) and previously rejected sites get a deeper, slower review, so don’t panic if it takes the full window.
In AdSense, go to Sites, open the site, and use Request review (the same Sites card shows the status, which flips to Ready once it passes). The typical decision lands in 1 day to 2 weeks, most often 3-7 business days. While the review is pending, avoid making more major changes.
Step 6: Confirm it’s fixed
You’ll know the cleanup worked when:
- AdSense -> Sites shows the site at Ready status (ads can serve).
- Search Console -> Indexing -> Pages shows the “Crawled - currently not indexed” bucket shrinking and your kept URLs moving to “Indexed.”
- After approval, watch AdSense -> Reports for per-URL coverage: pages with chronically low ad coverage are usually the next ones to lose value — refresh or remove them too.
When this is not on you
AdSense reviewers are partly automated, partly human. Some legitimate sites are rejected initially and approved after improvements. A first “low value content” rejection is common and is usually fixable in 1-2 cycles.
Easy to misdiagnose as
Adding 50 more pages does not help if they share the thin character. The fix direction is down, not up: cut the weakest pages, improve the strongest. Counter-intuitive but reliable.
Prevention
- Decide on a unique value angle (your data, your perspective, your audience) before producing 100+ pages.
- Don’t equate page count with site value — AdSense doesn’t.
- Audit content depth quarterly: if any page has been live for 6+ months with zero impressions in Search Console, refresh or remove.
- For programmatic / templated content, write the template once at full depth, then customize each instance with at least one original element.
- Tag and category pages need intros, not bare link lists.
FAQ
- What’s the minimum article count? Not officially defined. 20-40 strong pages with original elements often pass; volume below ~15-20 solid posts tends to draw a slower, harder review.
- Is there a minimum word count? No official number. As of June 2026, pages under ~400 words are frequently flagged, but Google grades on usefulness, not length — a tight 700-word page with first-hand data beats a padded 2,000-word one.
- Can I improve and resubmit immediately? You can, but wait 2-3 weeks first so the re-crawl reflects your changes, then expect another full review cycle (typically 3-7 business days, up to 2 weeks).
- Will deleting pages hurt my SEO? Deleting genuinely thin pages usually helps — Google rewards a higher average quality. Merge-and-301 where there’s any link equity or traffic worth keeping; hard-delete only true dead weight.
- Does
noindexcount the same as deleting for AdSense? For the content-quality sample, yes — anoindexpage is excluded from what reviewers judge. It still exists for users, which is the point when a URL has a reason to stay live.