Marketing Strategy & OperationsChecklist3 min readUpdated September 2026

Programmatic SEO: How to Avoid Thin, Templated Pages

A programmatic SEO page is thin when swapping the keyword out leaves the page reading the same. If every URL shares one template and differs only by a city, product, or tool name, Google can treat the set as scaled, low-value content. Each page needs something a visitor can't get from the others.

Google's spam policies specifically name scaled content abuse, meaning pages generated in volume mainly to rank rather than to help people. The fix isn't to avoid templates. It's to make sure each page has its own data, answers its own question, and would still be worth visiting if search traffic vanished.

Vendors Covered in this Article

Disclosure: We may earn a commission if you buy through some links on this page. It doesn't change what we recommend.

What makes a programmatic page thin?

Run a swap test. Take two pages from your set, replace the entity name (the city, the software, the job title) with a placeholder, and compare them. If the remaining text is nearly identical, the pages are thin.

Other signals to look for:

  • The only unique element is a name, a number, or an auto-filled sentence.
  • The page answers a question nobody asked, such as "plumbing services in a town with 400 residents."
  • Visitors land, find nothing specific, and leave within seconds.
  • Sections exist to hit a word count and say nothing a reader could act on.

If you can't explain in one sentence what a visitor learns from this page that they don't learn from its siblings, it doesn't earn its own URL.

Checklist: what each page needs before you publish it

Use this as a gate. A page that fails two or more items stays out of the index.

  • A data field that's genuinely different per page, such as local pricing ranges you collected, real inventory, or measured comparisons.
  • A specific answer near the top for the query the page targets.
  • Copy written or reviewed by a person who knows the topic, at least for the parts that vary most.
  • At least one element that isn't text: a table, a map, a calculator, a chart built from the page's own data.
  • Internal links to and from related pages, so the page isn't an orphan.
  • A clear reason to exist in the site's navigation or hub structure.

The data field is the one that matters most. Templates are fine when the data underneath is real.

How to test a template before you publish hundreds of pages

Don't launch the full set. Follow this sequence:

  1. Build 20 to 30 pages that cover the range of your data, including a few weak entities.
  2. Read ten of them cold. Would you be satisfied if you'd searched that phrase?
  3. Submit them and watch Search Console for two to four weeks: are they indexed, and do they get impressions?
  4. Check whether the pages are flagged "Crawled, currently not indexed" or "Discovered, currently not indexed." Large numbers of either suggest Google isn't seeing enough value.
  5. Improve the template, then expand in batches, checking each batch before the next.

Say you're building pages for "invoice template for [profession]." If accountants and photographers get the same template with the noun swapped, expand only the professions where you can add real line items and clauses for that profession.

How to handle entities with no real data

Some combinations will have nothing useful to say. You have three honest choices: don't generate the page, generate it but keep it out of the index with a noindex tag until it has content, or fold it into a parent page (a state page rather than 30 tiny town pages).

Skip the fourth option, which is padding the page with generic filler so it looks complete. That's how a good set gets dragged down by its weakest pages.

Also avoid auto-generating text with a model and publishing it unreviewed. AI-assisted drafting can be fine, but pages with no human check tend to repeat each other and state things that aren't true for the specific entity.

How to audit and prune a set that's already live

Pull all URLs in the set, then join Search Console data (impressions, clicks) with your analytics (engagement, conversions). Sort ascending.

For pages with no impressions after several months, decide one of three outcomes: improve with unique data, merge into a stronger page with a 301 redirect, or remove and let it return a 410. An Ahrefs or Semrush crawl helps you check thousands of URLs at once, and their keyword tools show which long-tail queries a template is actually catching.

Re-check indexing quarterly. A set that starts strong can decay as data goes stale, so put a refresh date on the calendar.

Executive Capability Standard

What Good Looks Like

Every indexed programmatic page carries data unique to it, answers its own query, and is linked into the site, and weak entities are noindexed, merged, or never generated.

Building The Capability (5-Stage Skill Ladder)

1. Learn:Read Google's guidance on scaled content and helpful content, then run the swap test on two of your own template pages.
2. Do Manually:Launch a pilot of 20 to 30 pages, read a sample cold, and check Search Console indexing status weekly for a month.
3. Delegate:Assign someone to research and load the unique data field for each entity, and to review copy before pages go live.
4. Automate:Generate pages from a data source with rules that keep low-data entities noindexed, and track cohorts of pages in a rank-tracking tool.
5. Buy:Purchase or license a reliable dataset for the field that varies per page, rather than scraping or inventing values.

How to Get Started

Disclosure: We may earn a commission if you buy through some links on this page. It doesn't change what we recommend.

Frequently Asked Questions

Is programmatic SEO against Google's guidelines?

No. Templates and generated pages are fine when each page gives visitors real, distinct value. Google's spam policies target scaled content made mainly to manipulate rankings. Pages built on unique data that answer a specific query are a normal, acceptable approach.

How many programmatic pages should I launch at first?

Start with 20 to 30 that cover the range of your data. Watch indexing and impressions for a few weeks, fix template problems, then expand in batches. Launching thousands at once makes it hard to tell what's working.

Should I noindex weak programmatic pages?

Yes, if the page may become useful later. Noindex keeps it out of search while you add data. If the entity will never have meaningful content, merge it into a parent page or don't create it.

Can AI write programmatic SEO pages?

It can draft them, but publish only after a person checks the facts and confirms each page differs meaningfully from the others. Unreviewed AI text tends to repeat itself across pages and state things that aren't true for the specific entity.

About the numbers

This guide doesn't quote a sourced benchmark. Figures in it are estimates or general guidance, so check them against your own numbers.

Related Guides