Skip to content

Glossary Programmatic SEO

What Is Programmatic SEO? Examples and Risks

Definition

Programmatic SEO is the practice of publishing many pages from one template and one dataset, each page aimed at a search that follows the same pattern, such as "plumber in Leeds" or "convert PDF to Word". It has nothing to do with programmatic advertising, which is the automated buying of ad space. Whether such a page is useful depends on the data that only that page carries. Without it, the template says the same thing under a different name.

On the upper left, a template with dashed gaps and, below it, a data table with one orange row; both paths meet in a grid of nine identical pages with their gaps filled, one of them in orange, on a dark blue ground.
On this page 13
  1. History and development
  2. How it works
  3. Types and variants
  4. Examples
  5. How to measure it
  6. Advantages and disadvantages
  7. How it differs from related concepts
  8. Tools
  9. Criticism and limits
  10. How to apply it, step by step
  11. Best practices
  12. Common mistakes
  13. How this entry was made
In brief

Programmatic SEO: pages built at scale from a template and a dataset. How it works, what Google has said since 2013, and how to measure it.

On the upper left, a template with dashed gaps and, below it, a data table with one orange row; both paths meet in a grid of nine identical pages with their gaps filled, one of them in orange, on a dark blue ground.

History and development

This is how Google has treated pages produced in bulk over the years:

  • August 2013: Google's Webmaster Tools Help already has a page on automatically generated content, defined as content "that's been generated programmatically". The examples include text translated by a tool without human review, text built with Markov chains, automated synonymizing and content scraped from feeds or search results. The date comes from the oldest copy in the Internet Archive, dated 24 August 2013; the page may be older.
  • 16 March 2015: in its webmaster blog Google announces a ranking adjustment aimed at doorway pages and warns that sites with "large and well-established doorway campaigns" may see a broad impact. The same post adds examples to the definition in its quality guidelines.
  • 18 August 2022: Google launches the helpful content update, which introduces a site-wide signal. According to Google at the time, on sites with relatively high amounts of unhelpful content, any content was less likely to perform well.
  • 8 February 2023: Google publishes its guidance on AI-generated content. It points out that automation "has long been used to generate helpful content, such as sports scores, weather forecasts, and transcripts", and says automation becomes spam when its main purpose is to manipulate rankings.
  • 5 March 2024: Google adds scaled content abuse to its spam policies, together with expired domain abuse and site reputation abuse. Google says the new policy applies whether the content is produced by automation, by people or by a mix of both. In the same month, according to Google's guide to its ranking systems, the helpful content system "evolved and became part of our core ranking systems". It no longer runs as a separate system.
  • 26 April 2024: Google reports that the rollout finished on 19 April and that search results now show 45% less low-quality, unoriginal content. Google had expected 40%. Both figures are Google's own measurement.

The version of the spam policies in force when this entry was written is dated 28 August 2026.

How it works

A programmatic SEO project has four parts, and each one can fail on its own.

The first is the search pattern. A head term is combined with a modifier that changes from page to page: a place ("employment lawyer in Bristol"), a comparison ("Notion vs Trello"), an integration ("connect Shopify to Mailchimp") or a conversion ("50 dollars in euros"). Each combination is a long-tail keyword with little search volume on its own. The total across all combinations is what justifies the work. A pattern is only worth building if a meaningful share of its combinations is actually searched, and that has to be checked combination by combination. The volume of the head term says little about the volume of each variant.

The second part is the dataset. Each row becomes a page and each column a piece of information shown on it: price, address, opening hours, measurements, ratings, availability. This is where most projects are decided. If the only column that changes is the name of the city or the product, every page will carry the same information, and the template will be the only thing a visitor reads. Good datasets usually come from the business itself: its own prices, its own stock, its own completed jobs, its own reviews, or measurements it took.

The third part is the template. It sets where each field goes, which fixed text surrounds it and which blocks appear only when there is data for them. A well-built template hides a block when the field is empty instead of filling it with generic sentences. A CMS or a site builder then fills the template with each row of the dataset.

The fourth part is publishing. URLs follow the pattern (for example /services/plumbing/leeds/), pages link to their siblings and to the parent category through internal linking, and each template gets its own XML sitemap. The separate sitemap is not only for crawling. It is what makes it possible to measure each template later in Search Console.

Its own data in each row, or only a different name

Types and variants

PatternExample queryWhat each page has to add
Locations"dentist in Bristol"Address, service area, opening hours, prices or reviews for that place
Comparisons"Asana vs Jira"Measured or documented differences between the two products
Integrations"connect Stripe to HubSpot"What can be automated and the steps to do it
Calculators and converters"100 pounds in euros"The result, with its source and date
Directories and glossaries"what is a CRM"A record or definition written for that entry

Source: our own overview; the example queries are illustrative.

Location pages carry the most risk, because the line between a useful local page and a doorway page is thin. The doorway page entry covers that pattern in detail. Comparison and integration pages depend on how much the author knows about both products.

Examples

A worked example with round numbers shows how the decision is made.

A renovation company works in 50 cities and offers 20 services. The template allows 50 × 20 = 1,000 pages. Before publishing, the company checks the search demand for each combination in a keyword tool. Suppose only 140 combinations have any searches, and for 90 of them the company has data of its own: jobs completed in that city, real lead times, photos of the work. The sensible start is those 90 pages. The other 910 would mostly repeat the template, and the 50 with demand but no data need that data before they deserve a page. The numbers are illustrative. What carries over is the method: count demand per combination, count own data per combination, and build only where both exist.

Google gives its own examples of helpful automation: sports scores, weather forecasts and transcripts (Google Search Central blog, 8 February 2023). In each case every page carries a new fact that the template does not contain: the score of that match, the forecast for that place and day, the words of that recording.

How to measure it

Measure each template separately. That is why each template should have its own folder in the URL and its own sitemap.

In Google Search Console, the Page indexing report can be filtered by a specific sitemap. This shows how many of the submitted URLs are indexed and why the rest are not. "Crawled - currently not indexed" means, according to Search Console Help, that Google crawled the page but did not index it; Google says this is common and "doesn't necessarily indicate an issue". For a new template, however, a high share of pages in that status is a reason to look at what each page actually adds. "Discovered - currently not indexed" means Google found the URL but has not crawled it yet. The example list of URLs in the report is limited to 1,000 items.

In the Performance report, the page filter with "URLs containing" or with a regular expression isolates the template's folder. Three figures are usually enough to judge a template:

  • The share of submitted URLs that are indexed.
  • The share of indexed pages that get any impressions.
  • The number of pages with no impressions at all after 90 days.

The 90 days are our own working threshold, not a Google recommendation. Compare the figures between templates: a template whose pages are indexed but never shown is telling you something different from one whose pages are not indexed in the first place.

Advantages and disadvantages

CaseAdvantageDisadvantage
Marketplace or portal with its own data (prices, stock, reviews)Each page carries information nobody else hasKeeping the data current costs as much as creating it
Service business with city pagesReaches local searches with little competitionWithout data for each place, the pages start to look like doorway pages
Glossary or directoryA fixed structure makes many entries easier to maintainIf entries have no content of their own, the risk of thin content grows

Source: our own overview.

How it differs from related concepts

ConceptRelationship to programmatic SEO
Programmatic advertisingSame adjective, different field: automated buying of ads
Scaled content abuseThe spam policy that applies when pages made in bulk add no value
Doorway pagePages for similar searches that only funnel visitors to another page; city pages without their own data can fall into this
Thin contentA possible result of programmatic SEO done badly
Duplicate contentWhat appears when many pages from one template say almost the same thing
Landing pageA single page built by hand for a campaign or a search

Source: ZDS glossary entries and Google's spam policies, accessed 17.11.2026.

Tools

In alphabetical order, with the description each vendor gives on its own website (accessed 17.11.2026):

  • Airtable: tables-based database; the company presents it as a platform for building apps on top of large data sets. Often used to hold the dataset behind the pages.
  • n8n: a workflow automation platform that combines AI with business process automation, in the vendor's words.
  • Webflow CMS: content collections with reusable templates; the layout is designed once and new content follows it automatically.
  • Whalesync: two-way sync between spreadsheets or databases (Airtable, Google Sheets, Notion) and site builders such as Webflow, WordPress or Wix.
  • WP All Import: a WordPress plugin that imports CSV, XML, Excel or Google Sheets files into any post type or custom field.

Criticism and limits

Google describes scaled content abuse as many pages "generated for the primary purpose of manipulating search rankings and not helping users", and adds that this applies "no matter how it's created" (Spam policies, version of 28.08.2026). Its listed examples include using generative AI to produce many pages without adding value, scraping feeds or search results and transforming them automatically, and creating many pages that make little sense to a reader but contain search keywords. The policy judges the pages that result, whatever method produced them.

Among the questions Google suggests for reviewing your own content is this one: "Are you using extensive automation to produce content on many topics?" (Creating helpful, reliable, people-first content, version of 05.10.2026). Since March 2024 these questions describe what Google's core ranking systems look for, because the helpful content system is now part of them.

There is also a technical limit. Thousands of pages without value fill the index and make Google crawl URLs that nobody searches for. Google says its guide to crawl budget is meant mainly for sites with more than a million unique pages, or more than 10,000 pages that change daily, and for sites with a large share of URLs in the "Discovered - currently not indexed" status. Google adds that these numbers are a rough estimate, not exact thresholds. For other sites, the same guide says keeping the sitemap up to date and checking the Page indexing report regularly is adequate.

Generated pages also age badly. A price or opening time that is not updated turns a page into a source of wrong information, and the more pages there are, the harder it is to notice.

How to apply it, step by step

  • Choose one search pattern and check demand for each combination, not only for the head term.
  • Collect, for each row, at least one piece of information of your own that changes from page to page.
  • Design the template with blocks that disappear when a field is empty, instead of filling them with generic text.
  • Publish a test batch of about 50 pages with its own sitemap.
  • Measure indexing and impressions for that batch over several weeks.
  • Scale up only if the test works; if it does not, improve the data or reduce the number of pages.
  • Merge or remove pages with neither impressions nor data of their own, redirecting them to the parent page.

Best practices

  • One folder and one sitemap per template, so each can be measured separately.
  • A visible update date on pages whose data changes.
  • Links between sibling pages and from each page to its parent category.
  • A manual review of a sample of pages before every expansion.
  • If generative AI writes part of the text, an editor checks it and the page says so.

Common mistakes

  • Publishing every possible combination, whether or not anyone searches for it.
  • Changing only the name of the city or product from one page to the next.
  • Having no plan for keeping the data up to date.
  • Measuring the project as a whole instead of each template.
  • Confusing programmatic SEO with programmatic advertising when looking for a provider or for information.

How this entry was made

This entry was written in November 2026 with the help of AI. The information comes from Google Search Central documentation and blog posts, Search Console Help and the vendors' own websites, all accessed on 17.11.2026. It was reviewed before publication and contains no client data.

Manuel Riveiro Rodriguez CEO & Digital Strategist

A technical audit covers this and everything else in one pass.

Request an audit

Frequently asked

Does Google penalize programmatic SEO?

Google does not use the term in its policies. What it acts against, since 5 March 2024, is scaled content abuse: many pages made mainly to rank and with little value for users, whether they were produced by automation, by people or by both.

What is the difference between programmatic SEO and programmatic advertising?

Programmatic SEO builds web pages for organic search from a template and a dataset. Programmatic advertising buys ad space automatically, impression by impression. The two share an adjective and nothing else.

What tools are used for programmatic SEO?

You need a dataset, for example in Airtable or a spreadsheet, a CMS with templates, such as Webflow or WordPress with an importer, and sometimes a sync or automation tool such as Whalesync or n8n.

Can you do programmatic SEO with AI?

Yes, but Google lists using generative AI to create many pages without adding value as an example of scaled content abuse. AI can write the text around the data. The data that makes each page worth having has to come from somewhere else.

What are some examples of programmatic SEO?

Google itself names sports scores, weather forecasts and transcripts as helpful automation. Other common patterns are city pages, product comparisons, integration pages and currency or unit converters.

Sources

  1. Google Search Central, "Spam policies for Google web search", version of 28.08.2026.
  2. Google Search Central Blog, "What web creators should know about our March 2024 core update and new spam policies", 5 March 2024.
  3. Elizabeth Tucker, "New ways we're tackling spammy, low-quality content on Search", Google blog, 5 March 2024, with update of 26 April 2024.
  4. Google Search Central, "A guide to Google Search ranking systems", version of 10.12.2025.
  5. Google Search Central Blog, "What creators should know about Google's August 2022 helpful content update", 18 August 2022.
  6. Google Search Central Blog, "Google Search's guidance about AI-generated content", 8 February 2023.
  7. Google Search Central Blog, "An update on doorway pages", 16 March 2015.
  8. Webmaster Tools Help, "Automatically generated content", Internet Archive copy of 24.08.2013.
  9. Google Search Central, "Creating helpful, reliable, people-first content", version of 05.10.2026.
  10. Google Search Central, "Google Search's guidance on using generative AI content on your website", version of 01.10.2026.
  11. Google, "Crawl Budget Management", version of 22.07.2026.
  12. Search Console Help, "Page indexing report".
  13. Search Console Help, "Performance report (Search results): Advanced filtering and comparison".
  14. Airtable, home page.
  15. n8n, "AI Workflow Automation Platform".
  16. Webflow, "Build scalable websites visually with the Webflow CMS".
  17. Whalesync, "Real-time data syncing between your favorite tools".
  18. WordPress.org, "WP All Import".