Discussion starter: a programmatic SEO question, posted with an answer from Niraj Raut to open the thread. If you have dealt with this on a site, reply with what you saw, especially where it differs.
- Technical SEO
Programmatic pages ranked for a year, then dropped in a core update. Which template patterns survived in your experience?
Disclosure: Niraj Raut, who posted this answer, runs the SEO consultancy linked at the end of it.
1 reply
-
adminAdmin
The public evidence here is documentation and visibility analysis rather than controlled tests, but it points one way. The templates that hold up through core updates tend to share one trait: each indexable page is the main place to do something, or to find information that isn’t easily available elsewhere. That can mean live inventory, first-party data, a working tool, or an entity that genuinely differs from its siblings. The templates that fall tend to be derivative layers: the same page with swapped nouns, public data restated, or generated filler wrapped around a thin data point. Aleyda Solis’s analysis of the May 2026 core update drew a similar line, describing the split as primary task destinations versus derivative layers rather than aggregators versus non-aggregators. Her data is domain-level Sistrix visibility for the US and UK, measured from 26 May to 2 June 2026 while the update was still rolling out, so it shows correlation, not cause.
A year of rankings followed by a drop doesn’t prove the pages were always weak. It suggests a broad re-evaluation ranked them below the alternatives, some of which may be new to those results pages. The answer that fits your site better than anyone else’s experience is in your own data: within each template, some pages probably held. Compare them with the ones that fell before you rewrite or delete anything.
Why a year of rankings didn’t protect the pages
Google’s core updates guidance describes these updates as broad changes that don’t target specific sites or individual web pages. In practice, three things can change at once:
- How your pages are assessed. Pages that ranked on close query matching and the site’s overall strength can lose ground when their usefulness is re-weighed against alternatives. That is a plausible reading, not something Google documents for any specific site.
- Who you compete with. The May 2026 analysis found job and travel marketplaces gaining in both markets, while reference aggregators and tools lost ground and established reference brands gained. A programmatic page that sits between the user and the primary source is exposed when the primary source gains.
- Market fit. The same analysis found local-market ecommerce sites strengthening in the UK while US.com marketplaces lost visibility there. A template that serves several countries from one set of pages can lose most where local sites gained.
Template patterns that tend to hold, and patterns that tend to fall
The table combines Google’s documented policies, the May 2026 visibility data and practitioner reasoning. The last column says which is which, so you can weigh each row accordingly.
Template More likely to hold when More likely to fall when Evidence behind it Listings and inventory (jobs, travel, property, products) The page is where the user completes the task: live inventory, useful filters, prices, apply or book Inventory is thin, stale or copied from a feed every competitor also carries Visibility data: job and travel marketplaces gained in May 2026 (correlation) Location pages (“[service] in [suburb]”) Real local differences: providers, prices, availability, local reviews, service areas The place name is swapped into identical copy and every page funnels to one booking form Policy: Google’s doorway abuse examples include pages targeted at specific regions or cities that funnel users to one page Data and reference pages First-party or proprietary data, with a visible method and update date Public data restated from the original source Visibility data: reference aggregators and tools fell while established reference brands gained Comparisons and “alternatives to X” Pairs people actually search for, with hands-on evaluation and structured differences Every possible pair generated, with generic pros and cons Reasoning, overlapping with the scaled content abuse examples Tools and calculators The page does the job better than a one-line answer could A thin wrapper around a single figure that an AI answer can state directly Reasoning A programmatic template is closer to a database with a user interface than to a set of articles. When the database holds something people need, each page is a useful view of it. When it doesn’t, each page is a search results page with extra words. That is close to one of the doorway abuse examples in Google’s spam policies: creating substantially similar pages that are “closer to search results than a clearly defined, browseable hierarchy”.
Features that make a page a destination rather than a derivative
- Each indexable page carries data that its sibling pages don’t.
- The page lets the user complete a task or make a decision, not just read a summary.
- Modules render only when there is data to fill them, so a thin page looks thin to you before it looks thin to Google.
- Pages exist only where search demand and a minimum data threshold are both met.
- The data has a visible source, method and last-updated date.
- Near-duplicate siblings are merged up to the level where the differences are real: region rather than suburb, category rather than every attribute combination.
- No generated filler. Intros and FAQs that restate the heading add length, not information.
The same spam policies define scaled content abuse as many pages generated for the primary purpose of manipulating search rankings and not helping users. The examples include using generative AI tools to generate many pages without adding value for users, scraping feeds to generate pages that provide little value, and stitching together content from different pages without adding value. A core update drop is not a manual action, but templates that resemble those examples are the ones I would audit first.
How to find out which of your own patterns survived
- Set clean windows. Compare a period before the rollout started with one after it finished, leave the rollout itself out, and match weekdays and seasonality where you can.
- Segment by template. Use regex filters on page URLs in the Search Console Performance report, or export through the API, so each template becomes its own series.
- Split each template by page features. Look at data depth (listings, filled fields), pre-update impressions, similarity to sibling pages, backlinks, age and conversions. Because the template is constant, differences between the pages that held and the pages that fell point to the features that mattered.
- Look at who replaced you. Marketplaces, official sources, local businesses, forums and AI answers each suggest a different problem: a quality gap, a shift in which type of source ranks, or a market-fit issue.
- Check indexing. In the Page indexing report, watch for template URLs moving into “Crawled - currently not indexed”. Growth there within one template suggests Google is choosing not to index a larger share of it.
- Check the rest of the site. If non-programmatic sections also fell, the programmatic volume may be weighing on how the site as a whole is assessed. Google’s guidance allows for that effect (see below) but does not quantify it, so treat its size as a hypothesis to test.
What to do with each cohort
Cohort Action Reasoning Real data and real demand, but rankings fell Improve and keep indexed: deepen the data, remove filler, make the task easier These pages have a reason to exist; the job is to make that reason obvious Real demand, thin data Consolidate up a level (suburb to region, for example) and redirect The demand survives at the level where the data is real Thin data and no demand Noindex or remove There is nothing to salvage Near-duplicate siblings Merge into one page Together they read like search results rather than a browseable hierarchy Pages that held Leave the template alone until changes are tested on a sample A global template change can break what survived Google’s core updates guidance describes deleting content as a last resort, only to be considered if you think the content can’t be salvaged. The same page adds that wanting to remove entire sections is likely a sign they were created for search engines first, in which case deleting the unhelpful content can help the good content on the site perform better. For programmatic pages the salvage test is concrete: if there is no data to add and no demand to serve, there is nothing to salvage. Pages with demand and some data are candidates for improvement or consolidation, not deletion.
Hypothetical example:
A home services directory has 14,000 “[trade] in [suburb]” pages. After the update, pages listing five or more verified providers with reviews keep most of their clicks. Pages with zero or one provider and generic copy lose most of theirs, and the trade hub pages slip as well. The team keeps about 2,000 suburb pages above a provider threshold, consolidates the rest into 300 regional pages with redirects, removes the generated introductions, and adds price ranges from its own quote data. It judges the result against the next core update rather than week to week.
What recovery can look like
Google’s guidance says that if a few months have passed without any effect from your improvements, that could mean waiting until the next core update, although you don’t necessarily have to wait for a major one, because Google also makes smaller core updates. Plan in update cycles, not weeks. Also plan for the possibility that full recovery isn’t available. If the results page for a query class has shifted towards primary destinations and AI answers, a derivative template may not come back whatever you improve. At that point the decision is where to put the next year of templates, not only how to fix this one.
If this were my site, I would not copy a pattern that survived on someone else’s site. I would find the pages within each of my own templates that held, identify what they have that the fallen pages lack (data depth, demand or a completed task, for example), and rebuild each template around those features: fewer pages, each with a reason to be the destination.
Need help with this on your own site?
Niraj Raut works with businesses in Nepal, Australia, the UK, Europe and the US on technical SEO, ecommerce SEO, local SEO and AI search.
Add a reply Cancel reply
Replies are open to approved members. Log in or apply to join.