There’s no doubt that AI is extremely powerful. But one thing that it did was to allow absolutely anyone the ability to generate and launch TONS of pages to the web. If there’s one thing that Google has been fighting for way over a decade, it’s index bloat. Sites, sections, and pages that are adding more and more indexing requests, with little to no new value to the web.
To take this one step further, we’re seeing thousands of examples of thin content sections and sites that are launched for ranking purposes, and Google killing them instantly or within days to weeks.
Some people still confuse programmatic SEO with AI spam. They’re not the same thing, and the gap between them is the difference between a page that ranks for years and one that gets de-indexed within days.
Programmatic SEO is the practice of building pages at scale off real, unique data. That’s it. Whether you use AI to help build it faster is a separate question. Using AI to mass-produce content isn’t programmatic SEO. It’s just AI slop at scale.
Here’s what real programmatic SEO looks like:
- Zillow ranks for homes in every location across the US and Canada. Same for apartments, rentals, recently sold properties, and pretty much every address in North America.
- DoorDash ranks for restaurant names.
- TripAdvisor ranks for hotel names, hotels by location, hotels by style.
- Pinterest ranks for design styles.
- Canva ranks for template names.
- G2 ranks for every software review.
- Investing.com ranks for every stock, commodity, and stock-plus-forum query, across 30 languages.
- LinkedIn ranks for each of us. Personal profiles and company names.
- SimilarWeb ranks for the traffic and engagement stats of pretty much any domain on the internet. SEMRush built something very similar.
- Glassdoor ranks for salary data and reviews, by company and by job title. In fact, Glassdoor ranks for most (large) companies. Similarly to Linkedin.
- Trustpilot ranks for company review pages, one per business, built entirely on user-submitted data.
This isn’t just about a strong vs. weak brand. It’s purely about slop vs. valuable.
None of these are “thousands of content pages” in the AI-slop sense. They’re sets of page types, each pulling from a brand’s unique data, each offering something a real person is actually looking for. That’s the part these folks skip: it’s not about the page count, it’s about whether the data behind the page is real and whether it answers actual intent. All while adding an extra layer of value to what the web has seen before.
What I’m seeing more of lately is different: template plus AI rephrase, with nothing underneath it. No proprietary data. No user intent. Just pages for the sake of having a large number of pages. This used to work in the past, but Google had enough. So unless you have a time machine, don’t bother. AI Slop at scale no longer works. And most importantly, it is not Programmatic SEO.
A quick test
Before you call something programmatic SEO, ask:
- Is there a real, unique data point behind every single page? Not a rephrased paragraph. An actual fact that didn’t exist as a page before (a price, a review count, a location, a statistic).
- Would the page still be useful if you stripped out all the SEO intent? Zillow’s listing pages are useful to someone buying a house even if Google didn’t exist. A “Best X in LOCATION” page, built by AI with no underlying data is not.
If the answer to any of these two is no, what you’re building is thin content at scale. It might rank for a few weeks. But it won’t be worth your time.
Why this matters more now, not less
With AI search, the gap between real programmatic SEO and slop is only getting more expensive to ignore. LLMs and AI Overviews are trained to prefer citable, structured, data-backed sources. A Zillow listing or a G2 review page gives a model something concrete to pull from and attribute. People also have more trust issues on the web than ever before, to say the least. So people go to brands they trust. A templated AI-rephrase page doesn’t, because there’s no real data point underneath the words, just a shuffled version of something that already existed somewhere else.
So the brands doing programmatic SEO right aren’t just building pages that rank in classic search. They’re building the data infrastructure that AI models and users increasingly treat as a source of truth. That’s a much bigger prize than a few weeks of thin rankings, and it’s exactly why the distinction is worth defending.
I’m a big fan of real programmatic SEO. It’s one of the most durable growth levers in SEO when it’s done right, and it’s becoming one of the more durable ways to earn a place inside AI-generated answers too. This isn’t a debate about whether AI belongs in the process. It’s a reminder that the data and the intent are what make it programmatic SEO in the first place. Skip those, and you’re just building slop faster, and Google had enough of that %hit.