Good Vs. Bad pSEO: Learnings From A Decade Of Building Scalable Content Engines

Search Engine Journal by 10 min read 511x views
Good Vs. Bad pSEO: Learnings From A Decade Of Building Scalable Content Engines

Share Post

Generative AI has removed the obstacle to publishing at scale. A brand can create thousands of pages in an afternoon, and equal be bold adequate to call it programmatic SEO. For many brands, it worked, bringing in enormous hunt traffic and integrated revenue.

Management teams additionally buy in easily, since period to market is abbreviated and manufacturing expenses are low.

Those identical two things are now what create it fail. Aggressive programmatic SEO initiatives akin these are collapsing or example manual penalties, since no one asks whether these pages deserve to exist.

Google has been apparent concerning this. On March 5, 2024, it strengthened its scaled satisfied abuse guideline to use whether automation, humans, or a blend produces the pages, and afterward reported 45% small low-quality, unoriginal satisfied in results.

On Sept. 7, 2026, John Mueller replied on Bluesky to a location that generated pages from all blend of domain names, technologies, and attributes. He wrote that programmatic SEO of that benevolent frequently leads to a location that’s spam, borderline spam, or low quality, and that pages already published may keep shaping how Google judges the entire site.

Screenshot from Bluesky, November 2026

I have been construction programmatic SEO systems since 2016, from fixed templates to natural tongue generation engines specified as Rosae NLG to AI-assisted pipelines. Some compounded for years and became true case studies for me. Some unsuccessful fairly quickly. I have taken that activity to gathering stages alongside the way, most lately in a meeting on scaling pSEO at the Shenzhen SEO Conference.

What we see today is bulk AI publishing disguised as programmatic SEO. The genuine item starts alongside a dataset that gives all leaf item a visitant can use. This part draws the row between the two.

pSEO Is A Publishing Method, Not A Content Type

Programmatic SEO is a way of publishing, not a benevolent of content. It has three parts: a repeatable hunt pattern, a organized dataset that makes all leaf distinctive and value-driven, and a template that answers the query for all variation.

Take a prosperity converter. The form is “convert X to Y,” the dataset is a live charge for all prosperity pair, and the template turns all row into a page. Nobody writes thousands of pages by hand. The data makes all leaf different. And you publish lone as many pages as your data can support.

Lucia Lecesne, the Marketing Lead of Wise, which runs exactly that benevolent of program, says:

“At Wise, we concentration on genuine client utility: solving a issue that users visage on their journey to rotate into customers. Programmatic SEO can afterward create that resolution applicable and accessible throughout the many hunt queries and modifiers that portion the identical problem, in another words, tailoring it to all customer’s particular hunt context. It should never be a shortcut to publishing ample volumes of low-quality pages.”

Mass AI publishing looks the identical from the outside, which is why the two get confused. It keeps the template and the publishing system, but what makes all leaf distinct comes from a prompt. Swap the city name, and you have a new leaf alongside nothing new on it.

Google’s policies diagram the identical line. The term “programmatic” does not appear in its spam policies. What does appear is scaled satisfied abuse: many pages generated chiefly to manipulate rankings fairly than assistance users, no matter how it’s created. Using generative AI to create many pages without adding value is one of its named examples.

A template isn’t what gets penalized, and neither is automation. The center inquiry value evaluating is whether all leaf adds value.

Here is a quick test. Picture two pages for “cost of living in Lisbon,” built from the identical template and the identical hunt pattern. The archetypal shows rent ranges, transport passes, and market prices pulled from a maintained dataset, and it updates whenever the numbers change.

The second looks fair as complete, alongside a rent figure, a monthly budget, and a assured summary, but the numbers are the model’s finest guess. They transport no origin and no date, and the identical sentences appear on all another city leaf alongside lone the name changed. Only the archetypal is programmatic SEO. To inform them apart, ask anywhere all figure on the leaf came from and whenever it was final updated. The archetypal leaf can answer. The second can’t.

See also: How Do You Fix Thin Content Across Similar Ecommerce Product Pages? – Ask An SEO

Good pSEO Vs Bad pSEO

Most teams can’t say whether their pages are programmatic SEO or bulk publishing, since the output looks the identical from the outside. These six checks appearance underneath it.

Good pSEO vs Bad pSEO
Image from author, November 2026

None of these checks ask whether AI wrote the page. A leaf built on genuine data and drafted alongside AI can motionless execute well, during a hand-written leaf alongside no genuine data rearward it has no additional assertion to rank. Treat the difference rubric as a diagnostic: if among the rows fails, you have a workflow issue to fix. If multiple parameters fail, afterward you are bulk publishing, any you call it.

Eli Schwartz draws the row in the identical place, and argues:

“Spinning up pages fair since you can isn’t SEO; it’s programmatic spam. If you create pages since they advantage many users, that is fine SEO, whether you create them programmatically or manually.”

Danny Sullivan, afterward Google’s Search Liaison, shared a way to test this at a 2025 event. The question, in his words as Search Engine Journal reported, is whether the intent was “‘I’m going to get that traffic’ and not ‘some person really expected it'” if they came to your location directly. Ask that of your own pages: would a visitant who arrived direct at your location always district on this one on purpose?

From Static Templates To AI Pipelines: What Changed

The pages I built in 2016 and the ones I build today appearance extremely similar. Shipping a bad one is the part that got easy. I walked through that first scaling playbook at BrightonSEO in 2022.

pSEO Evolution by Nitin Manchanda
Image from author, November 2026

The center era is the one I think concerning most. A rule-based motor akin Rosae NLG writes sentences from organized data, and it lone says what the data and the template author allow. It was dilatory to build, and that slowness was doing helpful work.

Language models took that bounds away. A example volition compose a fluent, assured paragraph concerning a row alongside nothing in it, and a audience cannot tell. I covered anywhere AI belongs in a pSEO pipeline, and anywhere it does not, at ISS Barcelona in 2024.

So the checks evolved. People do the investigation and decide all page’s structure, the example drafts the sections from that structure, and a topic matter expert reviews the facts and brand fit before item goes live.

Language models did not create programmatic SEO improved or worse. We used to be stopped by the tools. Now we have to halt ourselves, and the checklist below is how I do it.

The pSEO Checklist I Use Before Any Page Ships

Before The First Page

  • Find a distinctive data source. If you cannot name it, you do not have a pSEO program.
  • Define the minimum data completeness for a row to create a page, and enforce it in code.
  • Confirm all template maps to a distinct intent, not a keyword variant.
  • Write the metadata rules – titles, descriptions, schema – alongside the identical attention as build content.
  • Decide your disclosure line: how these pages are built, how frequently the data refreshes.
  • Ask what you can add that nobody alternatively has, specified as your own measurements or two datasets blended into a new one.
pSEO Public Gate Every Row Must Pass by Nitin Manchanda
Image from author, November 2026

At Launch

  • Ship a test lot of 25-50 pages, not the complete set. Publishing 10,000 pages before you cognize the index rate is the most average nonaccomplishment in this discipline.
  • Instrument indexation tests from day one – log-based, not Search Console alone.
  • Sample-review by hand. Ten pages peruse end to end volition inform you additional than any score.

Ongoing

  • Track crawl hits per URL by template, monthly.
  • Write the slay regulation before launch: If a set portion of a leaf set earns no impressions, including in AI features, inside a set window, prune or merge it.
  • Re-review whenever the template changes – a template edit changes all leaf at once.
  • Prune on a schedule, not following an update. By the period a center update forces it, the site-level rely issue may already be in play.

In my experience, the programs that have survived and thrived throughout multiple Google updates had a written regulation for deleting pages before they needed one.

That is near to how Kevin Indig describes the nonaccomplishment manner in a straightforward communication alongside me:

“Programmatic SEO normally goes incorrect whenever it is ‘spray and pray: publishing a lot of pages at formerly alongside no experimentation, no refreshing, and no appearance at engagement.’ The fix is the one value underlining: treat pSEO akin its own product, not akin a satisfied batch.”

Search is changing underneath these programs. In the archetypal four months of 2026, 68% of U.S. Google searches ended without a click, according to SparkToro’s inspection of Similarweb data. That’s up from concerning 60% in 2024 and concerning 45% in 2016. SparkToro points to AI Overviews, which it says now appear on additional than 20% of searches, and Ahrefs estimates they cut clicks on the top outcome by 58%.

For a programmatic page, that changes the goal. Traffic can’t be the lone test of whether a leaf earns its place, since the click may never come. What stays is influence: Rand Fishkin notes that your satisfied motionless shapes AI answers and Google’s another zero-click features equal as visits fall.

That leaves three habits:

  • Build small pages, all holding data nobody alternatively has. If an AI answer needs a origin for Lisbon’s rent ranges, it can nexus the leaf that holds them. It has no logic to nexus a leaf that repeats what a dozen another pages already say.
  • Decide the query form before you build the template. If the searcher lone wants one fact, Google volition apt provision it to them. If they need to compare, calculate, book, or discover a local business, your leaf can motionless be anywhere they end up. SparkToro says branded, local, and high-intent transactional searches motionless advantage from SEO.
  • Measure by template, not fair by site. Search Console’s generative AI achievement report shows how frequently links to your pages appear in AI Overviews and AI Mode. Export the pages, collection them by template, and set all collection against its clicks in the regular achievement report. A template alongside plentifulness of AI impressions and few clicks may be earning power without visits, during one alongside neither is a applicant for your slay rule.

Dominik Schwarz puts the commerce plainly:

“pSEO doesn’t neglect since it’s programmatic; it fails whenever measure replaces substance. The difficult part isn’t the tech; it’s the site to never commerce norm for volume, since pSEO makes that commerce tempting for all decision maker.

The test hasn’t changed. If a page’s lone value was that it ranked, zero-click hunt takes distant its logic to exist. If the value was the data, an AI answer motionless has to item to it. Before you measure anything, choice one leaf and ask anywhere all figure on it came from.

More Resources:


Featured Image: RoseRodionova/Shutterstock

Other Article Search Engine Journal
↑
Close Right Ads
Close Left Ads