Roundups
Programmatic SEO Done Well vs Badly: What Actually Separates Them
Programmatic SEO done well and done badly can use the exact same tooling, the exact same template structure, and produce the exact same page count — the dividing line is whether each page delivers genuine, distinct value or just swaps a keyword, city, or product name into otherwise identical text. Google's spam policies don't ban "programmatic SEO" by name; what they ban is "scaled content abuse," an intent-based policy that catches any mass-page-production method — automated or fully human — that exists primarily to manipulate rankings rather than help users. Zapier's 50,000+ integration pages driving roughly 5.8 million monthly organic visits sits on one side of that line; a law firm's ~42,000 near-identical city pages that lost 96% of their traffic sits on the other.
Last updated Aug 21 · 17 min read
Programmatic SEO isn't banned — scaled content abuse is
Google's scaled content abuse policy, formalized as part of the March 2026 core update, defines the violation as publishing many pages primarily to manipulate search rankings rather than help users — explicitly regardless of production method. That means the policy applies equally to AI-generated, human-written, scraped, translated, and hybrid content; automating the production process isn't itself the violation. A programmatic SEO campaign that generates 100,000 pages, each carrying genuinely distinct, useful information, isn't a scaled-content-abuse violation just because it was automated. A campaign that publishes 100,000 pages differing only in a swapped city or product name is a violation regardless of whether a person or a script wrote them.
What "done badly" actually looks like
Google's enforcement examples and independent analysis converge on three recognizable patterns:
1. Mass AI-generated content with no editorial review. Sites publishing 50-500 AI-generated articles per day across keyword clusters, with thin factual depth, no first-hand experience, and identical structure repeated across hundreds of pages.
2. Template-with-variable-substitution. The classic "Best [service] in [city]" pattern across hundreds of locations, where only the city name and a handful of variables change and the substantive content is otherwise identical.
3. Low-value aggregation. Pages that scrape or aggregate source data and add no additional context, analysis, or value beyond what's already available at the source.
A concrete documented example: a law firm site that generated roughly 42,000 near-identical city pages — same body copy, only the city name and firm name swapped — reportedly suffered a 96% traffic drop. A separate, frequently cited example is 5,000 "plumber in [city]" doorway pages that all redirected visitors to a single contact form — a doorway page, by definition, exists purely to rank for a query and funnel traffic elsewhere, which is one of the clearest-cut violations in Google's spam policies regardless of how the pages were produced.
What "done well" actually looks like
Zapier is the most consistently cited reference example for programmatic SEO at scale: over 50,000 integration pages ("Connect [App A] and [App B]"), each describing a genuinely distinct, functional integration, reportedly driving roughly 5.8 million monthly organic visits. The distinguishing factor isn't the page count — it's that each individual page answers a genuinely distinct question ("does Zapier connect to this specific pair of apps, and how") rather than repeating the same generic content with a variable swapped in.
Beyond SaaS, travel, real estate, and ecommerce platforms, along with sites like eBay, are cited as verticals where programmatic SEO has worked at scale — in each case, the pages carry data (prices, listings, availability, specifications) that's genuinely different per URL, not just a templated wrapper around identical prose.
The success factor, as one practitioner puts it: the goal isn't creating 100,000 pages — it's providing real value on those 100,000 pages, through original insight or data per page rather than volume for its own sake.
The AI-assisted publishing red flags Google watches for
Independent analysis identifies specific volume and velocity thresholds that correlate with getting flagged, even absent any single "smoking gun" of thin content:
– Sustained publishing of 10+ articles per day for months is flagged as a red flag, since human editorial teams typically produce at most 10-15 quality pieces per week — a 10x+ gap between a plausible human pace and the observed publishing rate is itself a signal.
– A "safe" AI-assisted productivity multiplier is roughly 2-4x over a human baseline; a 40-100x increase in output is treated as a warning sign of scaled abuse, independent of the content's individual quality.
– Launching all at once vs. ramping gradually. Practitioners recommend ramping up gradually — for example, roughly 1-3 posts/day scaling over time to 11-19/day — rather than dumping hundreds of pages simultaneously, since a sudden spike in programmatic content is itself a signal that can get a site sandboxed independent of the content's quality.
How much a scaled-content-abuse penalty actually costs
Reported traffic-loss ranges from real scaled-content-abuse penalties, by site type:
– Niche informational sites with 500+ AI-generated pages: roughly 60-80% traffic loss.
– Affiliate review sites: roughly 40-70% traffic loss.
– Location-based service pages (the "plumber in [city]" pattern): roughly 30-60% traffic loss.
These ranges, plus the specific 96% figure for the cited law firm case, illustrate that the downside isn't a modest ranking dip — it's frequently an order-of-magnitude collapse in organic traffic, which is the core reason "done badly" is a genuinely high-stakes mistake rather than a minor inefficiency.
Practical examples
Example 1 — A SaaS company building integration pages, done well. Following Zapier's model, a smaller SaaS tool builds one page per genuine third-party integration it supports, each describing the specific setup steps, use cases, and functional details of that particular pairing — a structure that scales cleanly because each page's content is genuinely distinct, not templated filler.
Example 2 — A local services directory, done badly. A directory site generates a page for "[service] in [city]" for every combination of 50 services and 500 cities, but the body copy is identical across all 25,000 pages except for the city/service name — this is functionally the same pattern as the cited 42,000-page law firm case and carries the same collapse risk.
Example 3 — A real estate platform, done well. A property listings platform generates a page per listing, each carrying genuinely unique data (price, square footage, photos, location specifics) — the page count is enormous, but each page answers a genuinely distinct query about a specific property, matching the pattern cited for successful real estate programmatic SEO.
*Illustrative only:* Example 1 and 3 are generic patterns following the cited Zapier/real-estate models, not confirmed case studies of a specific named company; Example 2 mirrors the documented law-firm pattern without naming an actual site.
Data and evidence
– Scaled content abuse policy applies regardless of production method (AI, human, scraped, translated, hybrid): Digital Applied.
– Google doesn't name "programmatic SEO" as banned; the operative policy is "scaled content abuse," intent-based: Growth Engineer.
– 5,000 "plumber in [city]" doorway pages redirecting to one contact form: Digital Applied.
– ~42,000-page law firm site, 96% traffic drop: Digital Applied.
– Zapier: 50,000+ integration pages, ~5.8M monthly organic visits: NoGood.
– 10+ articles/day sustained for months as a red flag; human editorial baseline ~10-15/week: Digital Applied.
– 2-4x AI-assisted productivity as sustainable; 40-100x as a scaled-abuse warning sign: Digital Applied.
– Traffic-loss ranges by site type (60-80% niche info, 40-70% affiliate, 30-60% local service): Digital Applied.
– Gradual ramp-up (1-3/day scaling to 11-19/day) recommended over simultaneous mass launch: Indie Hackers.
– 283 of ~400 pages indexed in a 2026 case study, ~2% stuck in "Crawled – Currently Not Indexed": Indie Hackers.
– Travel, real estate, ecommerce, and eBay cited as successful verticals beyond SaaS: NoGood.
– "The goal isn't 100,000 pages, it's real value on those 100,000 pages": Ian Nuttall.
Evidence not sufficiently verified: exact current per-site traffic figures for Zapier (the 5.8M/month figure) are third-party estimates rather than a figure Zapier itself has disclosed, and should be read as a directional industry estimate rather than an audited number.
Comparisons
Programmatic SEO vs. doorway pages. Programmatic SEO is a production method; a doorway page is a specific violation pattern (a page that exists purely to rank and funnel traffic elsewhere) that can occur with or without programmatic production. Conflating the two is a common mistake — most programmatic SEO isn't doorway pages, but the 5,000-page plumber example shows how easily one becomes the other.
Zapier's approach vs. thin templated sites. Both produce tens of thousands of pages from a template; the difference is entirely in whether the variable data behind the template represents genuinely distinct value (a real, functioning integration) or just a swapped keyword with identical surrounding prose.
AI-generated pages vs. programmatic SEO. These overlap but aren't synonymous — programmatic SEO is a page-generation architecture (a template plus a dataset), while AI-generated content is one possible way to fill that template's variable content. A programmatic SEO site can use AI-written content responsibly (with editorial review and genuine per-page value) or irresponsibly (mass-generated, unreviewed, thin) — the risk profile depends on execution, not on whether AI was involved at all.
Real-world use cases
Beyond SaaS integration pages, travel platforms use programmatic SEO for route- or destination-specific pages carrying genuinely distinct pricing, availability, and location data; ecommerce platforms use it for category/attribute combination pages (e.g., "[color] [product type]") where each page reflects a genuinely different, filterable product set; and real estate platforms use it for per-listing pages, where the page count scales with actual inventory rather than with keyword permutations layered onto identical content.
Common mistakes
– Confusing page volume with value. Believing that generating more pages is inherently the goal, rather than generating pages that each answer a genuinely distinct query.
– Launching the entire page set at once. A sudden spike in indexed programmatic content is itself a signal Google's systems can flag, independent of individual page quality.
– Skipping editorial review on AI-assisted content. The line between a 2-4x sustainable productivity boost and a 40-100x red-flag multiplier is largely determined by whether a human is actually reviewing and improving what gets published.
– Treating the template as the product. The template is infrastructure; the actual product, from a search-quality standpoint, is the unique data or insight that fills each instance of it — sites that lose sight of this end up with the "plumber in [city]" pattern.
– Assuming automation itself is the risk. The policy Google enforces is about intent and page-level value, not about whether a script was involved — this misunderstanding leads some teams to over-correct by avoiding automation entirely rather than fixing the actual per-page value problem.
Best practices
– Before building the template, confirm that the underlying dataset genuinely varies in a way that matters to a searcher — if two pages would read almost identically with the variable swapped out, the template needs more substance, not fewer restrictions.
– Ramp publishing volume gradually (roughly 1-3/day scaling toward 11-19/day, per practitioner-reported case studies) rather than launching the full page set simultaneously.
– Keep AI-assisted publishing volume within a roughly 2-4x multiplier of what a focused human team could plausibly produce, and maintain real editorial review at that pace.
– Monitor indexation rates after launch (the cited case study's 283-of-400 indexed, ~2% stuck result is a reasonable healthy benchmark) as an early signal of how Google's systems are receiving the page set.
– Study Zapier's approach directly: each page answers one genuinely distinct, narrow question, rather than being a generic template wrapper repeated with minor substitutions.
– Treat any page pattern resembling "doorway page" (exists purely to rank, funnels everyone to the same generic destination) as a hard stop, regardless of production method.
Key takeaways
– Google penalizes "scaled content abuse" — intent and per-page value — not programmatic SEO or automation as a production method.
– The clearest dividing line: does removing the variable (city, product, integration) leave genuinely different content behind, or near-identical boilerplate?
– Zapier's 50,000+ page, ~5.8M-monthly-visit model works because each page answers a genuinely distinct question; the cited 42,000-page law firm site lost 96% of its traffic because its pages didn't.
– A 2-4x AI-assisted productivity multiplier over a human baseline is treated as sustainable; 40-100x is a documented red flag.
– Gradual, ramped publishing (roughly 1-3/day scaling to 11-19/day) is safer than launching a full page set simultaneously, independent of individual page quality.
Relevant tools.scult.in resources
Before scaling any template-based page strategy, it's worth checking how visible your existing content already is to AI-driven search and answer engines — the AI Visibility Checker gives a starting read on that, which is a useful sanity check before investing in a large programmatic build. For teams working on the actual page templates and briefs that need to carry genuine per-page value rather than thin substitution, the page-template and content-scaling prompt patterns in this space are worth pairing with a hard look at whether the underlying dataset supports the distinct value this article describes.
Getting the strategy right before scaling — deciding what genuinely varies per page, how fast to ramp, and how to structure the dataset — is exactly the kind of planning SCULT's SEO/GEO service work is built around, worth a conversation before committing engineering time to a template that might land on the wrong side of the line described above.
For a related, free starting point, try the Schema Markup Generator.
Frequently asked questions
What is programmatic SEO?
A strategy of generating a large number of pages from a template combined with a structured dataset, so that page count scales with the size of the underlying data rather than being written one-by-one.
Does Google ban programmatic SEO?
No — Google's spam policies don't name "programmatic SEO"; the relevant policy is "scaled content abuse," which is intent-based and applies to any mass-production method, automated or not. (Growth Engineer)
What is scaled content abuse?
Publishing many pages primarily to manipulate search rankings rather than help users — a policy that applies regardless of whether the content is AI-generated, human-written, scraped, translated, or a hybrid.
What is a doorway page?
A page created purely to rank for a specific query and funnel visitors elsewhere, often to a single generic destination — a well-established, explicit spam-policy violation.
Is programmatic SEO still viable in 2026?
Yes, when each page carries genuine, distinct value — pages with unique data per URL and real user value remain fully permitted even at very large scale.
What's the difference between programmatic SEO done well and done badly?
Whether each page delivers genuinely distinct value, or is simply a template with a keyword/city/product name swapped in over otherwise identical content.
How many pages is "too many" for programmatic SEO?
There's no fixed page-count ceiling — Zapier's 50,000+ pages work because each is distinct; the risk comes from thin, templated repetition, not from raw volume.
What is Zapier's programmatic SEO example?
Over 50,000 integration pages, each describing a specific app-to-app connection, reportedly driving roughly 5.8 million monthly organic visits — the most frequently cited "done well" reference case.
What happened to the law firm that scaled programmatic pages badly?
A site with roughly 42,000 near-identical city pages (same body text, swapped city/firm name) reportedly suffered a 96% traffic drop.
What industries beyond SaaS have used programmatic SEO successfully?
Travel, real estate, and ecommerce platforms, along with sites like eBay, are cited as successful verticals.
What does "genuine per-page value" actually mean in practice?
Each page should answer a question a real searcher would have about that specific instance — a specific city, a specific integration, a specific listing — rather than restating generic information with one variable changed.
Why does production method (AI vs. human) not determine whether a site gets penalized?
Because Google's policy is explicitly about intent and outcome (does this page primarily exist to manipulate rankings, or to help a user) rather than the mechanism used to produce it.
What publishing volume is considered a red flag by itself?
Sustained publishing of 10+ articles per day for months, since that's well beyond a typical human editorial team's realistic output of 10-15 quality pieces per week.
What productivity multiplier is considered "safe" when using AI to scale content?
Roughly 2-4x over a human baseline; multipliers of 40-100x are treated as a warning sign of scaled abuse.
How much traffic can a scaled-content-abuse penalty cost a niche info site?
Roughly 60-80% for niche informational sites with 500+ AI-generated pages, per reported ranges.
How much traffic can it cost an affiliate review site?
Roughly 40-70%, per reported ranges.
How much traffic can it cost a location-based service site?
Roughly 30-60%, per reported ranges.
Should programmatic pages launch all at once or gradually?
Gradually — practitioners recommend ramping from roughly 1-3 posts/day up to 11-19/day over time, rather than launching hundreds of pages simultaneously.
Can newly launched programmatic pages get indexed reliably?
Yes, when built with genuine value — one 2026 case study reported 283 of roughly 400 pages indexed, with only about 2% stuck in "Crawled – Currently Not Indexed."
Is duplicate structure risky even without obvious keyword stuffing?
Yes — "data-template pages" that swap only a location, product, or keyword variant into an otherwise identical structure are explicitly flagged as a scaled-abuse pattern, independent of keyword density.
How do I build programmatic SEO pages that rank and stay ranked?
Start from a dataset that genuinely varies in ways searchers care about, build the template around presenting that variation clearly, ramp publishing volume gradually, and maintain editorial review throughout.
How do I scale content without triggering a scaled-content-abuse penalty?
Keep any AI-assisted productivity multiplier in the 2-4x range over a realistic human baseline, avoid launching large batches simultaneously, and ensure every page's core value proposition would survive having its variable removed and checked against a near-identical page.
How do I audit an existing programmatic SEO site for scaled-content-abuse risk?
Sample a set of pages and check whether removing the variable (city, product, integration name) leaves genuinely different substantive content behind, or nearly identical boilerplate — the latter is the specific pattern Google's enforcement examples target.
How do I decide what data should power a programmatic SEO template?
Choose data where each entry genuinely changes what a searcher would want to know — pricing, availability, specifications, functional details — rather than data that only changes a proper noun in otherwise static text.
How do I know if my publishing pace is safe?
Compare it against what a focused human editorial team could plausibly produce (roughly 10-15 quality pieces/week) — sustained output well beyond a 2-4x multiplier of that baseline is a documented red flag.
Advanced: how does Google's system distinguish scaled abuse from a legitimately large, well-built site?
Based on the enforcement patterns described here, the assessment centers on per-page value and publishing-intent signals (structure duplication, sudden volume spikes, near-identical content across pages) rather than page count alone.
Advanced: does internal linking structure affect scaled-content-abuse risk?
Evidence not sufficiently verified in the sources reviewed here — this guide can't confirm a specific mechanical relationship between internal linking patterns and scaled-content-abuse detection beyond the general principle that doorway-page-style funneling (all pages leading to one generic destination) is explicitly flagged.
Advanced: can a site recover after a scaled-content-abuse penalty?
Recovery specifics aren't detailed in the sources reviewed here beyond the general principle that removing or substantially improving the offending thin/templated pages is the standard remediation path for any Google spam-policy violation; specific recovery timelines are evidence not sufficiently verified.
Advanced: does mixing genuinely valuable pages with thin templated pages on the same site create risk for the good pages too?
The policy is described as applying to patterns of abuse across a site, which suggests a site with a large thin-content contingent could face broader scrutiny even where some pages are genuinely valuable — though this guide can't confirm the exact mechanics of how broadly a penalty's impact spreads across a mixed site.
Advanced: is there a reliable early-warning indicator that a programmatic rollout is at risk?
Indexation rate is one practical signal — the cited case study treated an ~2% "Crawled – Currently Not Indexed" rate as a positive result; a much higher stuck-rate could suggest Google's systems are already treating the page set with more skepticism.
Programmatic SEO vs. doorway pages — how do you tell them apart?
Programmatic SEO is a production method; doorway pages are a specific violation defined by existing purely to rank and funnel elsewhere — a programmatic site avoids being a doorway-page site by ensuring each page has a genuine, standalone purpose beyond redirecting traffic.
Zapier's approach vs. thin templated sites — what's the concrete difference in output?
Zapier's pages each describe a real, functioning, distinct integration; thin templated sites repeat the same body copy with only a proper noun swapped — same architecture, different substance.
AI-generated pages vs. programmatic SEO — are they the same thing?
No — programmatic SEO is a page-generation architecture; AI-generated content is one possible way (among others, including human writing) to fill that architecture's variable content, with the risk profile depending on execution rather than the mere presence of AI.
Gradual ramp-up vs. simultaneous mass launch — which is safer?
Gradual ramp-up is the practitioner-recommended approach, since a sudden volume spike in programmatic content is itself a signal that can draw scrutiny independent of individual page quality.
Human-written vs. AI-assisted programmatic content — does Google treat them differently?
Not by production method itself — the policy applies equally regardless of whether content is AI-generated, human-written, scraped, translated, or hybrid; what matters is per-page value and publishing intent.
My programmatic pages aren't getting indexed — what's usually wrong?
Check whether pages are too thin or too similar to each other relative to the variable they represent — under-indexation is one of the earlier, less severe signals that content may be read as low-value before a full ranking penalty follows.
My traffic dropped sharply right after launching a batch of programmatic pages — what happened?
This matches the documented pattern for scaled-content-abuse penalties on template-with-substitution content; audit a sample of pages for genuine per-page value versus repeated boilerplate as a first diagnostic step.
My programmatic pages were fine for months and then traffic collapsed — why?
Google's core updates (like the March 2026 update that explicitly named scaled content abuse) can retroactively re-evaluate previously-tolerated patterns — a sudden collapse well after launch is consistent with a policy/algorithm update catching up to an existing pattern rather than a new mistake.
I used AI to help write my programmatic pages and got penalized — was AI the problem?
Not necessarily AI itself — the policy targets thin, repetitive, unreviewed content regardless of how it was produced; if your AI-assisted output followed the mass-production pattern (high volume, no editorial review, near-identical structure) rather than the sustainable 2-4x-multiplier pattern, that's the more likely specific cause.
My programmatic site's traffic loss doesn't match the typical ranges cited for my site type — what else could explain it?
Other core-update factors unrelated to scaled content abuse can also affect rankings; a traffic drop should be diagnosed against multiple possible causes, not assumed to be scaled-content-abuse related just because the site uses programmatic pages.
Is it worth building a programmatic SEO strategy for a small site, or is this only for large-scale players?
It's viable at smaller scale too, as long as the underlying dataset genuinely supports distinct per-page value — the risk/reward calculation is about execution quality, not about needing Zapier-scale resources to do it safely.
Is it worth hiring an agency that specializes in programmatic SEO, or can this be built in-house?
Either can work; the deciding factor is usually whether the in-house team has the specific discipline to prioritize per-page value and gradual rollout over sheer page-count output, since that discipline — not the tooling — is what separates the "done well" and "done badly" outcomes described here.
Is a pSEO software tool enough on its own to do this safely?
Tooling handles the mechanical template-and-dataset production; it doesn't substitute for the editorial judgment about whether the underlying data genuinely supports distinct per-page value, which is the part most directly tied to penalty risk.
Should I prioritize AI-assisted or fully human-written content for a programmatic rollout?
Either can work under the scaled-content-abuse policy, since it applies regardless of production method — the decision should be based on which approach lets you sustain genuine per-page quality at your planned volume, not on avoiding AI for its own sake.
What's the single most important decision before starting a programmatic SEO project?
Confirming the underlying dataset genuinely varies in ways that matter to a searcher — everything else (ramp pace, review process, template design) is secondary to getting this right.
Should I ramp my programmatic rollout faster if early pages are ranking well?
Early ranking success doesn't remove the risk associated with rapid volume spikes; a cautious, gradual ramp remains the recommended approach even when initial results look promising.
Is it worth investing in a competitive gap audit before building a programmatic SEO template, to check what similar sites are doing well or badly?
Yes — studying both a working example (like Zapier's model) and a penalized example (like the cited law-firm pattern) before building your own template is a low-cost way to calibrate what "genuine per-page value" needs to look like in your specific niche.
How do I know if my niche is even a good fit for programmatic SEO?
Check whether there's a structured dataset in your space where each entry (a location, a product pairing, a listing) would independently be something a real searcher wants distinct information about — niches without that underlying structure aren't naturally good fits.
Should a growth team measure success by page count or by something else?
By something else — indexation rate, per-page organic traffic, and conversion from those pages are better health indicators than raw page count, which the sources here consistently warn against treating as the goal itself.
What's the honest, one-sentence takeaway on whether programmatic SEO is "safe" in 2026?
It's safe when it's built around genuine per-page value and rolled out with editorial discipline and a gradual ramp; it's a significant risk — documented at 30-96% traffic loss in real cases — when it's built around templated substitution with no real per-page substance.
Sources
- https://www.digitalapplied.com/blog/scaled-content-abuse-google-march-update-ai-pages-decimated
- https://growthengineer.ai/blog/programmatic-seo-google-penalty
- https://nogood.io/blog/programmatic-seo/
- https://www.indiehackers.com/post/looking-for-programmatic-seo-case-studies-and-success-stories-9439e1f9e7
- https://iannuttall.gumroad.com/l/pseo-examples
- https://www.digitalapplied.com/blog/programmatic-seo-after-march-2026-surviving-scaled-content-ban
- https://bulkbase.ai/seo/scaled-content-abuse-googles-policy-enforcement-how-to-stay-compliant-in-2026
Need this built into your business?
The free tools and prompts on this site handle the small, solved problems. If what you need is bigger — SEO, built and maintained for you — that's Scult's day job.
