One team published 748 programmatic SEO pages. Over 90 days, those pages generated 13 clicks. More than 57% never received a single impression. Not one.
That's not a case against programmatic SEO. It's what happens when you scale a page pattern before proving it works on a small batch. The template looked fine technically. The URLs got published. And then almost nothing happened, except the site now carried hundreds of pages competing for crawl budget and linking equity with nothing to show for it.
The power-law problem nobody budgets for
Programmatic SEO performance follows a power-law distribution, and the curve is steeper than most teams expect. A multi-site analysis across four large programmatic sites found that 97.5% of pages earned 100 organic visits per month or fewer. Over a third earned essentially zero. In a separate experiment with 162 pages, just 11 drove roughly three quarters of all organic clicks.
When that smaller team deleted most of the underperforming pages, total organic traffic increased by 29%. The dead weight wasn't neutral. It was actively hurting the pages that worked. Build 5,000 pages around an unvalidated template and you aren't just wasting effort on the failures. You're dragging down the ones that could have succeeded.
What "proving the pattern" actually looks like
Standard guidance says start with 50 to 100 pages. Reasonable, but the number matters less than what those pages represent. A pilot of 50 easy, high-volume variations confirms the template works in ideal conditions. It won't tell you whether it survives thinner data, weaker search demand, or edge-case combinations.
Pick pages that span the range you'd eventually scale. Building integration pages for a B2B SaaS product? Don't just test Slack and Salesforce. Include mid-tier tools with fewer workflows and less documentation. Doing location pages? Mix large metro areas with smaller markets where your local data is sparse. The pilot's job is to find where the template breaks, not to confirm it works under the best conditions.
Before publishing, record the primary query, the related query family, the page on your site it could cannibalize, and the action you want visitors to take. That baseline makes the pilot readable after 60 to 90 days.
Four gates, in order
Gate 1: Indexation. Are pages getting discovered and staying in the index? Compare indexed pages against dropouts. Look for patterns in data depth, content completeness, and internal link structure. If pages with thin underlying data consistently fall out, that's your first constraint on what's worth scaling.
Gate 2: Query matching. Pull Search Console data and compare the queries each URL actually receives against the ones you designed it for. If your "CRM for accountants" page mostly gets impressions for "CRM software" and competes with your main product page, the intent differentiation hasn't landed.
Gate 3: Characteristic segmentation. Aggregate performance hides the real story. Segment pilot pages by data depth, inventory size, internal link count, content completeness, SERP competition. One team testing 60 integration pages found that pages with detailed setup instructions and at least three supported workflows gained impressions consistently, while integrations with limited functionality stayed invisible. That finding reshaped their entire rollout order.
Gate 4: Business behavior. Traffic alone doesn't validate the pattern. Check whether visitors do something useful: product exploration, demo requests, trial starts, or progression deeper into the site. Also look at whether these pages appear in multi-touch paths that end in pipeline.
Three outcomes, and one of them is "stop"
Scale in batches. The pattern works across page types, not just easy wins. Pages index reliably, attract distinct query families, and the characteristics of strong performers are repeatable. Zapier's integration pages are the canonical example: 50,000+ pages targeting the "[App A] + [App B] integration" pattern, contributing roughly 16% of their 1.6 million monthly organic visits as of February 2023 (per Ahrefs). That scale works because each page maps to a real, distinct user intent backed by useful data.
Improve and retest. Part of the pilot works. Maybe location pages with 20+ active providers rank, but pages with a handful of listings don't index. Scale the proven segment; hold the rest until the underlying data or template improves.
Kill it. The template produces thin pages repeating the same copy with a few fields swapped. Queries overlap. Indexation is inconsistent. Stopping here prevents thousands of weak URLs from entering the site.
The measurement shift that matters
Stop measuring programmatic SEO by total pages published. Measure performance per template, per intent cluster. Indexation rate by page characteristic. Engagement by segment. Pipeline contribution by pattern type. The teams that get this right build template-level dashboards, not sitewide averages that hide a power-law distribution behind a single number.
The 748-page experiment didn't fail because programmatic SEO doesn't work. It failed because the pattern was never proven on a small batch first. Fifty pages would have surfaced the same indexation and query-matching problems at a fraction of the cost. The pilot isn't a speed bump before the real work. It is the real work.
