Programmatic SEO is the practice of generating large numbers of pages from structured data and templates rather than writing each page by hand. The examples that work share a common pattern: real, differentiated data fills each template, so every page answers a specific query no other page on the site already covers. The approach scales to hundreds of thousands of pages when the underlying data is rich enough to justify it.
The classic failure mode is mistaking volume for value. Google’s spam and core updates in 2021 and 2023 hit ZoomInfo and G2 hard because their programmatic pages were thin by design. The examples below are the ones that held up, and the reason each one held up is the same: proprietary data, real utility, and enough variation between pages to avoid cannibalization.
This page covers what programmatic SEO actually looks like in practice across multiple industries, what makes each implementation work (or fail), and how AI citation changes the calculus for pages built this way.
Wise: Currency Conversion at Scale
Wise built a programmatic SEO program around currency conversion that generates over 100 million monthly organic visits, according to Backlinko’s analysis of Ahrefs data. The /currency-converter/ subfolder alone covers thousands of currency pairs, each with a real-time mid-market rate, a 30-day performance chart showing highs, lows, and averages, conversion tables for common amounts, and a fee comparison against competitors like bank wire transfers. The pages answer the query (“USD to EUR rate”) and then convert users into customers by surfacing Wise’s actual product.
Three data layers make this work. First, Wise has proprietary rate data because it operates as a financial institution, not a data aggregator. Second, each page has interactive functionality (a live calculator, rate alerts) that static content farms cannot replicate. Third, the pages link laterally to related currency pairs, creating a dense internal graph that distributes authority across the cluster. The result is a programmatic program that survives algorithm updates because each page has genuine utility, not just keyword coverage.
The AI-citation angle matters here too. When someone asks ChatGPT or Perplexity “what’s the USD to EUR rate,” Wise’s pages are structured clearly enough (current rate, source attribution, historical context) that they’re regularly cited. The format satisfies what AI engines need: a direct answer followed by context, all on one page.
Zapier: App Integration Pages
Zapier operates more than 590,000 integration pages under its /apps/ subfolder, generating around 610,000 monthly organic visits from that section alone. The template is simple: one page per app (e.g., “Google Sheets integrations”), and one page per app-to-app combination (e.g., “Google Sheets and Notion integration”). Each page lists the supported triggers and actions for that combination, plus pre-built workflow templates users can activate in one click.
The programmatic layer is straightforward: Zapier’s integration database drives the template. Every time a new app integrates with Zapier, hundreds of new combination pages auto-publish. But the reason these pages rank is not the volume. It’s that each page contains the actual product: working workflows, not descriptions of workflows. A user searching “Google Sheets Slack integration” lands on a page where they can start that integration immediately.
This is the distinction worth building into any programmatic program. Pages that describe a thing rank less reliably than pages that provide the thing. Zapier’s app pages are the product expressed as SEO content.
Nomad List: Location Data Pages
Nomad List built a catalog of city pages targeting the digital nomad audience, each bundling a consistent set of data points: cost of living, internet speed, weather, safety score, air quality index, and community reviews from actual digital nomads. The keyword targets follow a predictable pattern across the long tail: “cost of living in Chiang Mai,” “best cities for digital nomads with fast internet,” “is Lisbon safe for remote workers.”
What makes these pages work is the combination of data sources. Internet speeds come from real speed tests contributed by users. Cost-of-living figures are community-sourced and verified over time. The pages are not scraped copies of existing data; they are aggregations of data that did not exist in combined form before Nomad List assembled it. That’s the threshold worth understanding: programmatic SEO works when the template produces a genuinely new information object, not a recombination of publicly available data that any scraper could replicate.
TripAdvisor and Zillow: Authority Compounding
TripAdvisor dominates “things to do in [city]” and “restaurants in [city]” searches by generating millions of location-based pages. The template layers multiple attributes: city-specific pages, cuisine-and-city combinations, neighborhood-level pages, and filtered views by price range, occasion, or dietary restriction. The result is a page inventory that covers every reasonable combination of location and restaurant attribute, including “vegan restaurants in Downtown LA.”
Zillow’s programmatic footprint is larger. Backlinko’s analysis found 173,000 home value pages by location, 1.6 million “for sale by agent” pages, 1.2 million rental pages, and 1.5 million “recently sold” pages, totaling 243 million monthly organic visits. The data driving these pages is real property records, real listing prices, and real transaction histories. Each page is unique not because Zillow wrote it differently, but because the underlying data is unique to that property or zip code.
Both examples illustrate a pattern: the highest-traffic programmatic programs sit on top of data that is inherently unique per record. A property listing, a restaurant with reviews, a currency pair with live rates. These pages survive because their differentiation is structural, not editorial.
Webflow: User-Generated Templates as SEO Pages
Webflow built a large catalog of template pages by surfacing the website designs its users had created as individual landing pages. Each template page describes the design, links to a preview, and targets keywords like “portfolio website template” or “SaaS landing page template,” capturing long-tail search demand that grows organically as more designers contribute.
The data source is user-generated content rather than first-party product data. This pattern works for platforms where users produce output that others search for: design templates, code snippets, app integrations, project examples. The SEO value comes from the catalog, and the catalog grows automatically as more users contribute. Programmatic generation simply makes the catalog indexable.
What the Failures Look Like
ZoomInfo and G2 both experienced significant traffic drops following Google’s May-August 2021 spam updates and the October 2023 core update. The common characteristic: pages built primarily to rank for a data point (a company’s phone number, a software category listing) rather than to deliver a meaningful experience. The pages existed because the data existed, not because anyone would find them genuinely useful.
Google’s guidance on “helpful content” is a useful diagnostic. Ask whether a person who reads the page would feel they got what they came for. Thin programmatic pages fail this test because the template answer and the user’s actual intent don’t align closely enough. The page might rank for “ZoomInfo revenue” but deliver a stub with a gated data point and no context. That’s the pattern that collapsed.
Programmatic SEO and AI Citation
AI engines change the incentive structure for programmatic content. When ChatGPT or Perplexity answer a query like “best currency converter” or “Zapier vs Make integrations,” they pull from pages that have a clear, citable answer format: a direct statement, supporting data, and a named source. Programmatic pages built around structured data are well-suited to this format, but only if the template includes a direct-answer paragraph near the top of each page.
The sites that get cited in AI answers from their programmatic pages are the ones that treat each page’s opening paragraph as a standalone answer, not a table of contents. Wise’s currency pages, for example, open with the current rate stated plainly before any navigation or promotional content. That opening sentence is what AI engines extract.
AI citation tracking lets you monitor which of your programmatic pages are appearing in ChatGPT, Perplexity, or Gemini answers, and which are invisible despite ranking on Google. The patterns are often different, because AI engines weigh structure and answer clarity more heavily than traditional SERP ranking signals.
Applying These Examples to Your Own Program
The examples above cluster around a few repeatable patterns.
Data-driven utility pages follow the Wise and Zapier model: the page does something (converts currency, starts a workflow) rather than just describing it. These pages have the highest retention and the strongest AI-citation rate because the format is self-evidently useful.
Location or entity combination pages follow the TripAdvisor and Nomad List model: one data template, applied across every instance of a category. Works when the underlying entity (a city, a restaurant, a property) has enough unique data to fill the template with real content.
User-generated catalog pages follow the Webflow model: users produce the underlying content, the platform generates the SEO layer. Works when your platform has a large user base creating differentiated output.
The decision table for whether programmatic SEO is viable:
| Factor | Required? |
|---|---|
| Unique data per page instance | Yes |
| Template answers a real, searchable query | Yes |
| At least one interactive or functional element | Strongly recommended |
| Domain authority to support indexing at scale | Yes for large programs |
| Each page solves a distinct user intent | Yes |
Fokal’s daily-article workflow is one form of automated content creation that sits adjacent to programmatic SEO: instead of pure templates, it combines research and AI generation to produce differentiated content at scale. The distinction is worth noting because search engines treat templated thin pages and researched long-form content differently, even when both are “automated.”
For a full breakdown of how programmatic SEO fits into the broader automation toolkit, see the SEO automation hub and the companion guide to programmatic SEO tools. If you’re planning a program, the programmatic SEO guide covers the template and data architecture decisions before you build.
Track whether your programmatic pages are earning AI citations alongside Google rankings. The two signals increasingly diverge, and knowing which pages are invisible in AI answers tells you where the template is failing the answer-clarity test.