The two options
Templated pages generated from a structured dataset, one per row.
- You hold a dataset with real per-row variation
- The page count runs to hundreds or thousands
- Queries follow a predictable pattern
Individually researched and written pages targeting specific, usually competitive, intents.
- The term is commercially critical and contested
- The answer requires judgement rather than data
- The page must persuade as well as inform
How they differ
| Programmatic pages | Hand-written pages | |
|---|---|---|
| Unit cost | Falls sharply with volume | Flat per page |
| Requires | A dataset with real variation | Subject expertise and time |
| Wins | Long-tail patterned queries | Contested head terms |
| Main risk | Thin or duplicated output at scale | Slow coverage growth |
| Typical volume | Hundreds to thousands | Tens |
What actually decides it
The threshold test is simple and rarely applied: if you removed the templated wrapper, would each page still contain something a reader could not get from any other page in the set? Where the answer is yes — different figures, different availability, different regulation — the approach is sound at almost any volume. Where it is no, volume is the problem rather than the solution.
Programmatic work fails most often at the data stage rather than the technical stage. Teams build the generation pipeline first and discover afterwards that their dataset supports fifteen genuinely distinct pages, not fifteen hundred. Auditing the data for real variation before building anything saves the entire exercise.
Contested commercial terms remain hand-written work. The pages competing for them carry original judgement, structure that took iteration to get right, and the kind of specificity a template cannot express. Attempting those at scale produces pages that read as generated, which readers detect faster than engines do.
Questions
Is programmatic SEO risky?
Only when the underlying data does not vary. Engines act against near-duplicate pages generated at scale, not against templates. A programmatic set built on genuinely distinct data is ordinary, defensible SEO.
Can AI writing replace the dataset?
No. Generated prose over a thin dataset produces pages that differ in wording but not in substance, which is the exact pattern that gets classified as spam. The data has to be real before the words are worth generating.
How much variation is enough?
Enough that the page would be factually wrong on any other row. If a find-and-replace turns page A into a valid page B, there is not enough variation to justify separate URLs.
Related services
- Programmatic SEO — SEO
- Topical Authority — SEO
- Technical SEO Audit — SEO
- Keyword Research — SEO
Still not sure?
Thirty minutes with a senior strategist. We pull your live visibility while we talk and tell you which of these two is actually binding for you — including when the answer is neither. Book a discovery call →