The problem with programmatic SEO is not making the pages
Anyone can merge a spreadsheet into a template and publish four hundred URLs this afternoon. The problem is that four hundred pages differing only by a city name is the textbook definition of what Google calls scaled content abuse.
Why most programmatic sets fail
A template plus a list of places produces pages that share ninety-five per cent of their text. They compete with each other, they give a reader nothing their sibling did not, and in aggregate they tell Google the site publishes filler.
The usual fix is to spin the wording so the pages differ on the surface. That changes the sentences without changing the fact that no page says anything specific, which is the thing being judged.
What makes a set of pages genuinely different
Each row is researched and written on its own, and the row's own data is handed to the writer rather than left to be inferred from the keyword. Then the result is checked before it is kept.
- Per-row research
- Every page gets its own live search against its own query. A page written with no research is the model paraphrasing itself, which is exactly the uniform filler to avoid.
- Your data reaches the writer
- The columns in your table, the local rate, the count, whatever you have, are given to the draft as facts specific to that page. This is the mechanism that makes the pages differ in substance rather than in wording.
- Near-duplicates are refused
- Each finished page is compared against what you already have. One that comes out too close is not saved, and the similarity is recorded so you can see the decision.
- A bad template stops the run
- After three refusals in a row the whole batch halts. Losing three generations to a template that does not vary enough is a lesson; losing four hundred is a bill.
- Nothing publishes with a hole in it
- Rows missing a value are rejected at setup, and a draft that still contains an unfilled placeholder is held back rather than posted.
Start with four rows and read what comes out.
Start freeWhen not to do this at all
If your rows carry nothing but a name, there is no honest version of this. A page about tutors in one city and a page about tutors in another need something true of each to be worth separating, and if you do not have it, the set should not exist.
Pages are capped per run and generated one at a time, deliberately. A tool that produces four hundred pages in an hour is optimising for the thing that gets sites penalised.
Near-duplicates are refused, not published.
Start freeQuestions worth asking first
How many pages can I generate at once?
Twenty-five per run, and each one counts against your daily article allowance the same as any other article. The cap is a decision rather than a technical limit.
What data do I need to start?
A table with a column for whatever varies, pasted from a spreadsheet. At minimum the thing that changes per page; ideally one or two facts that genuinely differ per row, because those are what make the pages distinct.
What happens to rows that fail?
They are kept and marked with the reason: a missing value, a slug that collides with another row, a page refused as a duplicate. You can read the list, fix the data and run again.