A provider proposes a hundred new SEO pages to a business. The AI writes texts for locations, services and variants within a few days. The calculation looks attractive because the unit price falls. But the number of published URLs says nothing about whether a potential customer finds a reliable answer on a particular page. This is exactly the gap the offer has to close.
What the number 100 in an SEO offer leaves out
Take a fictitious maintenance company as an example. One page is to be created for each neighbouring town. All pages list the same services, the same phone number and almost the same process; only the town name changes. This can be sold as local presence, even though the business provides no different information on response times, travel, available teams or special conditions. The reader still has to call to clarify the actual question.
Another page, by contrast, can have a real job: it explains which systems are serviced, which documents are needed before the appointment and which cases the business does not take on. This information can also be structured with AI. What matters is whether it comes from the real offer and was checked before publication. The technology used to create a draft is not a sufficient quality judgement.
If two planned pages answer the same question with the same facts, they probably do not need two publications.
What Google assesses in SEO texts written with AI
Google does not prohibit AI as a writing tool. In its guidance on generative content, Google even mentions it as an aid for research and structure. At the same time, Google expects manual fact-checking and an assessment of accuracy, quality and relevance before publication. This also applies to page titles, descriptions and image alt texts.
The spam policy on “scaled content abuse” does not start at a magic page count. It describes many pages generated primarily for search rankings rather than for users, often with little originality or added value. The creation method is irrelevant. A hundred AI pages are therefore not automatically a violation; a hundred interchangeable pages created to cover search queries are a serious warning sign. Conversely, a human-written but empty text does not make things better.
A provider who promises across the board that Google cannot detect AI texts is answering the wrong question. What matters to the client is whether content and structure make a real difference to the reader and whether every verifiable sentence fits the business. If a product that is not offered is mentioned or a service area is invented, the page is useless before any ranking problem arises.
Approve one page before a hundred go live
First have three deliberately different sample pages created: a specific service, an explanatory question and — if locally really necessary — a local page with verifiable location information. Check each page with someone who knows the work in the business. Which claims are wrong or incomplete? Which customer problem is answered? What can the reader decide better after reading it?
Then compare the pages with each other. If the same paragraph works under three titles, the plan is split too finely. A shared guide or a stronger service page can be more useful than many thin variants. If a separate page is to remain, it needs its own core of information: for example a different task, a different prerequisite or a documented regional particularity. Mere synonyms are not enough.
Document where the facts come from directly in the editorial draft. For a service page, this can be an approved product list, an internal process description or information from the responsible specialist. This way, the reviewer can later see which sentences come from the business and which were merely worded plausibly. Without this mapping, proofreading a hundred pages becomes a search for errors that nobody can record systematically.
- Evidence: Do service, area, prices and deadlines match the business?
- Distinctiveness: Which question does only this page answer?
- Next step: Does the reader understand whether and how they can get in touch?
- Maintenance: Who updates the page when the offer changes?
A limited pilot instead of a content gamble
Only after this approval is a larger plan worthwhile. Publish a small, justified selection, link it sensibly within the existing offer and watch whether the right pages become visible for suitable search queries. Search Console can show queries and clicks; whether they turn into qualified contacts has to be checked separately by the business. A rising number of URLs is not a business result.
In the contract, agree not only on a number of texts but also on research sources, expert approval, correction rounds and criteria for dropping planned pages. This also protects the provider from pointless production. If after ten pages it becomes clear that the remaining ninety would repeat the same answer, stopping is a good editorial result.
For the narrower question of when individual location pages actually carry their own information, see our article on city pages for service businesses. This one is about the decision before buying a quantity of content: whether someone can deliver a hundred texts is easy to answer. Whether a hundred pages are needed requires work on the offer and on customers’ questions.
