What would you be buying?
Programmatic SEO uses a template and a set of records to create pages. For example, real product records might supply sizes, compatibility, price conditions and availability. The template presents those facts consistently, while an assistant may help draft the explanation.
A replacement-parts shop with verified model compatibility can give each product page a useful answer: whether this part fits the customer’s machine. That depends on dependable records. Hundreds of broadly similar articles, distinguished by a town name or adjective, offer no such reason for another page.
For a small business, the alternative may be two excellent service pages, a useful product category or a clear FAQ within an existing page. Judge the bulk proposal against that achievable alternative.
When is a separate page useful?
Ask what someone can learn or do there that the existing pages do not already handle well. There should be a real difference: the relevant product specification, a distinct service, a meaningful delivery condition or a comparison that changes the choice.
Use actual customer questions, existing search reports and a few current search results to check the need. Do not assume every combination an assistant produces is something people search for, or that every search deserves another page on your site.
The source records should identify what is known and who maintains it. Keep prices, service areas, restrictions and availability connected to approved information. An empty spreadsheet cell means the fact needs checking or the page should wait; it is not a request for plausible filler.
What is the risk with town-by-town pages?
Consider this invented mobile bicycle-repair business. A proposal combines ten towns with two service types to create 20 pages. The business serves only three towns, so fourteen proposed combinations do not describe an available service. They should be rejected before drafting.
Six combinations remain. Suppose the same two services, prices and booking process apply across all three served towns: two helpful service pages with accurate coverage information may answer the customer more clearly than six copies with different place names. That outcome does not call for a programmatic system. Ask for a simple two-page quote and compare it before buying automation.
If one location has a genuine access restriction or a different available appointment pattern, assess whether a separate page helps customers resolve it. Include the actual condition and maintain it. Do not invent an office, local review or completed job to make a location page seem distinctive.
Google's policies address low-value scaled content and similar pages created to funnel searchers elsewhere. The concern is usefulness and purpose, not a rule that all templated pages or all AI writing are prohibited. Google's published policies
How do you compare the full cost?
The unrevised bicycle-repair proposal quotes £300 to generate twenty drafts and £100 to put accepted work into the website. Twenty minutes reviewing each draft adds up to six hours and forty minutes. Valued at £30 an hour, that is £200 of the owner’s capacity. These invented figures let you test the original bulk proposal, before requesting a cheaper or simpler scope.
If the two useful service pages pass the agreed factual and usefulness checks, cash production cost is £400, or £200 per accepted page. Combined cash and review capacity cost is £600, or £300 per accepted page. These are planning numbers, not supplier prices. The quote should first be rescoped to the two useful pages so the owner need not buy and review the rejected combinations.
Continue costing that unrevised bulk proposal over six months. Assume maintenance costs £20 cash and one owner hour a month. At the same £30 hourly value, that is £50 of monthly resources. Over six months the cash commitment is £400 plus £120: £520. The combined commitment is £600 plus £300: £900. Add any omitted licence, data or setup costs before deciding; the example assumes none beyond those stated.
Suppose a genuinely additional retained repair job contributes £100 after its direct costs and before this website programme. Six such jobs would exceed the £520 cash commitment, but nine would only recover the £900 combined commitment. Ten would exceed it. Those are requirements to test, not a forecast that two pages will create that many jobs. A customer switching from an existing page to a new one is not automatically an additional job.
The pages might also help existing visitors understand the offer or reduce repetitive enquiries. Judge that service value separately if it is the reason for the work; do not relabel it as new SEO customers. If the expected benefit cannot justify the commitment, improve the smaller existing set directly or defer the automation.
Can AI do the writing safely enough?
Give it the approved fields and explicit instructions to flag missing information. Read the complete output against those records, including the page title and description. Keep qualifications beside the claims they limit. The same checks apply to product details or information placed in structured data.
Google's AI-content guidance includes manual review for accuracy and trustworthiness. An automated workflow still needs an accountable person who can recognise a wrong claim. Google's guidance
Review the page in the real website layout and on a phone. Can a prospective customer understand the difference between options and complete the intended enquiry or order? Check the booking route and repair any broken step, even when the paragraphs read fluently.
What should a small pilot include?
Choose a few genuinely different pages. Include an awkward case, such as an unavailable option or a condition that changes the price. Record what each page should help the visitor decide, the source of its facts and what counts as acceptable.
Before publishing, test what happens when a field changes or becomes unknown. A price update must reach the relevant page; an unknown delivery promise should not remain confidently stated. Confirm who owns corrections and what the process costs.
Make the approved pages reachable through normal links in the site, and ask the provider to check their intended indexing status and main page addresses. Sitemaps can help search services discover the URLs; submission does not guarantee indexing.
After an approved launch, observe whether the pages are discovered, receive relevant visits and support suitable enquiries or retained orders. Keep a consistent date window and known internal tests separate. More indexed pages do not automatically mean more customers, and a quiet first week is not enough to diagnose failure.
When should you scale?
Expand when the small set helps customers, the underlying information stays reliable and the commercial evidence can justify the next amount of work. Keep a time and cash limit. Use the weekly measurement approach to connect page activity with the business.
For many small firms, a handful of useful pages remains the sensible first investment. Consolidate proposed pages whose only difference is a keyword. Where the business cannot maintain the facts, improve the source before automating publication.