Generic Pages, Real Costs: Auditing AI-Assisted Content in 2026

AI-assisted drafting has moved from novelty to routine for many UK SMEs. It can help a small team keep service pages current, answer common questions and cover more ground than a solo marketer could manage. But scale brings a less obvious cost. When pages are produced quickly and lightly edited, they often repeat the same structure, the same adjectives and the same claims. That is the relevance tax: the slow erosion of trust and topical clarity that follows when a site says more but proves less.

The tax rarely appears as a sudden ranking drop. It shows up as pages that receive impressions but few clicks, enquiries that stall after a quote request, or sales conversations that begin with the customer asking a question the website should already have answered. For UK small firms, where reputation and local recognition matter, generic copy is not a neutral shortcut. It is a drag on credibility.

Where the relevance tax hides

Three patterns are worth checking first. The first is the vague service description. A page may say that a firm delivers 'tailored solutions' or 'end-to-end support' without naming the actual tasks, sectors, locations or constraints involved. A customer in Leeds or Cardiff cannot tell whether the service applies to them.

The second is repeated boilerplate. The same paragraph about being 'customer-focused' or 'passionate about quality' appears across service pages, location pages and blog posts. Boilerplate is not wrong in every context, but when it occupies space that should carry specific proof, it flattens the difference between pages.

The third is the unrecognisable claim. Phrases such as 'industry-leading' or 'trusted by businesses nationwide' may be technically true, yet no local customer would use them to describe the firm. If a claim cannot be illustrated with a named sector, a project type, a time frame or a customer comment, it is probably costing more than it returns.

Audit the pages that matter, not every URL

A full content inventory is rarely necessary. Start with the pages closest to revenue: core service pages, high-intent location pages, comparison pages and the posts that already attract impressions. Export these from Google Search Console or a crawl tool, then sort them by impressions and by the number of internal links pointing to them. Pages with visibility but weak engagement are strong candidates.

Read each page as a prospective customer would. Underline every sentence that could apply to any competitor in the same sector. If most of the page is underlined, the problem is not a missing meta description. It is a lack of specific substance. Also note where the page asks the reader to take action without giving them a reason to trust the firm.

A simple scoring sheet

Use five criteria, scored from 0 to 2. Zero means absent or generic; one means present but inconsistent; two means clear and evidenced.

1. Customer specificity: does the page name the tasks, sectors, locations or situations the firm handles?
2. Local proof: is there evidence a UK customer would recognise, such as a project example, review, certificate or named service area?
3. Unique evidence: does the page contain information that is not repeated elsewhere on the site?
4. Decision support: does it help the reader choose, compare or prepare for a quote?
5. Maintenance value: is the page still accurate and worth updating in 2026?

A page scoring 8 to 10 is usually worth rewriting rather than replacing. A page scoring 5 to 7 may be merged with a stronger related page so that evidence is consolidated. A page scoring 0 to 4 should be removed, redirected or allowed to lapse if it has no traffic, no links and no strategic purpose. The threshold is a guide, not a rule; a low-scoring page that answers a genuine customer question may deserve a rewrite instead of removal.

Rewrite, merge, remove

Rewriting should add proof, not just synonyms. Replace abstract service language with the actual sequence of work, the types of client served and the questions customers ask before booking. Add one concrete example per page where possible. A short case note, a photograph of completed work or a sentence about a local constraint is often more persuasive than another paragraph of polished generalities.

Merging is appropriate when two pages compete for the same intent. A generic blog post and a service page may both explain the same process; combining them can reduce duplication and strengthen the remaining URL. Removal is the hardest decision, but it is sometimes the most honest. Thin pages that exist only because a content calendar required them tend to dilute rather than extend topical relevance.

Make the audit routine

The most practical approach is a monthly sample. Choose ten pages, score them, and rewrite, merge or remove the weakest two or three. Record the decision in a simple log: URL, score, action, owner and date. Over a quarter, that is enough to change the character of a site without stopping normal publishing.

The relevance tax is not an argument against AI-assisted content. It is an argument against publishing without a relevance check. UK SMEs do not need more pages. They need pages that a local customer can recognise as specific, credible and useful. An audit that scores for evidence rather than word count is a sensible place to start.