Why Most AI Search Tactics Are Guesswork (And What Works Instead)

A Search Engine Journal investigation in August 2026 revealed that many AI search optimisation tactics being sold to businesses rest on the same flimsy evidence you could use to justify a text file about cats. For UK plumbers, electricians and other service businesses investing in AI visibility, that matters.

Search Engine Journal published an investigation in August 2026 showing that many AI search optimisation tactics rest on evidence so weak it would justify optimising for cats. For UK service businesses spending money on AI visibility, this matters because you need to know which tactics actually work and which are expensive guesswork.

The investigation centred on llms.txt, a technical file some consultants recommend adding to websites to help AI systems understand your content. Search Engine Journal demonstrated that the four main arguments used to sell this tactic would equally justify creating a cats.txt file. Same logic, same evidence quality, different animal.

That is not a criticism of one specific file format. It is a warning about how the AI search industry operates right now. When a new AI platform launches or a new optimisation method appears, many consultants rush to sell it before anyone knows whether it works. The evidence comes later, if at all.

Why does this matter for a plumber in Manchester or a dentist in Bristol?

Because you are being sold AI visibility services. Someone is telling you that you need to optimise for ChatGPT, or Perplexity, or Google AI Overviews, or the next thing that launches next month. They are showing you screenshots, sharing case studies, and charging monthly fees.

Some of those tactics work. Many do not. The problem is telling the difference before you spend the money.

The Search Engine Journal piece, published on 7 August 2026, did not test whether llms.txt actually helps or harms visibility. It tested the quality of reasoning used to recommend it. The reasoning failed. The same arguments would justify optimising your website for any arbitrary thing, including cats.

That is the pattern to watch for across the AI search industry. Not whether a tactic sounds plausible, but whether anyone has tested it properly and shared the results.

What actually drives AI search visibility for UK service businesses?

Three things have evidence behind them, and all three predate the current AI search panic.

First, structured information. AI systems scrape the web looking for clear, factual answers to questions. If your Google Business Profile lists your services, hours, location and contact details accurately, AI can surface that information. If your website has a clear services page explaining what you do and where you do it, AI can quote it. This is not new. It is basic local SEO, and it works for AI search because AI search pulls from the same web as Google.

Second, reviews and reputation. AI systems cite businesses with strong review profiles because reviews provide social proof and answer the question every potential customer asks: is this business any good? A Leicester electrician with 200 five-star Google reviews and detailed customer testimonials gives AI something concrete to reference. A competitor with six reviews and no detail does not. Remy, The AIgency's reputation Agent, automates the process of asking for reviews after every job, so your profile builds consistently without you chasing customers manually.

Third, content that answers real questions. When someone asks an AI system how much a boiler replacement costs in Birmingham, or what to do about a toothache in Cardiff, the AI looks for content that answers that question clearly and names a location. A heating engineer with a page titled 'Boiler replacement cost Birmingham' that lists typical prices, explains the variables, and includes a local phone number has a chance of being cited. A competitor with a generic homepage does not.

Notice what is missing from that list. Experimental file formats. Obscure technical tweaks. Monthly subscriptions to optimise for platforms that may not exist in six months.

How do you avoid paying for astrology?

Ask for evidence. Not a screenshot. Not a case study from California. Not a consultant's opinion. Evidence.

Specifically, ask three questions before you buy any AI visibility service.

One: has this tactic been tested in a controlled way, and where can I read the results? If the answer is a blog post on the consultant's own website, that is not evidence. If the answer is an independent study published by a research firm or industry body, that is evidence.

Two: does this tactic work for UK service businesses specifically, or is it a general theory? AI search behaves differently in different markets and different industries. A tactic that works for e-commerce in the United States may do nothing for a plumber in Leeds.

Three: what happens if this tactic stops working? AI platforms change their algorithms constantly. Google AI Overviews launched, then changed how they cite sources, then changed again. If you are paying monthly for optimisation that depends on one platform's current behaviour, you are renting visibility that could vanish overnight.

The safest approach is to focus on the fundamentals that work across platforms. Accurate business information. Strong reviews. Clear content that answers local questions. Those tactics worked before AI search existed, they work now, and they will work when the next AI platform launches.

What should a UK service business do about AI search right now?

Start with your Google Business Profile. Make sure every field is filled in, every service is listed, and your address and phone number are correct. That information feeds into Google AI Overviews and every other AI system that scrapes Google's data.

Then focus on reviews. Set up a system that asks every customer for a review after the job is done. Remy automates this by sending review requests at the right moment, so you build a steady stream of fresh testimonials without manual follow-up. AI systems trust businesses with recent, detailed reviews because those reviews prove you are active and reliable.

Finally, publish content that answers the questions your customers ask. Not blog posts about industry trends. Not news about your company. Practical answers to practical questions, with your location in the title and the content. How much does a new bathroom cost in your town? What should someone do if their boiler breaks down at night? How long does a cavity filling take? Rosie, The AIgency's rankings and visibility Agent, publishes this kind of content automatically, targeting the local search terms and questions that drive enquiries in your area.

Those three actions—accurate business data, strong reviews, and question-focused content—give you AI visibility that does not depend on guessing which experimental tactic might work this month.

The Search Engine Journal investigation is a reminder that the AI search industry is young, chaotic, and full of people selling tactics they have not tested. Some of those tactics will turn out to work. Many will not. The difference between the two is evidence, and right now, evidence is in short supply.

For a UK service business, that means sticking to what is proven until the experiments are finished. Let the consultants test llms.txt and cats.txt and whatever comes next. You focus on the basics that have worked for twenty years and still work today.

Rosie handles the technical side of getting your business found in AI search, local search, and voice search by publishing optimised content and managing your online presence across directories and platforms. Get your free AI visibility report at The AIgency free report.

The AIgency home · Q&A · Blog