Most wasted AI search budget goes on tools and files. The spend that pays back goes on content, PR and technical access that AI can trust.
Most wasted AI search budget goes on tools, files and content volume, while the spend that pays back goes on clear expert pages, earned press coverage and making sure AI crawlers can actually read your site. If you only have money for a few things, fund the work that makes your business easy to find, easy to understand and easy to vouch for.
Every board now wants an AI visibility line in the marketing plan. That’s fair: buyers ask ChatGPT, Gemini and Google’s AI Overviews for shortlists before they ever visit a website. The problem is that a whole market of tools and quick fixes has grown up around that worry, and plenty of it does very little.
This is a spend review. We’ve gone through the activities we see businesses paying for most often and sorted them into what to stop and what to fund instead.
The quick version: stop doing this, do this instead
The pattern is simple. Money spent on signals AI systems can verify pays back. Money spent on gadgets that promise a shortcut usually doesn’t.
| Stop doing this | Do this instead |
|---|---|
| Paying for llms.txt setup as a visibility fix | Fix crawl access, page speed and clean HTML so AI crawlers can read your key pages |
| Tracking hundreds of prompts in an expensive tool | Track 20 to 40 prompts your buyers actually ask, monthly |
| Publishing dozens of thin AI-written articles a month | Publish fewer pages with real expertise, data and named authors |
| Buying links or paid placements dressed up as PR | Earn coverage in publications your buyers and AI tools already trust |
| Rewriting every page into short chunks for AI | Write answer-first pages that also work for people |
llms.txt and other special files
An llms.txt file is cheap to add and harmless, but it is not a visibility lever. In its 2026 guidance on optimising for generative AI features, Google said you don’t need special files or markup for its AI features, and that the normal foundations of search apply.
So if an agency quotes a project fee for “AI files”, treat that as a warning sign. Add the file if you like, but spend the real money on what AI systems need: pages that load, are indexable, and aren’t blocked from the crawlers you want reading them. If you’re unsure whether your site is accessible to those crawlers, that’s a technical check, not a content project.
Prompt tracking tools
Prompt tracking is useful in small doses and wasteful at scale. Answers from AI assistants change from one run to the next, so tracking hundreds of prompts daily mostly measures noise.
A tighter approach works better:
- List the 20 to 40 questions a buyer asks before they contact a firm like yours.
- Check them once a month across ChatGPT, Gemini, Perplexity and Google’s AI Overviews.
- Record whether you’re named, who is named instead, and which sources are cited.
- Use the cited sources to decide where to earn coverage next.
We explain the method in our guide to tracking AI Overviews. The cited sources column is the valuable part, because it tells you where your budget should go.
Content volume
Publishing more pages is the most common way AI search budgets get wasted. Large batches of generic articles rarely get cited, because AI tools favour sources with first-hand experience, clear authorship and something the rest of the web doesn’t already say.
Spend instead on fewer, better pages: service pages that explain exactly what you do and for whom, case studies with real numbers, and comparison pages that answer the questions buyers ask late in the process. For B2B firms in particular, a handful of pages written for the people who sign off the budget will do more than a hundred blog posts. That’s the core of how we approach B2B SEO.
PR and brand mentions
Earned coverage is one of the best places to put AI search budget. AI assistants build their view of a business from what other trusted sites say about it, which is why they so often cite review sites instead of your own website.
What pays back is coverage in trade titles, industry associations and well-known review platforms where your buyers already look. What doesn’t is paying for placements on low-quality sites that exist to sell links. If you’re new to this, our beginner’s guide to digital PR covers the basics.
Technical access
Technical access is the cheapest win and the most often skipped. If AI crawlers are blocked by your robots.txt, your firewall or a slow server, nothing else on this list will help.
Check three things first: that your robots.txt allows the AI crawlers you want, that your key pages are indexable and return quickly, and that your main content is in the HTML rather than loaded later by scripts. These are one-off fixes, and they make every other pound you spend work harder.
How to split a modest budget
For most UK firms we’d put the bulk of an AI search budget into expert content and earned coverage, with a smaller share for technical fixes and a light monthly tracking routine. The exact split depends on where you’re weakest, which is why we start with a baseline.
Our AI SEO agency team can show you where your budget is leaking. Start with the free AI visibility check, or book a free strategy session and we’ll talk through your plan.
Related reading: our guides to how to rank in Google AI Overviews and entity-based SEO.
Common questions
What should an AI search budget be spent on first?
Start with technical access, so AI crawlers can read your key pages, then fund expert content and earned coverage. Tracking tools come last and should stay small.
Does llms.txt improve AI visibility?
There is no good evidence that it does. Google says its AI features don’t need special files, so treat llms.txt as optional housekeeping rather than a reason to spend.
How many prompts should a business track for AI visibility?
Twenty to forty prompts that match real buyer questions is enough for most firms. Check them monthly and focus on which sources get cited.
Is publishing more content the fastest way to appear in AI answers?
No. AI tools tend to cite pages with clear expertise and original information, so a few strong pages usually beat a large batch of generic articles.
