16 July 2026

There is a category of conversation happening about your UK business right now that you almost certainly cannot see. It is happening inside ChatGPT. It is happening inside Perplexity. It is happening inside Gemini. Someone asks one of these AI systems which digital marketing agency they should hire in London, or which UK accounting software handles Making Tax Digital correctly, or which CRM integrates best with Xero — and the AI responds with a recommendation that either includes your business or does not. You receive no notification. No Search Console impression. No GA4 session. No Ahrefs ranking change. The conversation happens, the recommendation is made, and the prospect either considers your business or does not — based entirely on what an AI system believes about you, sourced from content and signals you may or may not have deliberately cultivated. This is the AI brand mention problem that most UK businesses have not yet structured a response to. Not because the problem is new — it has been developing since ChatGPT’s public launch in late 2022 — but because the monitoring infrastructure that makes brand mentions in traditional search visible (Google Search Console, brand tracking tools, referral analytics) simply does not exist in an equivalent form for AI search systems. There is no native AI Brand Console. There is no official API exposing how frequently an AI recommends your business. There is no referral tag identifying that a customer found you through a ChatGPT recommendation. The data is partially dark, partially manual, and partially inferrable from downstream signals. But “partially” is not “entirely” — and UK businesses that build the monitoring infrastructure available to them today will accumulate a meaningful competitive intelligence advantage over those who wait for the problem to become fully measurable before addressing it. This guide covers every available method for tracking AI brand mentions across ChatGPT, Perplexity, and Gemini — from fully automated tools through semi-automated workflows to manual testing frameworks — and explains how to connect what you discover to strategic action that improves your AI visibility over time. Why AI Brand Mention Tracking Is Fundamentally Different From Traditional Monitoring Before the methodology, the foundational difference between tracking traditional brand mentions and tracking AI brand mentions must be understood — because applying traditional monitoring frameworks to AI systems produces systematically incomplete data. Traditional brand mention monitoring works through crawling and indexing: a tool like Brand24, Mention, or Semrush’s Brand Monitoring crawls the web, news sources, and social platforms, identifies every instance of your brand name appearing in published text, and reports it with source URL, sentiment classification, and reach metrics. This approach works because traditional brand mentions live on specific, crawlable, publicly accessible URLs — a news article, a forum thread, a social post — that a monitoring tool can find and index. AI brand mentions do not work this way. When ChatGPT recommends your business in response to a user query, that recommendation does not exist at a publicly accessible URL. It exists within the conversational context of one user’s session, delivered in real time, and — unless the user explicitly shares the conversation — disappears when the session ends. The recommendation is ephemeral by design. No crawling tool can monitor it directly. The implication is that AI brand mention tracking requires a fundamentally different approach: instead of monitoring what has already been published about you, you must actively query the AI systems on the questions your prospects are likely asking, observe whether and how you appear in responses, and systematically track those observations over time. You are not monitoring a published record — you are conducting systematic market research into a dynamic, query-responsive AI landscape. This shift from reactive monitoring to proactive querying is the conceptual foundation of every tracking method described below. Level 1: Manual Query Testing — The Foundation of Every AI Monitoring Programme Manual query testing is the most accessible, most reliable, and — for UK businesses beginning an AI monitoring programme — the most important starting point. It requires no specialist tools, no API access, and no technical resources. It requires only a structured approach to querying AI systems and a consistent methodology for recording and comparing results over time. Building your UK query set. The first step is constructing the set of queries your target audience is most likely to submit to AI systems when seeking a recommendation in your category. These are not the same as your SEO keyword set — they are conversational, specific, and decision-oriented in a way that keyword-formatted queries are not. For a UK digital marketing agency, relevant queries include: “What are the best SEO agencies in London for small businesses?”, “Which UK digital marketing agencies specialise in ecommerce SEO?”, “I run a professional services firm in the UK — which SEO agency would you recommend?”, “What should I look for in a UK SEO agency and can you recommend some?” — and dozens of variants covering different service types, client sizes, locations, and industry verticals. Build a spreadsheet of 30 to 50 such queries. Cover: direct competitor-framing queries (“best [service] in [location]”), problem-framing queries (“I need help with [specific problem] — which UK provider would you recommend?”), comparison queries (“[Provider A] vs alternatives for UK businesses”), and criteria-based queries (“what makes a good [service provider] for UK [industry] companies?”). The testing protocol. Run each query in ChatGPT (GPT-4o via chat.openai.com), Perplexity (perplexity.ai), and Gemini (gemini.google.com) on the same day, in fresh sessions with no prior conversation context. Record: whether your brand name appears in the response, where in the response it appears (first mention, middle, closing recommendation, not at all), what specific language the AI uses to describe your business, which of your competitors appear in the same response, and whether the AI cites any specific sources alongside its recommendation. Document everything in your spreadsheet with the date tested. This baseline dataset is the reference point against which every subsequent test will be compared — the foundation of your AI visibility trend