I asked three different AI tools to research my market. Here's what happened.

THIS WEEK: Market research used to mean either paying someone else to do it or spending a weekend buried in spreadsheets. AI has genuinely changed that. This week is a practical session, a real example showing exactly how to use AI tools to size up a market, map out competitors, and understand what customers actually think, using nothing but the free tools available right now.

Freelance market research for a small business typically runs from around £43 for a basic competitor summary up to £400 or more for anything comprehensive, and that is before you get anywhere near agency territory, where a proper UK market research project can run from £15,000 into the tens of thousands. Most small business owners never commission any of it. They guess instead, based on whatever they happen to notice about their competitors and whatever their customers happen to tell them face to face.

AI tools close a lot of that gap. Ask the right questions in the right order and you can get a genuinely useful picture of your market, your competitors, and your customers, built in an evening rather than commissioned and waited on for weeks. The tools are not a replacement for talking to real customers, but they are a serious upgrade on guessing.

Researching a Market Before Opening a Second Location

To show how this actually works, here is a walkthrough using a genuine scenario. An independent coffee roastery with one shop in a market town is considering opening a second unit in a nearby town, and wants to know if it is worth it before committing to a lease.

Step One: Sizing the market

The first question is whether there is enough demand to justify a second site at all. In Perplexity, the prompt was: "What is the current state of the UK specialty coffee shop market in 2026, and what trends should a small independent roastery be aware of before opening a second location in a market town in the Midlands?"

Perplexity's answer came back with cited sources covering UK coffee shop growth trends, consumer spending patterns on specialty coffee, and the shift toward independents over chains in smaller towns, each claim linked back to where it came from. That last part matters. Every figure could be checked against its source rather than taken on trust, which is not something a plain ChatGPT answer gives you by default.

Step Two: Mapping the competition

Next, the actual competitors in the target town. Five independent coffee shops and two chain locations were identified through a normal Google Maps search first, then their websites, menus, and Google review pages were pulled together and pasted into Claude with the prompt: "Here are the websites and reviews for five competitors in [town]. Build me a comparison table covering pricing, opening hours, what they are known for, and what their reviews complain about most."

The result was a genuinely useful table. It flagged that three of the five competitors had no evening opening hours at all, and that a recurring complaint across two of them was slow service during the Saturday rush, exactly the kind of gap a business can plan around before it opens rather than discovering the hard way.

Step Three: Understanding the customer

The final step was pulling real customer reviews, not for the roastery's competitors this time but from its own existing shop, and asking ChatGPT to identify recurring themes: "Here are 40 reviews of my coffee shop. What comes up most often, both positive and negative, and what would customers most want to see if I opened a second location nearby?"

This surfaced patterns that are genuinely hard to spot by reading 40 reviews one at a time: repeated praise for a specific staff member, repeated requests for oat milk options that were already on the menu but clearly not well signposted, and several mentions of wanting somewhere to work with a laptop for an hour, a gap none of the five competitors were filling either.

None of this replaces standing in the target town on a Saturday and watching who walks past. It does mean that Saturday is spent testing a hypothesis rather than forming one from scratch.

Perplexity is built specifically for this kind of research because every answer comes with sources attached, which turns "AI told me" into "AI told me, and here is where it got that from." The free plan gives unlimited basic searches on its standard Sonar model, plus five Pro Searches a day for more complex, multi-step questions. The paid Pro plan, at $20 a month, adds access to premium data sources including PitchBook and Statista, which is worth knowing about if the free tier's daily Pro Search limit starts to bite. Link: https://www.perplexity.ai/hub/pricing

Claude is the stronger choice once you have raw material to work with, competitor websites, reviews, spreadsheets of data, because it is better at synthesising a large pasted document into something structured like a comparison table. Web search is included on the free plan, not held back for paid tiers, and the free plan runs on Claude Sonnet 5 with a 200K token context window, large enough to paste in a genuinely long set of reviews or several competitor pages in one go. Link: https://claude.ai

ChatGPT is the most familiar option and remains a solid choice for the customer sentiment side of things, spotting patterns across pasted reviews or survey responses. The free plan runs on GPT-5.5 Instant with web browsing included, though it is capped at 10 messages every 5 hours before dropping to a lighter fallback model. Link: https://chatgpt.com

Pick one genuine question about your market, something like "who are my real competitors within five miles" or "what do customers complain about most in my industry." Put the exact same prompt into two different AI tools on the same day and compare the answers side by side.

Where they agree, you have got something worth acting on. Where they disagree, that is exactly where you need to dig further yourself before trusting either one.

NEED HELP? In over your head? Wearing too many hats already? No idea where to start? I can help you. Alternatively, feel free to WhatsApp me on 07800596333!

Companies House (find-and-update.company-information.service.gov.uk) is free and lets you look up the filed accounts of any UK limited company, including most of your competitors. Smaller companies often file abbreviated accounts that leave out turnover, but even a basic filing shows you who the directors are, how long the business has been trading, and whether accounts are being filed late, which is sometimes a useful early signal on its own.

Google Business Profile Insights, sitting inside your own free Business Profile dashboard, shows how customers are finding you and what they search before they land on your listing, which is worth comparing against whatever the AI research turns up about broader market demand.

AI research tools will occasionally get things wrong with complete confidence, stating a figure or a fact that sounds entirely plausible and simply is not accurate. Perplexity's citations make this easier to catch because you can click through and check. ChatGPT and Claude answers without sources attached need the same scepticism you would apply to a stranger's opinion at a networking event, useful as a starting point, not as the final word.

Used properly, these tools do not replace market research. They replace the version of market research most small businesses were doing anyway, which was guessing, and give you something closer to an educated first draft to test against reality.

Need someone to do this properly rather than doing it yourself? I have been doing this for 17 years. Get in touch. Alternatively, feel free to WhatsApp me on 07800596333!