← All use cases / Small business
A small business owner used Claude's help with the analysis to work out that, over about nine years of trading, they'd worked with three-quarters of all universities in the UK. It gave them a rare chance to step back and see what they'd built.
I run a small business that sells to [TYPE OF CUSTOMER, E.G. UNIVERSITIES, SCHOOLS, CAFES] in [COUNTRY OR AREA]. Attached is my client list from [YEARS]. Please: 1) match each client to a single organisation, merging duplicates and different spellings; 2) compare them with this full list of [ORGANISATIONS IN MY MARKET]: [PASTE OR ATTACH]; 3) tell me how many and what percentage I've worked with, broken down by [REGION / TYPE / YEAR]; 4) list the ones I've never worked with. Flag any matches you're unsure about.
Replace the [PLACEHOLDERS] with your own details before you send it.
How to try it
- Export your past clients or invoices into a spreadsheet, with names, dates and amounts if you have them.
- Find a full list of the organisations in your market, such as an official register or a public directory.
- Give both lists to Claude with the prompt above and ask it to match them carefully.
- Check the uncertain matches yourself, especially organisations that have merged or changed name.
- Use the result: celebrate the reach, and look at the ‘never worked with’ list for new leads.
Why this works
When you’re busy running a business month to month, it’s hard to see the bigger picture. Client lists build up in invoices, emails and spreadsheets with inconsistent names, and matching them against a whole market by hand is slow. An AI can do the tidying and matching in minutes, leaving you to check the edge cases. For the owner who shared this, the answer was a genuine morale boost after years of hard work. The same analysis also shows where the gaps are, which feeds straight into a simple marketing plan with AI.
Tips to get a better result
- Ask it to show its matching table, so you can check every pairing it made.
- Break the results down by year to see whether you’re still reaching new customers or mostly serving existing ones.
- Remove personal contact details before uploading, if you don’t need them for the analysis.
- Our guide to analysing customer feedback with AI can help you understand why customers come back.
Common mistakes
- Using an incomplete or out-of-date list of the whole market, which skews the percentage.
- Counting different departments of the same organisation as separate customers.
- Sharing confidential client data without thinking about your own privacy obligations.
Try a variation
Here’s a variation of our own to try next:
Using my client list, show me which of my customers haven't bought from me in over [NUMBER] months, sorted by how much they used to spend, and suggest a friendly reason to get back in touch with each type.
Related
- AI for work and business: our hub of ideas
- AI for small business: where to start
- Make a simple marketing plan with AI
More small business use cases
Brainstorm shop slogans and product descriptions with ChatGPT
A ZDNET writer set up a print-on-demand Etsy shop in under a week, using ChatGPT to brainstorm slogans, product names and descriptions that he then edited heavily. He's candid that sales were poor and ChatGPT didn't help with Etsy SEO.
Write 20 [TONE, E.G. WITTY] slogans for a shop called [SHOP NAME] that sells [PRODUCTS]. Then write a product description for [PRODUCT] covering [MATERIALS, SIZES, CARE] in a [STYLE] voice, under 150 words.
Clear a backlog of unanswered customer WhatsApp messages with ChatGPT
A freelancer spent an hour on a local business owner's WhatsApp chats that had been left unanswered. They used ChatGPT to draft clean, friendly replies based on what the business actually sells, tidied them up and sent them over, so customers stopped feeling ignored.
I run [BUSINESS], which sells [PRODUCTS/SERVICES] at [PRICES OR PRICE RANGE]. Opening hours: [HOURS]. Delivery or collection: [DETAILS]. Here are some customer messages I haven't answered yet: [PASTE MESSAGES, REMOVING PHONE NUMBERS AND SURNAMES]. Draft a short, friendly reply to each one in a [WARM / PROFESSIONAL] tone, apologise briefly for the slow reply where it's been more than a day, and only use the information I've given you. If a message needs details I haven't provided, write [CHECK] instead of guessing.
Find new customers in local building applications with Grok
The owner of a trucking business used to spend tens of hours a week looking for house construction and renovation applications that had been lodged or approved, to find work. They now hand that search to Grok Bot, and said it has saved them so much time that they've made their veg garden ten times bigger.
I run a [TYPE OF BUSINESS, E.G. SKIP HIRE, HAULAGE OR LANDSCAPING] business in [AREA]. Search the public planning or building application listings for [COUNCIL OR AREAS] for applications lodged or approved in the last [NUMBER] days that involve [TYPES OF WORK, E.G. NEW HOUSES, EXTENSIONS OR DEMOLITION]. For each one, give the address, the date, a one-line description of the work, the application number and a link to the official listing. Put them in a table, newest first, and mark any where you're unsure of the details.
Reference: X @soppystern