Prospecting around your existing customers: the neighbours who look like them
Your best next customers often sit a few streets from the ones you have. How to search around each customer, keep only look-alikes you do not serve, and give reps a reference.
Part of the guide AI employee for sales prospecting
Prospecting around existing customers means taking each business you already serve, searching a short distance around it for businesses of the same kind, removing the ones that are already customers, and handing the rest to the rep who covers that area with the nearby customer named as a reference. It works because neighbours share suppliers, visit each other, and trust a name they can see down the street. It also keeps routes short: a rep visiting a customer can call on two prospects on the same walk.
This article is for sales managers and reps in any business that sells to local premises: food and drink suppliers, cleaning and maintenance firms, software for shops and clinics, equipment rental. It covers choosing the radius, defining what counts as alike, an example from three cities, the AI employee, and the mistakes that make the list feel like spam to the people on it.
Why the neighbourhood works
A cafe owner in Barcelona who hears that the bakery two doors down switched coffee supplier is more likely to take a meeting than one who receives a cold email. A dental practice in Seoul that sees a neighbour using a booking system has already seen it working. Proximity turns a customer into proof, and a rep who can say 'we deliver to your neighbour every Tuesday' starts the conversation with something the prospect can check.
There is a cost reason too. Serving a new customer near existing ones is cheaper: the same delivery round, the same technician, the same rep's afternoon.
Choosing the radius and the kind
- Radius. Walking distance for dense streets, about 300 to 500 m. A short drive for business parks and suburbs, 2 to 5 km. The aim is a list a rep can work in one visit to the area, not a city.
- Same kind. Search the types your customer is: a bakery customer leads to bakeries, cafes and patisseries, not to every food business. Pick types from the list in Nearby places.
- Same shape. If your best customers are independents, add Independents only. If they are rated well, consider a minimum rating, knowing it drops every place with no rating at all.
- Not already yours. The step that matters most: remove anyone who is a customer, a lost customer you agreed not to call, or a prospect another rep is working.
An example across three cities
The following is made up to show the method. A coffee roaster supplies 120 cafes across Barcelona, Seoul and Atlanta. The flow searches 400 m around each one for cafes, coffee shops and bakeries. Before removing customers it finds about 1,100 places; many are found around two or three customers at once. After removing duplicates and current customers, 380 remain. A scoring step marks 90 as strong fits: independent, well rated, similar in type to the customer they were found near.
Each rep's Monday email lists their strong fits grouped by the customer they sit near: 'Near Cafe Mirador (customer since 2023): four prospects within 400 m'. In Atlanta, the rep combines delivery days with visits. In Seoul, the team uses the customer's name only after asking the customer, as is polite there and good practice everywhere.
The AI employee
- Start with a Schedule, every Monday.
- Bring in your customers with My data (a CSV with name, address and position), or from your own database with Database query.
- Add Nearby places on Each incoming place, passing on the places found, with your types and radius. Each place found carries a near field naming the customer it was found around.
- Add Remove duplicates, since a prospect between two customers is found twice.
- Add Match with my table against the same customer table, matched by website or phone, keeping only the places not found.
- Add Only new, so a prospect is sent once and the next weeks bring only newly found neighbours.
- Add an xMap LLM block that scores each prospect against the customer it sits near and writes a one-line reason to call that mentions the neighbour.
- Split the list with Split by territory between your reps, then Email each their own, with a map.
What arrives is a list per rep with the prospect's name, type, address, phone, website, rating, the customer nearby and the distance to it, the fit score and the reason. A map shows prospects clustered around customers, which makes route planning obvious.
Mistakes that make it backfire
- Naming a customer without asking. Some customers are happy to be a reference; some are not. Ask once, record the answer in your table, and let the scoring step use only names marked as allowed.
- Matching customers by name only. Names are spelled many ways. Website and phone matches are far more reliable, and the table lookup can compare either.
- A radius that is too large. At 5 km in a dense city, every prospect is near some customer and the reference means nothing.
- Forgetting competitors' customers are fair game but competitors are not. A rival supplier's showroom will appear in a broad type search. Leave such types out.
Questions
How far around each customer should I search?
About 300 to 500 m on dense streets and 2 to 5 km in suburbs or business parks. Small enough that the customer is a meaningful neighbour.
How do I keep current customers off the list?
Add Match with my table against your customer table, matched by website or phone, and keep only the places it does not find.
Can it use customers from my CRM?
Yes. Read them from your own database with Database query, or upload an export to My data and refresh it when the CRM changes.
What if two reps cover the same street?
Split by territory shares the prospects between named reps by area, so each prospect goes to one person.