What POI data is, and what a business can actually do with it

POI data (points of interest) is a list of real places: shops, clinics, restaurants, offices, schools, fuel stations. Each record carries a name, a type, a position on the map and a handful of details such as the address, phone, website, rating, whether it belongs to a chain and whether it is open. A business uses it to count a market, find customers, watch competitors and choose where to open.

This guide is for owners, sales leads and analysts who keep hearing the term and want to know what is really inside such a dataset, how to read one record, and which everyday questions it can answer without a research agency.

What a single place record holds

Think of one record as an index card for a building's front door. A pharmacy in Toronto, for example, would come with fields like these:

• Identity: the name, and a stable ID so the same pharmacy can be recognised next month.
• Type: a category (Pharmacy), plus the wider group and sector it sits in (Health).
• Position: latitude and longitude, which is what lets you put it on a map, draw a circle around it or measure a distance.
• Contact: street address, city, phone number and website, when the business publishes them.
• Reputation: the average rating and how many reviews it has.
• Ownership signal: a brand name and a yes or no for belonging to a chain.
• Life cycle: open or closed, and, where known, the month it opened.

None of these fields is exotic on its own. The value comes from having the same fields for every place of a kind across a whole city, so you can compare, count and sort them.

What POI data does not tell you

It is just as important to know the edges. A place record is a description of a front door, not of the business behind it. It does not hold revenue, staff numbers, opening margins or how many people walk in. A rating says what reviewers thought, not what every customer thinks. An opening month is only present for some places, and a place that never published a phone number simply has none.

Coverage also varies by kind. Large, public-facing businesses such as supermarkets and hotels are almost always listed; a home-based bookkeeper may not be. Treat a count as "places listed", which is exactly how xMap's market pages word it.

Five jobs it does for a business

1. Sizing a market

The simplest question is "how many are there?" xMap's places database lists 866 pharmacies within 10 km of the centre of London and 519 within 10 km of the centre of Toronto (November 2026). The same count, repeated around several candidate sites, tells an expansion team where a category is crowded and where it is thin. See the London pharmacy page for how such a count is laid out.

2. Building a prospect list

A company that sells to businesses (coffee machines to cafes, software to clinics, fuel cards to garages) can turn a category in a city into a call list with phone numbers and websites, instead of buying an old directory.

3. Watching competitors

Run the same search every week and the differences are the news: a rival's new branch, a competitor that closed, a rating that slid. That is what an AI employee is good at, because nobody remembers to check by hand.

4. Choosing where to open

Places around a candidate site describe its surroundings: how many direct competitors, which anchors (supermarkets, schools, gyms) are close, and whether the rivals nearby are well rated or weak.

5. Keeping your own records right

Your customer list ages. Looking each customer up against current place data shows which ones closed, moved or changed their phone. The guide on cleaning a business list covers that job in detail.

How to read a place list without fooling yourself

1. Check the area first. A count within 10 km of a city centre is not a count for the whole metropolitan area, and a circle can include water or parkland.
2. Read the type as the database names it. "Cafe" and "Coffee shop" may be separate types; decide which you mean before you count.
3. Look at how many records carry the field you care about. An average rating across 300 rated places is solid; across 30 it is a hint.
4. Separate chains from independents before drawing conclusions. A market of 400 places where 40% are chain branches behaves differently from one where nearly all are owner-run.
5. Compare like with like: the same radius, the same type and the same month.

A worked example, made up for illustration

Say you sell point-of-sale terminals and want a first territory in Sydney. You search bakeries, cafes and small grocers within 5 km of your office. The list comes back with a few hundred places. You drop the chain branches (they buy centrally), keep those with a phone number, sort by number of reviews as a rough sign of trade, and keep the top 60. That is a week of calls, built from public place records in minutes. The numbers in this example are invented; the steps are the real ones.

Doing it with an AI employee in xMap Flows

In xMap Flows a flow is a short chain of blocks, and a flow that runs on its own is called an AI employee. For place data the usual chain is:

1. Schedule: every morning, or every Monday.
2. Area of interest: the cities or the areas you draw.
3. Nearby places: the types of place, from xMap's own database of more than 300 million places refreshed every month, with filters for chains, ratings and opening date.
4. Place classifier when you need one clean category per place.
5. Only new so each place reaches you once.
6. Email or Show on a map to deliver the list.

What lands in the inbox is a short list of places with their details and a map link, and on the next run only the ones that were not there before. To start from something ready, the templates New places every morning, only new and Market size of a category across cities do most of the setup for you.