What Is POI Data? Points of Interest, Explained

POI data (points-of-interest data) is structured information about physical places: a store, restaurant, gym, or clinic, each tagged with a name, category, coordinates, and attributes like hours or brand. In retail real estate, it is the map of who is already on the ground, and it sits underneath your trade areas, competitor analysis, and demand modeling.

The term gets muddy fast, because "POI" and "point of interest" mean different things in different rooms. To a tourist, a point of interest is a landmark. In an acronym dictionary, POI can mean a person of interest or a military point of impact. This article is about the business and mapping sense: POI as a dataset of real-world locations that retailers, analysts, and site-selection teams use to decide where to open next.

What is POI data?

POI data is a structured dataset where each record represents one physical place. Every row carries, at minimum, a name, a category, and a pair of coordinates. That structure is what separates a POI dataset from a phone book or a pile of addresses: the category and coordinates make it queryable by "what is here" and "where exactly," not just "who is listed."

A single POI is one location, not one company. A national coffee chain with 3,000 stores shows up as 3,000 separate POI records, each pinned to its own address and geocode. That granularity is the whole point. When you are choosing between two shopping centers, you do not care that the chain exists somewhere. You care whether one of its stores sits 400 feet from the pad you are evaluating.

What is in a POI record?

A useful POI record answers who, what, and where with enough precision to act on. The core fields are consistent across most datasets:

• Name. The business or place name, ideally the operating name a customer would recognize, not just a legal entity.
• Category. What kind of place it is, from a taxonomy like NAICS codes or a provider's own scheme (coffee shop, urgent care, tire dealer). Categories are how you filter a dataset down to competitors, complements, or traffic drivers.
• Coordinates. Latitude and longitude, produced by geocoding the address. This is the field that lets you measure distance, draw a radius, and place the location inside a trade area.
• Address. Street, city, state, and postal code, used both for display and as the input to geocoding.
• Brand or chain. The parent brand, so you can roll 3,000 records up into one franchise footprint or compare two banners.
• Attributes. Category-specific fields: opening hours, phone number, whether a restaurant has a drive-through, whether a gym is 24-hour. These are the fields that vary most in quality between sources.

The first three fields are close to universal. The richer a dataset gets past them, the more useful it becomes for real decisions, and the harder it is to keep accurate and current.

Where does POI data come from?

POI data comes from four broad kinds of sources, and almost every production dataset blends more than one because no single source is complete, current, and accurate everywhere.

Mapping platforms. Google Places, Apple, and Foursquare maintain large POI databases built from business submissions, user edits, and their own collection. They are broad and reasonably current in dense markets, and access usually comes through paid APIs with usage terms.

Open datasets. OpenStreetMap is a community-maintained map with millions of POI records that anyone can use. The Overture Maps Foundation, backed by Meta, Microsoft, Amazon, and TomTom, publishes an open POI dataset that has become a common base layer. Coverage is strong in some regions and thin in others, depending on contributor activity.

Commercial aggregators. Providers such as Advan (formerly SafeGraph) and Esri clean, merge, and enrich POI data from many upstream sources, then sell it with coverage guarantees, category standardization, and freshness commitments. This is what most large retail and analytics teams buy when they need consistency at national scale.

Government and first-party records. Business licenses, tax rolls, and health-department permits are authoritative for existence and address, if slow to update. And your own store list is the highest-quality POI dataset you own: you know exactly where your locations are and how they perform.

For a wider view of the data categories that feed retail decisions, see our guides to site selection data and commercial real estate data.

How is POI data built and kept current?

Turning raw sources into a usable POI dataset takes four repeated steps: collect, geocode, deduplicate, and refresh. Each step is where quality is won or lost.

1. Collect. Pull records from the chosen sources. This is where coverage gaps enter, because a place missing from every upstream source is missing from the result.
2. Geocode. Convert each address into coordinates. Good geocoding puts the pin on the front door; weak geocoding drops it on the street centerline or the center of a large parcel, which quietly corrupts every distance measurement downstream.
3. Deduplicate. Resolve the same place appearing in several sources under slightly different names or addresses into one record. Under-merging leaves duplicates that inflate counts; over-merging collapses two real neighbors into one.
4. Refresh. Re-check the dataset on a schedule and retire closed locations. This is the hardest step, because the physical world changes faster than most datasets update, and a closed store looks identical to an open one until someone checks.

The quality pitfalls that trip people up

POI data is never perfectly complete or current, and the gaps are systematic rather than random. Knowing the failure modes is the difference between trusting a dataset and trusting the wrong parts of it.

• Staleness. Closed businesses linger. A restaurant can shut its doors and stay in a dataset for months, so a naive competitor count includes places that no longer exist. Every record needs a freshness date, not just a value.
• Duplicates. One place listed three ways (an old name, a new name, a suite variation) triple-counts, which distorts saturation math and cannibalization analysis.
• Miscategorization. A place filed under the wrong category disappears from the filter that should have caught it. A ghost kitchen tagged as a general restaurant, for example, will not read as delivery-only.
• Geocoding drift. Coordinates on the roof or the road instead of the entrance skew drive-time and distance calculations by exactly the margin that decides close calls.
• Chain vs. independent gaps. National chains are usually clean and well-attributed. Independents, the places that often define a neighborhood's character, are patchier, so a dataset can overstate how corporate an area really is.

The practical rule: a POI dataset is a strong starting point, not a verified fact base. For any location that actually drives a decision, confirm the records that matter rather than trusting the raw count. This is the same discipline behind fixing underperforming stores with location data, where a wrong assumption about the surrounding places sends you chasing the wrong fix.

How POI data powers site selection

In site selection, POI data is the substrate under the questions that decide a deal. On its own it is a list of places. Combined with foot traffic, demographics, and a scoring model, it becomes the answer to "what is around this site, and does that help or hurt."

Three uses do most of the work:

• Competitor and complement analysis. POI records tell you which competitors sit inside a candidate site's trade area and which complementary draws (a grocery anchor, a gym, a pharmacy) feed the same visitors. The category field is what makes this filterable.
• Trade-area characterization. Drawing a trade area is step one; describing what is inside it is where POI data earns its place, layered with demographics and foot traffic to describe the commercial texture of a catchment.
• Demand and saturation modeling. The density and mix of POIs, weighted by traffic, feed models that estimate demand and flag cannibalization before you sign.

This is where GrowthFactor treats POI data as an input rather than the deliverable. The platform pulls POI, foot traffic, demographics, and competitor data into a single site score that shows the inputs behind the number, so a real estate team can see why a location scored the way it did and defend it in any room. The data behind real estate analytics is only as good as what your team can act on, and a raw POI feed is the beginning of that work, not the end.

Frequently Asked Questions about POI Data

Here are concise answers to common questions about POI data from retail and real estate professionals.