What is point-of-interest (POI) data?
In its simplest form, a POI can be any real-world location that a person may find interesting or valuable.
This ranges from anything like a stadium, park, tourist attraction, to something more simple like a grocery store, retail location, or ATM. It’s typically defined by its geographical location and other attributes like hours of operation or phone number. POI data also shouldn’t be seen as static – it will change as events occur, such as a polling place during an election or business closures.
Ultimately, it provides valuable and actionable information about a specific place. When searching for POI data, many companies selling POI data or POI APIs will refer to their products as “Places” or “Places API.”
Where does it come from?
For the most part, POI data providers rely on their own crawlers and algorithms to search the web and source POI data. An example would be a crawler searching for Starbucks and using all publicly available information from Starbucks to log and update POI data. Each provider will then validate their own data through a series of algorithms or neuro-learning platforms. In addition to crawlers, providers will often use additional sources such as:
- Businesses and Syndicators: Businesses and listing syndicators willingly provide their information so their specific POI may have the most up-to-date attributes for the end-user.
- Government Sources: Census data and other publicly-available government data are sourced and used to validate place data.
- 3rd party data and partnerships: POI data is licensed through third parties.
- Field Reps/Human Verification: Data providers employ individuals or teams to canvas an area to collect new place data and validate existing data.
- 1st Party User-Generated Content: With the rise of social media, companies have used user-generated check-ins to validate data across their Places data set. Companies like Foursquare have their own native apps like City Guide and Swarm that provide human-curated data that is later cleaned and validated by further human verification.
POI data IRL
Depending on the industry, POI data has abundant use cases and applications. Location data can be used to help businesses with their site selection and enable them to make an informed decision on where to open up a new location. For app and web developers, it can power navigation apps in real-time to a specific POI or recommend nearby POI data with rich content.
One should look at their method of delivery to understand the application of POI data. It can be transferred via two methods, the first being a flat file similar to a CSV but usually a JSON or TSV file. The second is through an API that allows developers to power their applications and websites in real-time by providing users with POI search and recommendations as well as rich content. Both delivery methods provide the user with plenty of places data but serve different functions for the use case and implementation.
A flat file is a static object containing all the data and is typically sent on a recurring basis. Flat files are great for retailers, food & dining and real estate who will use POI data to get a clear understanding of a target area. Having a clear view of an area helps influence things such as site selection, neighborhood assessments, trade area analysis, or competitive analysis. Although those tend to be the most common, POI data is also used by government agencies to optimize services or by marketing and ad-tech companies to better measure their advertising attribution.
Alternatively, an API functions in real-time within a developer’s app or website. This allows developers to deliver accurate POI data to users and push relevant content such as POI recommendations, photos, reviews, and tips in the surrounding area. Examples of this may be an app using your location to pull up nearby POI for recommendations or a ride-hailing app sourcing relevant drop-off points as the user types.
Evaluating POI data
POI data is tricky, and having the most extensive set of POI data doesn’t necessarily mean that a specific POI provider will have the best data set for your use case. When evaluating, you should clearly define your campaign’s goals and scope of work, and align each provider’s capability with your use case. There are many challenges to POI data, and the criteria below will help you determine what to look for in your specific use case.
- Accuracy: Accuracy is arguably the most critical metric because your data may be useless without it. You should assess how each provider classifies their POIs, attributes, and ultimately, the process and investment they put into validating their data. This will help ensure that the POIs listed are actual businesses, not a simple LLC operating out of their home or a business that has gone out of business.
- Recency: The recency of your data could actually indicate its accuracy and usability. POI data is ever-changing, and many businesses open and close over time. It’s essential to check with your provider and know the frequency in which they update their database.
- Coverage: A provider should have the necessary coverage in the area your use case focuses on. All the major providers have some global coverage, but some may excel compared to others on a market-to-market or vertical basis. While a provider may have the highest number of POI, they may lack fill rate and not provide the data your business is looking for.
- Delivery: Your data should be delivered to you in a way that best suits your needs. APIs allow you to develop powerful apps and user experiences but don’t allow for bulk analysis or optimization. Flat files, on the other hand, allow for a greater volume of data and the freedom to refine your data but require you to maintain your database and are static outside of their delivery frequency. Ultimately the best delivery method is the one that best suits your needs and use case.
In addition to the general points of reference above, checking each provider’s schema allows you to identify which attributes are most relevant to your use case and how granular their data is. All POI providers provide general amounts of data, but Foursquare sets itself apart by operating its own apps that also serve as a way to generate and validate their datasets.
POI data should be seen as a living data set that must grow and expand. Because the world is changing, your data will be too. Even the best machine learning algorithms will make mistakes. You should be able to see how your provider is aggregating and validating their data and what they’re doing to make sure you’re receiving the most accurate data.
At Foursquare, our award-winning API has consistently delivered more detailed and accurate geocodes than notable providers like Google, and a more flexible pricing structure that ranges from 25% to 75% less.
Foursquare also boasts 60+ data attributes and 1100+ categories – a significant difference to Google, which comes in at around 26 data attributes and 96 categories. The difference in data and attributes enables companies to drive better user experiences at a fraction of the cost. A real-life example is using a POI as a drop-off location. Ideally, when navigating to a POI or calling a ride-sharing app, users should be dropped off at the most convenient location for entry or designated drop-off point. Foursquare’s data not only takes into account street addresses but also locates the preferred drop-off location to ensure the user is left at the most convenient location. For ride-hailing apps like Uber or delivery services, this could make all the difference between a horrible transaction and a great one.
However, what ultimately sets Foursquare apart from other providers is our data restrictions and caching policies. Foursquare currently has no limitations on use cases, verticals, or use cases while allowing 24-hour caching for enterprise customers.