Fletcher Berryman from SafeGraph explains why the dates POIs open or close are important for geospatial analysis, and why they can be hard to verify.
Fletcher Berryman from SafeGraph explains why the dates POIs open or close are important for geospatial analysis, and why they can be hard to verify.
Fletcher Berryman is a geographer and remote sensing specialist currently working as a Product Manager at SafeGraph. Previously, Berryman led emerging markets data startup dataPlor as Head of Sales, grew enterprise relationships at CARTO, and analyzed satellite imagery at Apollo Mapping. Outside of work, he is a Research Associate at the University of Chicago's Center for Spatial Data Science, co-chair of the largest geospatial meetup in the world (GeoNYC), and a year-round NYC surfer.
You may not always think much of a new store opening up or closing down, especially if it doesn’t sell anything you want or need. But whether a point of interest is opening, open, or closed can be incredibly valuable information in use cases such as investing, urban planning, and retail site selection.
For example, if you own a retail chain, you may want to place a new store near an area that has seen a surge in foot traffic. But how do you know that surge isn’t caused by the recent opening of a competing store? Conversely, let’s say you’re thinking of closing down stores in an area, or are hesitant to open new stores in an area, because there are too many competitors nearby. But if some of those competitors aren’t operating anymore, and you’re not aware of that, you could end up making very erroneous and costly decisions.
SafeGraph product manager Fletcher Berryman goes through scenarios like these as he makes a case for why POI (points of interest) open and close data is a critical component of geospatial analysis. He also discusses some of the challenges in accurately determining whether a POI has opened or closed, as well as some ways geospatial data scientists can get over these hurdles. Here’s a brief look at what he’ll be talking about in this webinar:
To give some opening context, we’ll describe what we mean by “open and close” (or “open/close”) data, and give a general explanation of why it’s important for geospatial analysis.
Open and close data refers to a set of descriptive attributes for points of interest data. It denotes the date when a place first opened to the public, as well as the date when it ceased operations and closed down. Obviously, the latter date will be null if the place hasn’t actually closed yet.
Open and close data may not be all that helpful when it refers to just a single POI. However, it can become immensely valuable when used to compare many POIs across categories, geographies, brands, and/or industries. This can provide insight into broader social or economic trends within a given area over a specified time period.
So how specifically can open and close data be used in geospatial analysis? Furthermore, how do geospatial data scientists determine precisely when a POI opens or closes, and why is this so difficult? The answers to these questions are five important insights you can take away from this webinar.
Time in Video: 11:57
Open and close dates, among other attributes, are indicators that POI data can be just as dynamic as the foot traffic data it’s often contrasted with. Places don’t just “always exist”; they need to be conceptualized and built up over time. And they don’t remain static in the same place forever.
This is especially true of stores. They could change their branding or operating hours; they could move to other locations or go out of business; and the buildings they occupy could be renovated or even demolished. All of these changes can be measured across space and time to show broader patterns.
Time in Video: 14:00
Foot traffic data doesn’t necessarily reveal much unless it’s put in the context of the actual physical places nearby. Open and close data can form part of this context. For example, it could explain footfall in an area increasing because new POIs have opened up and curious residents want to have a look at them. Or it might show foot traffic dropping because a popular POI has closed or moved somewhere else. This type of information can inform things like investment decisions in certain brands, businesses, or industries.
Time in Video: 15:24
Another scenario in which open and close data can be useful is when mobility data, or another type of geospatial data, is incomplete. This can happen in certain countries that may pay relatively less attention to detail in cataloging geospatial data, especially in remote areas where POIs are sparse anyway. It can also happen when POIs for a specific brand are confined to a particular country, region, or city.
In these circumstances, it may be comparatively easier to locate a POI (or branded set of POIs) based on whether it’s open or closed. This is because there’s usually some sort of clue or record pointing to it opening, being open, or closing. For example, an open POI will usually have a functional website that’s updated periodically, as it helps the owners advertise their POI’s location, products/services, prices, contact information, and so on. The website may even have announcements relating to a new store about to open, or a store planned for closure.
Time in Video: 18:38
Open and close data can even be used as a tool to validate a brand’s reporting of its financial situation. In response to market circumstances, certain companies may either close stores temporarily or significantly shift their store footprint in an area. Others, meanwhile, may expand as economic conditions improve, or find creative new business strategies for keeping stores open.
A good place to start is with a brand’s quarterly earnings report. Check this information against reliable POI data, and filter out the stores marked as closed for a brand. Then you can calculate the correlation coefficient between store counts in the open/close data and what the brand reports over the same time period. This can give you an idea of how accurately a company is assessing its own performance.
Time in Video: 21:07
There are several challenges to accurately determining when POIs open or close. One is in processing the correct metadata from multiple information sources, which can be complicated by things such as a brand changing how the store locator on its website works. Another is how brands are often more reluctant to report store closures than store openings, for PR reasons. This can mean there’s a delay in getting some of this information.
By far the biggest obstacle, however, is in differentiating between a POI that has permanently closed and one that has closed only temporarily. This is especially common in places where there are few other signals that can give hints as to the status of a POI. That is, a place may look like it’s closed from the street, but it may not be clear for how long (e.g. an hour? The rest of the day? The weekend? A week? A month? A season? Permanently?). Or it may not technically be ‘closed’ and merely changed locations. This makes it necessary to weigh different signals over time to accurately determine whether a POI is either temporarily or permanently closed.
Open and close data for POIs isn’t always a complete geospatial analysis solution. However, it’s a useful tool that can add context to – or even sometimes fill in for – other geospatial data. For example, it may help to explain why foot traffic has increased, decreased, or shifted in an area as POIs open, move, or close. This could be informative for commercial real estate investors, for instance, as they weigh how much footfall an area gets versus how long POIs tend to last at nearby locations. Or a retail chain manager may see an opportunity to fill a gap left by a certain type of business that recently closed in, or moved away from, an area.
In another use case, open/close data might be able to locate POIs that other geospatial data doesn’t point to yet. Predominantly in developing countries (and even in some well-developed ones), there can be gaps in geospatial data for remote areas or hyperlocal brands. If there is foot traffic in an area, but not enough to confidently determine the presence of a POI nearby, it may be helpful to search for some sort of other digital footprint for that place.
To illustrate, geographers could do a crawl or search of the Internet to find a website for that location, or some other page that mentions it. Then they could analyze that information for clues, including how recently it was posted or last updated, as well as any explicit announcements of store openings or closings. Some brands may also have store directories or locators on their websites. All of this can help to find a POI that hasn’t been registered in other geospatial data yet, as well as determine whether it’s still open or has closed.
One other way to use open and close data is to check a brand’s location footprint against what the company actually reports. This can be especially informative during volatile economic times, when a company may significantly scale back or shift its operations (or even aggressively scale up, in some cases) but not always be prompt in announcing these changes. Looking at the correlation between a brand’s open/closed POIs and what is included in their public financial filings may give a more accurate picture of the brand’s situation and short-term intentions.
SafeGraph includes open and close attributes as part of our Places dataset. This dataset gives information about millions of points of interest around the world, including parking lots in the US. It is updated monthly to account for the frequency with which information changes about POIs, including if they newly open, move, or close.
We track open and close data using three attributes. “opened_on” denotes the month and year a POI opened, as it first appeared and was flagged in our data source pipeline. “closed_on” shows the month and year a POI closed, once we have enough metadata to confirm a closure within this timeframe. If there is no date for this attribute, it is assumed the POI is still open. “tracking_closed_since” indicates the month and year we began reporting a POI as being closed, as it may take some time between first flagging a place as potentially closed and confirming when it closed.
We can use a number of different metadata sources to verify these attributes. For example, we can look at the quarterly financial reports of publicly-traded companies to get an idea of how many of their stores are still open. For smaller private brands, we can check a brand’s website or social media feed for announcements of store openings or closures. A brand may also have a store directory or locator on their website that will show which of their stores are open. Other times, we can source information from people who are actually at specific locations and collecting information about POIs with mobile apps.
See our Places dataset documentation – particularly the section on how we designate POIs as open or closed – for more information.
Fletcher Berryman is a geographer and remote sensing specialist currently working as a Product Manager at SafeGraph. Previously, Berryman led emerging markets data startup dataPlor as Head of Sales, grew enterprise relationships at CARTO, and analyzed satellite imagery at Apollo Mapping. Outside of work, he is a Research Associate at the University of Chicago's Center for Spatial Data Science, co-chair of the largest geospatial meetup in the world (GeoNYC), and a year-round NYC surfer.
