May-2024 Release Notes

April showers bring... weeds and flowers? ๐Ÿค” This month, we exterminated a forest of weeds so that our new buds (data) can bloom! ๐ŸŒป Welcome to the May 2024 release notes.

Places Highlights

  • Deduplication model update removes 325k duplicate places! ๐Ÿ’ฅ
  • Significant lift in European places precision ๐ŸŒ
  • High quality polygons for over 2,000 US colleges and universities! ๐Ÿง‘โ€๐ŸŽ“

Places Growth

In the May release, SafeGraph has a grand total of 52,984,620 places, including POIs with or without geometry, closed POIs, and parking lots. This is a net decrease of -1,468,947 places from last month. We added millions of new places despite this net decrease, and the total count changes per country are a result of the following major initiatives.

Down with the Dupes!

For the past few months, we were working hard behind the scenes to release an update to our patented places deduplication model, and we are finally ready to unleash it on our data! This update removed 325k non-branded, duplicate places globally, and and the following 2-digit naics_code groups saw the largest improvements:

  • naics_code like '81%' (Services): -90k ๐Ÿ’‡โ€โ™€๏ธ๐Ÿ’…
  • naics_code like '44%' or '45%' (Retail Trade): -88k ๐Ÿ›
  • naics_code like '72%' (Accommodation and Food Services): -67k ๐Ÿจ ๐Ÿ”

Below is an example of a duplicate pair that is now resolved in the May release:

Impactplacekeylocation_namestreet_addressnaics_codelatitudelongitude
Removed from May release โŒ227-222@5pw-6bw-7kfSharks Fish and Chicken 79 Street100 E 79th St72251141.751257-87.621063
Remains in May release โœ…zzw-222@5pw-6bw-7h5Shark Fish & Chicken100 E 79th St72251141.7513-87.621127

European Precision

With the help of our customers, we identified a large set of low quality categories representing millions of our European places. We made some concerted efforts to improve the veracity of existing coverage - which included deleting some noise - and onboarded new sources to fill obvious gaps. As a result, the total POI count decreased in some countries and increased in others. Based on manual, ground-truth validations, we are pleased to report that the overall precision of European places increased by +17 percentage points! ๐Ÿ“ˆ

May Release Change Tracker

If you've read this far, you're as tired as we are ๐Ÿ˜‰. Below is a high level summary of placekey adds (net new POIs), placekey drops (intentionally removed POIs), and total POI count changes grouped by each major project and also by placekey "churn" (the unintended drops we report each month) and other small changes.

ProjectPlacekey AddsPlacekey DropsPOI Count Delta
Deduplication Model Update728k1.10M-372k
European Precision2.77M3.71M-940k
Monthly "churn" / other small changes112k270k-158k
Total3.61M5.08M-1.47M

The top 3 countries with largest POI count increases are:

Country (iso_country_code)Placekey AddsPlacekey DropsPOI Count Delta
Great Britain (GB)513k128k+385k
Spain (ES)485k342k+143k
Netherlands (NL)152k38k+114k

The top 3 countries with largest POI count decreases are:

Country (iso_country_code)Placekey AddsPlacekey DropsPOI Count Delta
Poland (PL)106k682k-578k
Italy (IT)14k367k-353k
Germany (DE)587k911k-324k

As always, you can visit our Places Summary Stats for more granular metrics.

Brands

We've added a total of 105 brands across 54 countries ๐ŸŒ including:

  • Bank of India (SG_BRAND_156067e8b88c1439) with 5,131 POIs ๐Ÿฆ
  • Migros Turkey (SG_BRAND_057af42a2a3e151c) with 3,117 POIs ๐Ÿ›’๐Ÿฑ
  • PicknPay (SG_BRAND_7e69eab7c01782de) with 2,254 POIs ๐Ÿ›’๐Ÿฅ™

๐Ÿ‘€ Are we missing a brand or country? ๐Ÿ‘€ Please let us know here!

Brand Openings and Closings

  • We rely on POI metadata to track store openings and closings, and we are especially interested in understanding open/close dates for branded POIs. It can take more than a month to infer open/close dates, so we report brand open/close metrics on a one month delay.
  • In this release, we flagged 1,944 brands with at least one store closure in March 2024 and 1,939 brands with at least one store opening in March 2024. Learn more about our open/close columns here.

Brands Known Issues or Bug Fixes

  • We had previously assigned naics_code = 311811 (Retail Bakeries) to 52 distinct brands, but after some feedback and further review, naics_code = 722515 (Snack and Nonalcoholic Beverage Bars) is a better fit for all 52 brands. See the full list here.

College and University Polygons

We've added precise campus polygons for over 2,000 colleges and university POIs across the US. These polygons improve the accuracy of our spatial hierarchy (parent_placekey) data, making it easier for users to query for and distinguish POIs that are located within universities (e.g., QSR locations on a university campus).

223-222@5pw-6c2-6kz - University of Chicago

223-222@5pw-6c2-6kz - University of Chicago