April-2024 Release Notes

We forgot to release new data this month ๐Ÿ˜ž... Now that we're past the cringey yet obligatory April fool's joke, welcome to the April 2024 release!

Places Highlights

  • Key source refinements add hundreds of thousands of real POIs and remove hundreds of thousands of "fake" POIs ๐Ÿ“ˆ
  • +204 new brands across 67 countries ๐ŸŽ‰
  • Polygon source updates:
    • Improvements to Geometry precision across the US ๐Ÿ‡บ๐Ÿ‡ธ
    • Massive improvements to Geometry coverage in Spain ๐Ÿ‡ช๐Ÿ‡ธ

SafeGraph evaluates the quality and completeness of our US POI data on a quarterly basis using two key metrics:

  • Coverage Rate: The percentage of real and open SafeGraph POIs compared to the industry standard. In the US, this is Google.
  • Real Open Rate: The percent of SafeGraph POIs we claim to be real and currently open compared to those that actually are real and open.

Each quarter, we randomly sample two zip codes (one urban and one rural) to measure our accuracy. This month we're looking at:

  • 60601: Chicago, IL - Representative of Urban
  • 80863: Woodland Park, CO - Representative of Rural

Here's the summary (full breakdown on the Metrics page):

Geo AggregationCoverage RateReal Open Rate
Total Sample83%68%
60601 (Urban)84%66%
80863 (Rural)79%74%

Check out our Accuracy Metric Methodologies for more detailed info on how we arrived at these calculations.

Interested in a particular zip code for a future measurement? We're taking suggestions!

Places Growth

This month, SG Places has a grand total of 54,453,567, including POI with or without geometry, closed POI, and parking lots. This is a net decrease of 93,965 places from last month.

We refined our collection and ingest strategies for some key global sources, and this yielded +196k POIs in Canada ๐Ÿ‡จ๐Ÿ‡ฆ , +46k POIs in ๐Ÿ‡จ๐Ÿ‡ณ , and +37k POIs in France ๐Ÿ‡ซ๐Ÿ‡ท among other countries. Aside from this effort, new brands is always our favorite tool for adding new places. This month, we added a grand total of 204 brands across 67 countries including:

  • Agricultural Bank of China (SG_BRAND_2a4d94ba2c983f02) with 20,882 POIs ๐Ÿฆ
  • OYO (SG_BRAND_869bdbc72aa33ba6) with 2,022 POIs ๐Ÿจ
  • International Paper Company (SG_BRAND_5abf2e68c3de0817) with 322 POIs ๐Ÿ“ƒ
  • Beter Bed (SG_BRAND_ea2aebbc773f48cb) with 99 POIs ๐Ÿ›
  • Complete Nutrition (SG_BRAND_c4c6bd3f2b3b8146a7544065700755c4) with 29 POIs ๐Ÿ’ช

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

Refining key sources also resulted in better detection and removal of low quality POIs. We take pride in ridding our product of junk, so these net decreases are worth celebrating ๐Ÿฅณ. The following countries saw the largest net decreases:

  • Italy: -192k POIs ๐Ÿ‡ฎ๐Ÿ‡น
  • US: -77k POIs ๐Ÿ‡บ๐Ÿ‡ธ
  • Portugal: -23k POIs ๐Ÿ‡ต๐Ÿ‡น
  • Germany: -15k POIs ๐Ÿ‡ฉ๐Ÿ‡ช

Of course, you can always visit our Places Summary Stats to find more details on our continued growth.

Geometry Improvements

This month, we made two major updates to the building footprint datasets that power the Geometry product.

In Spain ๐Ÿ‡ช๐Ÿ‡ธ, we added a nation-wide building footprint dataset. Our coverage of non-synthetic polygons in Spain rose from 3% of POIs in March to 91% in April ๐Ÿš€.

Some of our new building footprints in Spain.

Some of our new building footprints in Spain.

In the US ๐Ÿ‡บ๐Ÿ‡ธ, we ingested a new, more granular nationwide building footprint dataset. This new dataset not only reduced our US synthetic polygon rate from 3.3% in March to 2.6% in April, it also increased the overall precision of the polygons we are providing for our POIs. As shown in the image below, the new dataset features a higher density of polygons, and polygons are more granularly split along buildings' rooflines, allowing us to better capture each POI's unique footprint.

Our previous polygons (in red) vs. our new polygons (in blue/green)

Our previous polygons (in red) vs. our new polygons (in blue/green)

Drops โฌ‡๏ธ

  • We are ingesting many sources and due to source changes and processing changes, Placekeys do drop over time. In this release, we dropped 2,570,861 Placekeys (187,606 branded and 2,383,255 non-branded).
    • We made some widespread improvements to location_name for transit stop POIs (naics_code like '485%' and geometry_type = 'POINT' ), which accounts for 1.54M drops globally. The POIs are still in the product, but they unfortunately have a new placekey. Please reach out to your customer success manager for help mapping back to the original placekey if this is a nuisance.