I’m working with POIs that are tanning salons and interested in foot traffic patterns - however, about 75% of our salons with foot traffic appear to have visits in the middle of the night when the vast majority of them are closed

Hi all, I’m working with POIs that are tanning salons and interested in foot traffic patterns - however, about 75% of our salons with foot traffic appear to have visits in the middle of the night when the vast majority of them are closed. I initially thought this was just a small amount of random noise, but about 2% have 100+ visits per hour at these odd hours (which is big for these types of establishments in SG data). Googled a few of these, which appeared to be in the first floor of an apartment or larger commercial building. I’d really love some help with this issue! Thank you! @Spencer_V

I think (@Spencer_Vail_SafeGraph correct me if I am wrong) that the vertical resolution of the GPS pings isn’t always so good.

I had the same problem with businesses at the bottom of apartment or condos.

@Mackenzie_Wehner_MD_Anderson Thanks for all the info on this. We will look into it more, and hopefully can get back to you in the next day or two.

In the meantime… There may be a way to work around the issue. In particular, you may be able to correct for and remove the effects of the foot traffic from people who live there, because they would show up every week/month relatively consistently. However, there could be a problem if you also have the same customers going to the salons regularly.

Thanks for the replies! I am hopeful these aren’t people who live in the apartment building, given that SafeGraph designates where each phone’s ‘home’ is and should not be coding that someone spent all night at a tanning salon every time they were at home asleep…

@Spencer_Vail_SafeGraph I could really use your or someone on your team’s help with this!

Hi @Mackenzie_Wehner_MD_Anderson, we’re looking into this. Sorry for the delay, but hopefully we will have some resolution early next week.

Hey @Mackenzie_Wehner_MD_Anderson, just wanted to let you know that we’re still looking into this!

Hey @Ryan_Kruse_MN_State, thanks for keeping me updated. Sorry I didn’t realize you were with SafeGraph at first. Would be happy to email you examples etc if that would be helpful.

Hey Mackenzie, heard back from our product team and unfortunately we don’t have great vertical data so they would look like they are all visits to the ground floor retail while they are more likely to be associated with the residential areas on higher floors.

Our recommendation would be to filter out the visits during times you would not expect and those that have a higher than expected dwell time. Or exclude these locations from the analysis altogether as outliers.

cc @Jeff_Ho_SafeGraph

Hey @Spencer_Vail_SafeGraph @Ryan_Kruse_MN_State, thank you all for looking into this. I’ve been so excited about using this data to better understand consumer use of tanning salons (in our work to prevent skin cancer). I am using a tiny portion of data (~1600 POIs), and I know tanning salons might not be typical POIs for SafeGraph - do other brands or POI types have this same issue with middle of the night visits when they are closed? I am not trying to nitpick, I think these data are great and I’m delighted to have access to them, but it seems that being unable to distinguish between retail visits and people being at home in their apartments would be a big issue for SafeGraph data overall. I would love your continued insight on this and really appreciate the help.

@Mackenzie_Wehner_MD_Anderson I believe someone on the product team is looking into this further, but there is no timeline on it because the fix is not necessarily easy. Ideally, we would know when apartments are stacked on top of tanning salons (and other POIs) so that visits can be more correctly attributed at nighttime hours. We’re very sorry that this issue is an obstacle to your work.

Generally, this could happen with other POI types, but I would not expect it to be as pervasive as you are seeing with tanning salons. Is there a particular region you are working with? Perhaps a workaround could be limiting to tanning salons in regions where this would not be a problem—like smaller cities with less dense populations? I’m happy to help you come up with a reasonable path forward.

Hi @Ryan_Kruse_MN_State @Spencer_Vail_SafeGraph, thanks for the updates. We went ahead and looked at all the POIs we have in the NAICS code that tanning salons fall info (812199) - we have about 34,000 with foot traffic, and 60% of them have visits every hour 12am-5am. I fully realize that your business model does not center around catering to researchers using the data for free. However, I’m having trouble with validity - I want to be able to use your data for public health, but I’m concerned that this issue is pervasive. I doubt that it’s limited to big cities, or to tanning salons. I would love to have a meeting with someone on the product team, if that’s possible.