Historically, types of geolocation in the marketplace have been of a couple different flavors:

1. IP Address-based

2. Fingerprint / Cookie-based

IP Address-based geolocation is inherently inaccurate and is not sufficient for resolving anything beyond a city level. The best IP geolocation services make no claims beyond about 85% accuracy for CITY level resolution.

Device fingerprinting and cookie-based geolocation methods attempt to use a combination of probabilistic and deterministic methods to match a user's device to a previously disclosed location. The process works something like this:

• A user willingly provides their address/name/email etc on a website. Unknowingly, this site also records a device fingerprint and/or sets a third-party cookie on their browser.

• When the user visits a different site, this data is accessed and used to match that visitor across the 2 sites and disclose their location.

This approach seems good initially, but it has some issues:

1. The user has not offered to share their information on the second site, using this information is somewhat of a legal gray area at the moment but will likely become impermissible as consumer protection laws ramp up.

2. Device fingerprinting is not unique enough. Very often 2 totally different users may have the same device fingerprint causing you to incorrectly identify the visitor on your site as someone who was never on your site.

3. Both device fingerprinting and third-party cookies are on their way out. Safari has already disabled third party cookies and device fingerprinting and Chrome is set to follow suit in the future. This effectively ends this as an option since Safari/Chrome represent the bulk of internet traffic.

Our approach is different from the start. We begin with user-consented GPS data, requested when they visit a website with our script installed.

When a user consents to provide their GPS location, we record it, along with a unique anonymous ID we generate, that allows us to determine when this person comes back to the website. We then process the GPS data through several steps:

1. We check the accuracy of the GPS measurement, if its not a high-accuracy reading we flag it so we know not to use it to determine exact physical locations. It is still a useful signal for determining interest and behavior down to about a neighborhood level - but never something we could use to market to a physical location.

2. We then check the provided location against previously provided nearby locations for that user. If this location is more accurate, we update our data to use it, if it’s less accurate we continue to use our last best measurement.

3. Finally, we take the best measurement we have for that user in that location and, if it’s accurate enough, we will use it to determine an address.

Next, we take our processed, high-accuracy addresses and use a combination of public records, and opted-in marketing databases to determine if an address is commercial or residential and the current resident(s) of that address. Resident data is then compared against opt-in email and direct mail data for current household email addresses.

At this point, we have a high accuracy, high confidence match of the residents of an address and all of their browsing and interest history on our website. All accomplished using user-consented data, and no third-party cookies or device fingerprints.