The geographic location a user is searching from can be one of the most important indicators of their value to a paid search program and Enhanced Campaigns now allow advertisers to target users based on their location in a variety of ways. Average household income levers, proximity bidding, and betas measuring the in-store impact of ads are all powered by the measured location of the user, not to mention traditional geographic location targets.
But how much of an advertiser's traffic can Google measure precisely? We took a look at a few advertisers and found some interesting trends.
More Users Now Tracked at Zip Code Level
While the most granular location ID Google's API click performance report will pass is at the city level, geographic reporting in the dimensions tab of the AdWords UI gets all the way down to the zip code level. Using this, we can tell how much traffic Google is able to attribute at the most granular level it can currently report on.
Here's a look at the percentage of United States AdWords traffic attributed to zip codes for three large advertisers since Q1 of 2012, the first quarter that data was made available at a zip code level.
As you can see, these advertisers have seen remarkably similar trends every quarter for the past two years. While the dip in zip code level attribution in the middle of 2013 is difficult to explain, 2014 has seen Google tracking the location of users more precisely than ever.
However, even with all of the insight Google has, including through the Android OS, it has still only been able to track about 57% of users to the zip code so far in Q2 this year. That figure matches up well with about how much traffic we're seeing attributed to HHI income levers, which are based on zip code average income. Thus, it seems Google is reporting on pretty much all of the traffic that they're able to measure at this level.
What Traffic is Google Not Able to Locate?
While we don't know all the details of when a searcher's location is more difficult for Google to track, digging into location data passed via click performance reports does offer us one clue. Across the board, advertisers typically see about 3% of desktop traffic attributed by Google to the state and national levels, the least precise possibilities. By contrast, that figure can be between 8% and 15% for mobile clicks.
This indicates that Google has a harder time placing searchers who are on mobile devices as precisely as they are able to place desktop searchers. Thus, we can probably assume that the share of traffic that cannot be attributed to a zip code likely contains a disproportionate amount of mobile clicks compared to the percentage of mobile traffic for the entire account.
However, mobile traffic is also likely where Google is making gains in terms of locating users, with newer devices with GPS capabilities making tracking easier than in days past. Thus, mobile may not be a weak spot for long.
What Do Google's Geo Tracking Capabilities Mean for Advertisers?
One key takeaway is that Google's ability to measure the location of a searcher is clearly changing over time. Whether this is the result of Google leveraging new measurement techniques, or simply the product of more geo-measurement friendly devices coming onto the market, it means that advertisers need to account for this change as they measure performance in the years to come.
For example, if measurements for airport proximity bidding had existed in Q2 of 2013, an advertiser utilizing such metrics might see a 30-40% lift in traffic from searchers around airports year over year just as a result of Google being able to measure more precise location for more individuals (using the percentage increase in share of traffic attributed to zip codes as a proxy for what the increase in airport proximity measures would look like). Thus, conclusions from historical measurements need to be based on data that accounts for this natural increase in measurability over time.
Google's growing ability to measure more precise attributes of searchers also extends beyond geographically-based optimizations. While Google can attribute gender and age to those logged into Gmail accounts based on user-supplied information, they're forced to use less precise indicators for other users. Google's ability to more correctly segment searchers into these and other buckets is likely to change in time, and will thus impact advertisers' measures of these segments.
Related to this is that while Google is in the process of rolling out all of these amazing new targeting options, they are certainly not perfect because Google can only measure so much. That's not to say that Google isn't doing a pretty incredible job of creating targetable data points around searchers, which they are, but rather that the technology that powers these measures does not allow for 100% of traffic to be split up in each bucket.
And while Google does provide estimations to fill in the holes with some metrics, such as cross-device conversions, they are just that: estimations. The numbers in these measurements, then, should not be taken as perfectly accurate for the time being.
It's sound to assume that Google is going to keep getting better at this stuff, based on the data above and the push for even better targeting. How Google goes about pushing their ability to track user attributes to the next level is going to be limited to an extent by privacy concerns, but I'm sure they'll find responsible ways to give their advertisers improved data in the years to come.