Google captures location data by zip codes: Some of the implications

[Dave]

The other day I took a hard look at some google adwords data for a campaign and was startled to see that about 1/2 of 80,000 "impressions" for searches over the last 3 months were identified by zip codes. The other 1/2 were identified by city name and then a relatively few didn't have location data.

1/2 of locations for 80,000 impressions by zip code. That is a lot and very granular. For city locations for this campaign we could peer into the city itself and see which zip codes were generating more searches for the industry/vertical. I then went back to our proprietary data for leads and sales. Similarly higher numbers of both leads and sales from the zips where the most search phrases were generated and we had similarly high numbers of leads and sales.

I looked at the same adwords campaign for year to date--> about 500,000 impressions. I saw about 43% leads with zip data. Could it be that google continues to improve its drill down location data as time moves on?

Wow!!! three important locational data points and they all coincided. I saw a similar pattern with the zip codes for a larger regional "town" with 3 major zips and 2 smaller zips. Lots of search phrases, lots of leads, and lots of sales.

For the bigger city analytics wasn't a help. Currently analytics drills down to the town/city level. But not zips. Nevertheless google has that higher level of granular data. They might not be showing it currently...but its there.

Per this article, though analytics might be able to drill down into the zip code level via Universal Analytics and Dimension Widening. Google Analytics for Lead Generation in 2013 and Beyond (Part 1)

I haven't looked at that yet. has anyone here implemented this and used analytics in this manner?????

I think the implications at zip code level are extensive and interesting including some of the following:

1. Its very relevant for campaigns into cities with multiple zip codes. For the types of services that are ubiquitous and highly competitive, as google gains greater confidence as to the more precise locations of a user it could alter the pac to show smb's that are closer by zip not just the entire city.

After all if you are on the west side of Chicago do you really want to fight traffic to visit a dentist or an accountant on the east side of the city when there are a plethora of dentists or accountants far closer and far easier to access.

2. With the above example it could affect the PAC that we see...adjusting rankings by zip code proximity.

3. As I looked at the data I was able to see towns with multiple zips where their was significant traffic by zip rather than the bigger region around it.

That is sweet. We can design micro campaigns for local media that get exactly where there are large groups of possible buyers rather than spread it throughout a city including zip codes and areas that don't work.

It really grabbed me as a possibility to adjust the PAC by zip's. Dang...that could further complicate local seo efforts into the PAC.

Anyone have further experience with this? I was overwhelmed by how granular the data is at this point.

 

[Phil Rozek]

Sounds interesting, Dave. So people weren't actually including the ZIP codes into their queries? (According to the area in AdWords where you can drill down and look at the actual queries that triggered your ads.)

Or are you saying most of your clicks simply came from people in a specific ZIP? (Click fraud, maybe?)

 

[sergiuliano]

Here is another POV you should to consider when thinking to ZIP-CODE targeting.

IPs are not related to a certain zip code. When a user is browsing the WEB on his desktop, from Chicago for ex, without using any custom locations or zip code in the search query, is almost impossible for Google to know which zip code is the user located.

For mobile devices is even more difficult as most of the users have the GPS switched off most of the times, and without the exact Location data, by using only IP data, Google will display results based on city level only.

There were some other posts about different results based on the user distance to centroid ( not business address in this case ) - which has no sense when analyzing the things from the POV mentioned above.

 

[Dave]

Phil:

the information came from adwords. It is related to impressions and clicks. If you run adwords campaigns you can break down data by the "dimensions" button and its drop downs. Two of the drop downs deal with geographic locations.

That is where the data came from.

Its DEFINITELY NOT search phrases with zip codes. This campaign never sees many of those. never. Not over years.

Its that google has the ability to drill down deep with regard to user location.

@sergiuliano. I think google can drill down deep. With pc's at a business and with the business claiming its listing and identifying its address google could with reasonable confidence assign an exact address to those pcs. That is very precise.

I don't know if they are doing this and I'm sure they would never tell us, but its possible. Of course its more difficult to do this type of location identification with mobile. But mobile enables you to switch locations. Mobile can tell where you are located if you let it. That is very powerful. If you fly from Boston to Denver you can adjust your location and mobile will identify where you are. Then search for pizza and you'll be served Denver pizzas rather than Boston pizzas.

I was initially grabbed by the volume of adwords data for impressions and clicks that showed the most granular location with a zip code as opposed to a city or town. When I separated all the location data with zips from all the location data without zips for the data on 80,000 impressions---half had data down to the zip code level.

I thought that was impressive and a lot. And it did surprise me.

Again, we then ran that data against our own in house data for actual leads and actual sales.

Within our target city as part of a large regional campaign we found something interesting.

2 city zips had relatively high volume data for impressions. In other words a lot of searches were being made from those two zips for phrases that tied to our very thorough adwords campaigns.

We found internal data showing we had a relatively lot of leads from those 2 zips and a relatively lot of sales. I also know our competitors can similarly address folks in those zip codes.

From a business perspective, knowing that there are a lot of searches, a lot of inquiries and a lot of sales into those zips is the kind of data we want to know. We can use additional and more focused marketing into those areas to try and pick up more sales. If its successful we'll be picking up sales from what appears to be fertile ground and probably beating our competition in some cases.

BOY I LOVE THAT LAST POSSIBILITY!!!!

I also thought and hypothesized that google could deliver PAC and geographically oriented personalized search results on this basis at a more granular level than at a city level.

I don't know if they are doing this...but it seems possible.

Hey!!!! If I'm on the West Side or North Side or any side of a big city with a lot of traffic and I want pizza...do I want to fight traffic to get some slices by driving cross town??? heck no. I'd rather have the closer "good piza" and a cold one!!!!

 

[sergiuliano]

Hi Dave,

Zip code targeting will be amazing, but what I can tell you 100% after comparing tens of thousands of IP addresses for GeoRanker, there is a difference of over 20% between the IP2Location DB from RIPE or Arin DB and Google IP2Location DB .

When verifying a local business via Google Local you are verifying and address for a business ( and a phone number ), not an IP address, I can verify an address from Australia with an IP from UK as long as on Australia I have a landline to answer or an address to receive mail - so this hypothesis is wrong

The only chance to target Local Data is for the users with the Location Feature activated on their mobile phones or for the ones where the ISP IP is located in the same are as the one Google has on file - for now, this data has a very low accuracy.

Also, users with the custom location set to a certain area will receive results ( Organic or CPC ) from that exact area.

For example I am running a search query right now from 20 Km from the city center, in a residential area and I get results from the city and not close to my location.

When I try on the mobile, with location discovery off, I get the same type of results.

 

[Dave]

@sergiuliano: All of what you write may be and is probably true. You have experience and expertise there.

I reported on what I saw in a reasonably sized adwords campaign. (over 500,000 impressions) Roughly half the searches were geographically described down to the zip code level. How good is that data? I don't know. But google thinks its pretty good and is reporting on it.

I know and they know their geo breakdowns aren't perfect. But after years of running adwords campaigns I know they are pretty good and accurate when you set geo parameters as to size of market or target specific areas.

I also know that certain smallish towns have one zip code and one town name. In those cases identifying a town as a location is tantamount to identifying the zip code.

Simply the volume of data that went down to the zip code level surprised me. I wasn't aware there was SO much data provided at that level.

Then I cross referenced the impression data from adwords against our own business proprietary data with leads and sales from individual customers.

For that campaign and smb it has a regional focus with a major city. Within the city 2 zip codes stuck out with a relatively lot of impressions. ie...within that zip a relatively lot of people were searching on industry phrases. I cross checked the data against our own information. A relatively lot of leads from people in those two zips and happily a relatively high number of sales.

I love that data. Search phrases, leads, and sales all overlapping from some very distinct and identifiable neighborhoods/zips within a far bigger city. These are neighborhoods wherein our competitors can also easily attract customers.

What we are now acting on is to increase our marketing into these very identifiable neighborhoods. They seem rich to me.

I appreciate the fact that google is providing deeper data that I didn't know they had. In those cases it seems to make sense and it plays out with our own data.

I'm not saying the data is perfect or exactly how google uses it or had absolutely much veracity it has. But in our initial looks...it appears to be far richer than I had realized before....and it corresponds on the impression side with our own results.

I intend to take advantage of this new level of information.

 

[Dave]

Speaking about Micro Local Data look what Macy's is doing in two stores


​

Google is not the only entity that can track to more and micro locations. Check Ins at foursquare and yelp Identify what store, neighborhood etc you are in.

But Macy's is taking it to a new level via Apple. They'll know when you are in the store and where in the store you are located. Macy

---------- Post Merged at 02:36 PM ---------- Previous Post was at 02:26 PM ----------

Speaking about Micro Local Data look what Macy's is doing in two stores


​

Google is not the only entity that can track to more and micro locations. Check Ins at foursquare and yelp Identify what store, neighborhood etc you are in.

But Macy's is taking it to a new level via Apple. They'll know when you are in the store and where in the store you are located. http://bit.ly/1dlfOqx

 

[Phil Rozek]

Thanks, Dave. Gotcha. That wasn't clear to me. Although I've logged many an hour in AdWords, I haven't spent much time in the Dimensions are (tsk, tsk).

 

[Dave]

I was just looking at some analytics information by city. Was very struck to see further and greater levels of granular location data. It sort of relates to this extensive access of zip codes that we might not realize.

In comparing where traffic was coming from by city and various suburban towns between this year and last year there were some significant examples where areas that were defined by "larger towns" in the past are now subdivided into more specific smaller towns, possibly related to more specific zip codes, or simply further breakdowns of geo information.

Google's data and its access to accts via people signing into google + is giving them more and more granular data than they used to have.