Revisiting the Local Queries Question

One question making the rounds — without a good answer — is “what percentage of mobile search is local?” Let’s put aside app usage, which can be a form of search, for the time being.

Google previously said that on the PC “20% of searches on Google are related to location.” Because of the challenges in defining a local search — is “flat panel TV” a local search because 95%+ are bought in stores? — this 20% figure is probably conservative.

On the mobile side of the house, two Google employees have been cited (Paul Feng and Diana Pouliot) as sources for the metric that roughly one third (33%) of mobile search have a local intent. I asked Google to unpack this figure for me and I was told by a Google PR person that this number came from BIA/Kelsey Group.

I’m not confident that’s actually the case, but if it is it’s really weird that Google would repeat a piece of third party data when it has actual mobile query data. BIA’s data are survey based and self-reported. Surveys are good directional indicators but not as accurate as actual behavioral data. 

Previously Microsoft presented data (from 2008) that argued if “local intent” is the value, at least 50% of mobile searches featured local intent:

Picture 218

Source: Microsoft (3/10, based on 2008 data)

This morning I was playing around with Google’s keyword traffic estimator tool, which allows you to look up mobile search volumes for specific queries. I looked at a number of categories and query terms: doctors, restaurants, movies and a few others. One caveat here is that I selected terms that are predominantly local terms. Product-related terms are far less local (in Google’s attribution) though most products are going to be bought in stores. 

 Picture 33

What I discovered was that the portion of queries attributed to “local” on a monthly basis varied, as one might expect, by query and category. The range was 29% at the low end to 67% at the high end for the terms I selected. The average of this small sample came out at around 43%. 

Someone with API access could probably do this calculation across a broad range of queries and get a better “average” than I got off my small non-scientific sample. But the point here is that there’s a lot of LBS action in the query mix on mobile devices. And with specific local-intent queries there’s going to be an accompanying “need it now” interest in many or most cases as well.

So let’s not call it “one third.” Instead, let’s say there’s a “local intent” mobile query range from about 20% in ambiguous categories to as much as 70% (or more: e.g., restaurants) in some categories where fulfillment is offline.