Google Local Carousel hides non-branded searches

Google recently released the new Local Carousel, and local search experts have been scrambling to try to figure out how it’s going to affect search traffic and user behavior. Mike Ramsey studied the reactions of 10 random people, and there have been several Local Carousel heat map studies.

We’ve been running our own tests since we first got access to the test version of the Carousel. Personally, I’m not that worried about how user behavior might change or how search traffic might flow differently… because I’m blown away by the biggest flaw of the Local Carousel (that no one else seems to have noticed yet):

Because of the way that Carousel clicks work, local businesses will lose the vast majority of their non-branded search traffic.

Yep… You read that correctly. Nope, I’m not exaggerating for dramatic effect.

(Stick with me on this – it’ll be a long post, but I want to fully explain our process)

If you haven’t read my original post about the Local Carousel, I pointed out that a click on a Carousel listing doesn’t take you to the venue – it simply updates the search results to a branded search for that venue. With the new layout, users are two clicks away from a venue’s website (if they use the Carousel), and we were curious how the search queries would be reported in Google Analytics. Would you see the original non-branded query? The modified branded query? Would Google stick it to us with more (not provided) results?

I’ll admit that we don’t yet know how many users will click on the carousel, but come on – we have to assume that after people become accustomed to the carousel, it’s going to get a vast majority of the clicks. Even if we’re conservative and assume that only half of a venue’s current organic clicks will go to the carousel, that’s still going to result in a huge potential loss of data.

One of our SEO team members does some freelance work for a few restaurants in McKinney, Texas – a small town about 25 minutes north of Dallas. We decided to run our Local Carousel tests on one of the small restaurants, since the low traffic volume would make it easier to run a clean test.

We used a unique non-branded search term that hadn’t appeared in their Google Analytics data within the last 60 days. We used Internet Explorer for every query, and we were logged out in order to pass through keyword data. We monitored the results through real-time reporting in Google Analytics, then backed up our results with standard reports and Google Webmaster Tools.

Our test query was “Italian restaurants downtown McKinney” for all 10 users. Half of the users clicked the standard organic listing for the restaurant:

half clicked organic results

The other 5 users searched for “Italian restaurants downtown McKinney,” then clicked on the Local Carousel listing for the venue. That updated the search query to “Sauce on the Square McKinney,” and they clicked on the venue’s link in the organic results.

clicking the carousel, then the organic result

We had Google Analytics open while our test subjects were clicking, and the results backed up our fears…

If a user conducts a non-branded search, clicks a Carousel listing, and then clicks that venue’s organic listing, the modified branded query is what gets reported to Google Analytics.

Here’s a screen capture of our Real-Time Report:

Local Carousel live click data

Frustrating, right? We thought it might be possible to pull some extra data out of Google Webmaster Tools, but unfortunately, you still get stuck with the modified branded search. Check out the screen capture of our Webmaster Tools data in Google Analytics for the day of our test:

Local Carousel test impression data

We still lost the 5 impressions/clicks for “italian restaurants downtown mckinney” (green) from the users who clicked the venue on the Local Carousel.

If you’ve got clients who are showing up in the Local Carousel and you’re tracking non-branded search traffic for their reports, get ready for some uncomfortable conversations… but hey, at least it’s not (not provided), right?