Yesterday at Local University Advanced Joel Headley, one of the main people behind the development of Google Maps answered a number of very intriguing questions related to Google+ Local. But before I start discussing some of those Q&A’s I should note that I didn’t participate personally in the event, so everything I know is sourced from the social networks. One of the topics that was of greatest interest to me was the Google reviews one. It all started with the following tweet:
Joel Headley is talking about Google’s clean up of the review space. They deleted any review that they found had a duplicate. #localu
— Mike Ramsey (@niftymarketing) October 1, 2012
I was curious to get a clearer idea of which reviews have been targeted and that is why I asked:
@niftymarketing Duplicate in Google’s database, or anywhere on the web???
— Nyagoslav Zhekov (@Nyagoslav) October 1, 2012
Mike was kind enough to ask Joel for clarification and here is what turned out:
if google finds content on the web that is the same as a google review it will be removed.Dont add as a site testimonial. #localu
— Mike Ramsey (@niftymarketing) October 1, 2012
And a bit more details:
@nyagoslav I had to ask the question to which was vaguly answered. He couldn’t say for sure but alluded to the fact that content anywhere.
— Mike Ramsey (@niftymarketing) October 1, 2012
As I have learnt not to trust Google for anything (they even show me as part of a company I left more than half a year ago for crying out loud), I decided to check this myself. Obviously, the biggest problem in such an experiment would be to find an appropriate sample. It would have to be dated, and it would have to be blatantly fake. How do I find these? In fact, it is very easy. I look for businesses that are very unlikely to be reviewed many times, but in fact do have a decent number of reviews.
I first decided to “target” the plumbing industry. I really “love” exact match domains such as this one, as well as keyword-stuffed business names, so the reviews associated with this business seemed like a promising laboratory rat. I used the following review that dates back a year ago (i.e. Google had plenty of time to figure out if this review is fake or not):
“We had 4 different plumbing companies come to our house before we committed to letting Dallas Plumbing do our work. They were, by far, the most professional and trustworthy! We have given out their cards to all of our family and friends. I would highly recomend them to anyone.”
How unsurprised I was to discover that there was a very similar review left for another plumbing company on another site (Yahoo! Local) that very much reminds of the above mentioned one (note: it has been deleted by Yahoo – talk about efficiency of combating fake reviews, but it still does exist in Google’s index, i.e. Google thinks it still exists):
“We had 2 different plumbing companies come to our house before we committed to letting YB Plumbing Dallas do our work. They were, by far, the most professional and trustworthy! We have given out their cards to all of our family and friends. I would highly recomend YB Plumbing Dallas to anyone.”
My second sample was a company in the moving industry. I used the following review, written 10 months ago:
“Awesome!!! Called before arriving, on time! Moved everything out of the old house and into the new house in just under 2 hours. They worked hard, were very careful about corners, walls and banisters. Polite, respectful and I can’t reccomend them more!”
And I found a very close match on Judysbook for another moving company. Here it is:
“Called before arriving, on time! Moved everything out of the old house and into the new house in just under 4 hours. They worked hard, were very careful about corners, walls and banisters. Polite, respectful and I can’t reccomend them more!”
Note: what I really “love” the most in both of these examples are the matching misspellings of “recommend”.
My third, and last, example was a carpet cleaning company, and more specifically the following review (from 10 months ago):
“After finding that my dog urinated all over my couch, first of all I put him outside, and second of all, I called Los Angeles Carpet Cleaning. They gave me a really affordable price for the cleaning so I hired them and they were able to extract the urine and few other stains I had and the smell is gone.”
There is an exact match review for the same business (probably) on Insiderpages.
What is the conclusion?
Google claims to be fighting spammy reviews (check here how long it might take them to get this done in some obvious cases) and they are hopefully getting smarter at doing so, using different signals and thus making the anti-spam filter more sophisticated. Unfortunately, they do not seem to be using the signal of finding exact (or close to exact) match reviews across the web. It should be noted that these should be reviews on business directory and/or review websites, but not simply found anywhere on the web, because in that case it would take, as mentioned here, just a dummy site where all reviews for all competitors could be stuffed, so that Google could match them as duplicates and delete them. There are two very distinctive features of the spammers, which Google could leverage in their favor: they are working in bulks, and they are lazy. This means that Google could easily track down whole networks of fake reviews by finding patterns (such as the one I discussed in this post).
Update (2 October 2012, 9:15 AM EST): Joel gave the following clarification on what he had said during the SMX session:
“Communication by tweet always lacks context. Specifically, I said duplication of content was against our terms and doing so could result in removal of said content from our reviews system. I didn’t make specific claims about what had been done in the past, but rather was discussing the policy of duplicating/using the exact same text when leaving reviews.“