How Long Has Marissa M Been Testing Hotspot?

Gary Danko, is not the sort of restaurant that I would typically eat at, not that I wouldn’t like to try it one day. Dinners run towards $200/ head. While, it is unlikely I ever will eat there, the many, many reviews provide for some interesting reading.

Amrita – Aug 12, 2010
The seared foie gras is to die for. also, don’t go with more than three courses unless you are *really* starving

I presume that the word starving is not meant literally in this context.

But it also appears to be a favorite haunt for Googlers, particularly those on the Maps team. This includes both long time veterans and some recent arrivals. In the body of reviews there are some interesting narratives and insights (both real and imagined).

Carter doesn’t appear all that cracked up about the place giving it but a one star Hotpot rating. Lior Ron, a senior product manager on the Maps team on the other hand ate there on September 18th and gave it 5 stars.  Marissa M also seemed to really enjoy it here, giving it a 5 star rating.

Marissa M rates a restuarantHer Hotpot rating, a feature not available to mere mortals until this month, was created on January 21st, 2010. She shows two periods of intense Hotpot ratings prior to introduction ( August 13th and January 21st ). Hmm… perhaps she was thinking of working in Local long before her job description rewrite in October?

Theirs weren’t the only “early” ratings to show. Gary Danko also seems to not just be a hangout for Googlers but to be a testing grounds for Hotpot with 250 of the 290 Google reviews appearing to be ratings, many of which occurred long before Hotpot was released.

The reviews combined with the many ratings also provide an inkling into the relative algorithmic importance of reviews vs ratings. Take a look at the general ordering of the different review types on this page and note that long form reviews are shown first in date order (with the exception of the most recent), followed by Hotpot ratings that include brief reviews again in date order, followed by ratings with no comments. It would appear that if you want your review to really count, take some time and add some detail!

So what have we learned with this guided tour of Gay Danko’s reviews? Carter and Marissa M seem to have dramatically different tastes in restaurants. Googlers are obviously spending way too much on eating out. They typically are NOT eating at restaurants that serve hotpot.

The key take away for me is….

Upon entering Gary Danko’s never shout loudly with an Android glued to your ear: I have a call here for a Google Employee From Sergey! You might get hurt.

Please consider leaving a comment as your input will help me (& everyone else) better understand and learn about local.
How Long Has Marissa M Been Testing Hotspot? by

15 thoughts on “How Long Has Marissa M Been Testing Hotspot?”

  1. @Will

    Its the “pleasure” of living in an “open” world so long promoted by Google.

    Hey next time we are in SF, you can take me out! 🙂

  2. Mike: Wonder if you can find some early test spots in NYC. There are a lot of Googlers there: of course its a big company and the NYC Googlers might not have been in on the early Hotpot testing.

    I’d love to see what these folks w/ so much time to eat out have to say abt deli’s and places w/ bialys.

  3. Interesting find. I’m wary of the reviews vs ratings. It’s good that Google seems to value reviews over ratings by listing reviews first, however ratings still play a huge part in a SMB listing in the SERPs by being displayed in the overall star rating of the SMB. Say I’m looking for a service. I do a Google search and the first listing in the SERPs has 2 stars, followed by other listings with 3.5-5 stars. I’m inclined to skip over the first 2-star listing without digging much deeper.

    In my small circle I’m already seeing businesses’ star ratings taking a hit from their competitors, who have quickly discovered ways to game the system.

  4. Mike, I don’t know how you consistently find great stuff but it is truly amazing!

    Looking at the list of ratings, I find myself wondering at the utility of a bunch of 3/4 star ratings with no comments attached. Fast forward years from now and assume that Hotpot is a success…are the ratings without comments signal or noise?

  5. @Earl

    I think its your turn for research. 🙂

    Yes it does have impact and yes it will be gamed and as you point out, the visual impact is likely to be the biggest fallout/outcome

    Whether it becomes noise or not, depends on whether it is used for recommendations or ranking. The liklihood is that if it is just used for ranking then it will be noise and it won’t be too many years either…. at which point they will need to write a new algo to deprecate their value.

    Oh, and I almost forgot… you wanted to know how I found so much interesting stuff…. Its because I spend way too much time looking at Places! I was trying to find examples of business responses to bad reviews and I cam across this… the names literally jumped off the page at me 🙂

  6. I saw ratings without reviews few months ago on some hotels in Las Vegas, NV. Thought it was a bug. To be honest every day I see things that can be considered as bugs. Mike, you need tip form here so we can report such things 😉

  7. @Plamen
    You are right, I do. If I can find the time for a redesign it will be there.

    But as you know, I love tips and tidbits. You have been kind enough to email me a few ( mike at yourself…. this job of watching Places is much, much easier and a lot more fun when there are many hands on the task.

    Well, next time you fly me to the Googleplex, perhaps we can eat there on the expense account. 🙂 I am sure that it is good. You are not alone, 20 others noted it as best ever. Just promise me that you won’t proclaim that the foie gras is “to die for”. What’s Carter’s story?

    Sorry about the continual errors on the name. I spelled it correctly 8 out of 9 times… lets see that’s a 4.5 star rating…

    I find it a very difficult name to proof read. Perhaps Google needs to create a self correcting plug in for WordPress. 🙂

  8. Great find, Mike.

    More than 60 ratings (without reviews) were created on Nov. 16 and another 28 or so on Nov. 17. That’s an amazingly high level of activity for one establishment.

    It appears the raters on those days also rated A LOT of other businesses (mostly restaurants) on the 16th, too… and most don’t look to have rated any other businesses since that day.

    Was there a push at G to get employees to use Hotpot when it was announced? Probably so. Will the Googlers (and others) continue to use it in their daily dealings with local businesses? Only then will Hotpot rival a service like Yelp.

    BTW, I did get a little chuckle when seeing one of Carter’s maps edits marked ‘denied’. Guess it even happens to the best of us. 😉

  9. @Cathy
    Yes, this restaurant, along with a few other SF establishments, were clearly used for testing. I wonder how many of the Google raters actually ate there…. obviously at least one did (Lior Ron)…

    Hotpot, and its features, do not really need to exist in a standalone way to be successful… when paired with reviews the ratings (whether from a quickie star on a mobile device or a fuller review in Hotpot itself) will create a lot of additional signals that will be valuable to Google.

    I think it is necessary to evaluate the whole of the Google review system when comparing it to Yelp. Yelp may always be the goto site in urban foodie markets where people talk about foie gras as the end all be all… but the world is a big place with lots of non restaurant establishments and businesses not in SF or NYC.

    I wonder whether Google’s backfill (aggregate), fill (review) and over fill (ratings) might not be the most successful over time in industry and locales that by their nature are going to be generally lightly reviewed. And in those industries and locales that are heavily reviewed Google shows the scraped data… in the end they may not care as much about the content of reviews as the quantity and sentiment which they get regardless.

  10. More strange reviews:

    Bernardo ? – Dec 2, 2010
    A bit to pricie for what it is, but nice.
    Liked: Food, Atmosphere, Value, Portions, Presentation

    As far as I know you can only rate Food, Service, Atmosphere and Value from Hotpot.

  11. @Plamen

    good find. I wonder if they are testing a more dynamic sentiment analysis and generating options based on existing sentiments… I don’t have an Android to test if it is there or not… certainly not on the desktop.

  12. to dispel some more of the mystery:
    @Cathy – these were our launch days, so that’s the obvious spike post launch
    @Plamen – we’re offering people to rate based on the popular aspects of that place, for Gary Danko its the one you see in the review, for other places it will be different, and the default for the restaurant category is what you’ve mentioned

Leave a Reply to Will Scott Cancel reply

Your email address will not be published. Required fields are marked *

Comments links could be nofollow free.