Facebook Graph Search, Search Quality, and the Meaning of Like

Facebook’s new Graph Search feature promises to revolutionize our searches by tying together the social graph of our friends and Liked pages.

At first glance, the ability of the new search to bring together different bits of information as an answer to a natural language query is almost awe inspiring.

Ask for “people in my company under the age of 25 who like skiing,” and Graph Search will give you exactly that, at least insofar as your co-workers have been honest about their age and favorite sports activities.

But what about a search with more commercial intent?

Let’s say I’m looking for the Mexican restaurant nearby me that my friends think is the best. So I search on Facebook for “mexican restaurant nearby that my friends like.” And Facebook gives me a list of the most popular south-of-the-border eateries, according to my Durham NC area friends. Perfect, right!

You Keep Using That Word. I Do Not Think It Means What You Think It Means

A US mailbox designed as the R2D2 robot from S...

“These are not the search results you’re looking for.” (Photo credit: Wikipedia)

Not so fast. Those “likes” by my friends, remember are Facebook Likes. And Like on Facebook may not always mean what you and I mean by “like” in the real world.

In the real world, if my friend Matt says, “I really like Sol Azteca in Morrisville.” It’s reasonable for me to assume that Matt means that he “likes” Sol Azteca because of their great food, and maybe other relevant factor like their atmosphere and service.

But what if Matt has “liked” Sol Azteca on Facebook? Does that Like necessarily mean the same thing?

Very often, it probably doesn’t. “Like” on Facebook can mean all sorts of things, and indicate all kinds of intent.


For example, it is not uncommon for people to “like” a Facebook brand in order to participate in a contest or offer with the brand. In the real world, if the only reason Matt liked Sol Azteca was they let him have an entry in their contest, he’d probably tell me that. He’d say something  similar to, “I like that I might win $50 from Sol Azteca! Their food? Eh, not so great.”

But I have no way of knowing why Matt liked Sol Azteca on Facebook. So if they show up in response to my query for “mexican restaurants nearby that my friends like,” what I’m getting as a result may not be really recommendations of particular restaurants, at least not in the way we think of recommendations in real life.

The Coming Like Inflation on Facebook

Not only are recommendations based on Facebook Likes of dubious value, Graph Search may be about to cause them to get even worse, in an inflationary spiral.

Within an hour of the announcement of Graph Search, I was already seeing blog posts with tips for Graph Search search optimization. Inevitably one of the recommendations is to start getting as many Likes for your brand page as you can. It isn’t hard to see what will happen here. There has always been a race to run up the number of Likes for Pages. Now it will turn into an inflationary spiral that will make pre-WW II Germany look like a Golden Age.

Expect to see even more contests and coupon offers to get Likes. And expect there to be a sudden bulge in the pocketbooks of the “Likes for sale” businesses online.

But here’s where Facebook might win anyway: when people are getting crappy search results, they usually don’t know they are getting crappy search results. So if the average Facebook user never stops to wonder about the real meaning of her friend’s “recommendations,” she’ll continue to search for them and follow them anyway.

Related: See “Facebook Is Recycling Your Likes to Promote Stories You’ve Never Seen to All Your Friends


Like this post? Please upvote it at Inbound.org!

Enhanced by Zemanta
Facebook Graph Search, Search Quality, and the Meaning of Like by

Mark is Director of Digital Outreach for Virante Inc. Mark helps businesses build strategies to increase brand influence and attract natural links and social signals. He has a special reputation as an expert on Google+ and Google Authorship. A former teacher, Mark has worked directly in Internet marketing since 2005, but has been involved in social media and online community formation since the mid 1990s. When not helping Virante clients improve their online presence, Mark participates in competitive storytelling, plays with a Dixieland street band, and (surprise) spends more time on the web.


  1. Well written post with a lot of good points. The only way they are going to overcome all this is to somehow beat Four Square but no one has been able to as of yet. Four Square tips are usually right on. I am less familiar with Yelp but there is even more information there because it is not just tips but reviews. I think Google learned the hard way that information is not king. The Right information is king. Their search function is getting much better now. As Google+ grows it will only get better as it morphs into social search.

    The thing that may allow Facebook to win in the end if if they use status updates. That is people being real. But people are going to scream bloody murder when they do it for privacy reasons.

  2. I “Like” your article and analysis Mark! ;-)

  3. Sounds as though the perception of Graph Search will outweigh the reality. Thanks Mark.

  4. Hi Mark, off topic but probably worth mentioning – I noticed your ‘upvote on inbound.org’ link and visited the site. They seriously need an about page, or faq, or features page. Currently it’s not a useful link to visit for the uninitiated. :)

  5. I read a bit about the real meaning and use of the Facebook “Like” and other “likes” in social networks. I can say that some of my “likes” were only to follow the brands for news because I really do like their content. Others were to support them even if I weren’t familiar with their content, products or services. I can attest to seeing pages having massive amounts of “Likes” but hardly any interactions or “likes” for their individual posts.

  6. This is where the issues will surface because as you say – Graph Search looks awesome and awe inspiring but the reality is that not only “likes” but also friends are dubious on FB and friends of friends because many originally utilised it to network and do business so a friend is not necessarily someone you really are friends with. Likes are often “just because” and information that is shared is often not reality or honest either.

    I think it is a great idea but is not the world domination search solution that people are suddenly thinking is coming – I have seen many comments about it and some people saying that Facebook is finally going to kick Google where it hurts – Hmmm – for goodness sake Google is not only about search it is about business! :)

  7. When I heard the announcement I had two immediate thoughts: FGSO (Facebook Graph Search Optimization) will be a new trend, and also that the value of the like just went up — although your inflation argument is certainly valid. Great job pointing out that a “like” on Facebook doesn’t necessary have the same endearing emotions as a real life like. Very important distinction.

  8. Mark I think the system is ripe to be gamed. Facebook already has issue with dead people liking things, throw in search as an incentive and I suspect facebook’s user database will grow exponentially. What will stop savvy businesses from liking everything in sight to show up higher in search? What prevents like brokers from sending 20,000 likes your way to rank higher?

  9. You bring up some very good points I haven’t thought about yet. A like isn’t a recommendation.. Now you got me thinking.. But this could mean people will be stricter with what pages they like knowing that their friends will be basing decisions off of it. Which further leads to brands running better targeted ads.

  10. Absolutely right about it not necessarily meaning they like the restaurant or whatever that comes up in a result, but my point would be if I wanted to know which of my friends liked a certain place and valued their opinion, I would ask them direct not a search engine.

  11. Excellent food for thought. Perhaps some value in locating photos etc. but the value of a “like” is at best dubious. Nice work Mark


  1. [...] Facebook Graph Search, Search Quality, and the Meaning of Like [...]

  2. [...] Facebook Graph Search, Search Quality, and the Meaning of Like – Virante Orange Juice Should you trust that Facebook’s new Graph Search will give you reliable recommendations? Probably not. Here’s why. [...]

  3. [...] 1. “Like” gamification – Mark Traphagen wrote in his blog post:Facebook Graph Search, Search Quality, and the Meaning of Like: [...]

  4. [...] Facebook Graph Search, Search Quality, and the Meaning of Like [...]

  5. [...] all the time in order to run up their Like counts. Facebook has incentivized what I call “Like inflation” as they have tightened up EdgeRank, making it more difficult for brands to get their content [...]

  6. [...] pages all the time in order to run up their Like counts. Facebook has incentivized what I call “Like inflation” as they have tightened up EdgeRank, making it more difficult for brands to get their content [...]

  7. [...] was a copy of SnapChat and that had about 2 minutes of fame. Facebook then released Graph Search. Industry experts pointed out how terrible this product is for everyone involved. Then came the new Facebook profile [...]

  8. […] all the time in order to run up their Like counts. Facebook has incentivized what I call “Like inflation” as they have tightened up EdgeRank, making it more difficult for brands to get their content […]

Leave a Reply