We’re all critics, whether we Like it or not. Our Facebook posts are implicit reviews and recommendations of restaurants, apps, movies, news articles, dry cleaners, and even whole cities. Now, Facebook’s new keyword post search is unlocking the value of the opinions we share but never put on review sites like Yelp, Annie’s List, MetaCritic, or the App Store.
Of course, your Facebook search results are only as interesting as your network of friends. If you roll with a Luddite crew, or your friends just aren’t that into Facebook, you won’t get much help. But if you have a share-happy graph, you may be pleasantly surprised. After spending some time testing the new Facebook Search, I was amazed by how useful it is for digging up recommendations. Over time, thanks to artificial intelligence and natural language recognition, Facebook could get even better and categorizing and filtering the results.
Whether I searched “dentists” or “Berlin clubs”, recommendations for real-world experiences were plentiful on Facebook because that’s what most people post. Media suggestions were a bit more basic, but I could find out if my friends generally raved about a movie or not. Reviews of apps or more niche things were scarce, and web search engines like Google will remain the go-to.
I found Facebook Search to be a powerful tool for determining the sentiment of my network regarding a specific news topic. Posts about “Ferguson” were mostly rightful outrage or news articles condemning the Grand Jury decision, accurately portraying my liberal community. The main problem, though, is that you can’t display results in reverse chronological order like on Twitter, and instead get a News Feed-style personally set of results from over the past few months and years, rather than today’s pulse.
The new Facebook Search is still rolling out to all US English web and iOS users. For more info on how it works and why it’s important, check out my full launch story on the new Facebook Search.
Now, Here’s my review of the experience for a bunch of common searches. Note, as a student who got Facebook in 2004, San Francisco resident, and tech worker, my network is unusually active on Facebook, so I get unusually rich results. Others with less active networks may have a lot less luck, so let’s consider this a look at how Facebook Search works under ideal conditions.
Reviewing The Results
Local Businesses: “Dentist”
Why would anyone post about their dentist? If they really love the care they get, or they themselves were looking for recommendations. Facebook Search did a stunningly good job. It found me several friends in my area praising their dentist, and even more who’d asked for suggestions and received huge reels of comments of the best ones from their friends. All I had to do was dig into those comments to find someone who could drill my teeth without deriving pleasure from my pain.
Specific Local Businesses: “Fun bar”
I wanted to see if Facebook Search could find me a high-quality local watering hole. There was a fair amount of noise in the results from people using “fun” and “bar” without actually rendering an opinion of a pub, but I did get a few solid Instagrams of bars tagged with #fun. Looking for places with a specific adjective may be a bit too narrow with Facebook search right now. It could get much better if Facebook understood synonyms so it could surface posts with “great”, “awesome”, “best”, or “pub”.
Travel Recommendations: “Berlin”/”Berlin Recommendations”
People on social media commonly ask what they should see when the visit a foreign city. Of course, there are plenty of other reasons to post about something related to a metropolis, so I had to sift recommendations for clubs and restaurants out of random news articles and other noise in my initial search for “Berlin”. But once I searched for “Berlin Recommendations”, I saw post after post of crowdsourced travel guides as friends asked for help and got reams of answers in the comments. “Recommendations” is probably the most powerful word you could enter into Facebook Search.
Place-Specific Photo Search: “Dolores Park Photos”
While Facebook had photo search in Graph Search before, you could only see ones tagged in a specific place. Now you can bring up any photo with a caption mentioning a place, and it works nicely. A query for pics from my favorite San Francisco relaxation spot Dolores Park produced gobs of pretty Instagrams and Facebook photos of smiling friends and summer sunsets.
Movie Reviews: “Interstellar”
I don’t care if the professional film critics liked Christopher Nolan’s new movie. I want to know what my friends think. There were plenty of low-utility “I’m going to see Interstellar!” posts, but I saw enough “Wow, I loved it” and “can’t stop thinking about Interstellar” posts to know it was a must-see. I was also able to peek in on the various debates between friends about whether the story was actually plausible. Those might be something I’d actively avoid in the feed until I saw the movie, but then would be excited to search for after.
App Reviews: “Best App”
Here’s where Facebook Search failed, or at least I couldn’t figure out how to make it work. Though I know friends occasionally post about their most beloved apps, no modifier like “great”, “loved”, “favorite” or “best” returned me a high-signal set of results. Instead, I mostly got random news articles, which makes me think a way to purposefully filter out link shares or news link shares from results would be a great feature.
World News/Trending Topic: “Ferguson”
Twitter is the king of real-time chatter, and its search makes it easy to see what the whole world is saying about a topic. Facebook Search doesn’t show Public posts by strangers, so it won’t give you the pulse of the planet, but instead it can show you what your real friends think.
A search for “Ferguson” correctly showed that most of my friends were angry about the Grand Jury’s decision not to indict the Ferguson police officer who killed Michael Brown. I saw posts as recent as 45 minutes old alongside a slew of links about racism and the police state. Rather than a pulse, it was more like a high-level overview of the sentiment of my network. What I really wanted, though, was the ability to display the posts in reverse chronological order like on Twitter so I could see the most recent posts from friends.
Privacy Search: “Josh Constine Drunk”
Facebook keyword search doubles as a privacy check if you turn it on yourself. Rather than meticulously browsing through your old posts to make sure you didn’t say anything too stupid when you were younger, now you can just search your name and some bad words. I was happy to find I’d never posted about being drunk and neither had my bosses. My parents were actually the tipsy ones:
I did find that I curse like a sailor, though. If I had a more buttoned-down job, I might want to clean those up. Luckily, swearing is regarded as an asset by my current employer.
Strong v1, Big Opportunity For Categorization
Overall, Facebook Search does a strong job of dredging up posts washed down the News Feed’s content river and turning them into convenient recommendations. If I want a more trusted review from someone I know, Facebook Search could be more helpful than Yelp or Annie’s List. The feature proves that a lot of the most useful reviews never end up on review sites. They’re just things we share casually on social media. Answers we previously had to go to Yelp or Google for can now be found on Facebook, which should give those companies a reason to worry.
Facebook Search’s biggest downfall is that it doesn’t understand why we’re searching for something. If I’m looking for a random memory or something about a specific friend, the comprehensive nature of the results it shows now work fine. But if I’m looking for more actionable recommendations, link shares and photos are often just noise getting in the way. Hopefully one day Facebook will be able to recognize the different types of value that posts can offer, and let us filter by our intent.
My other big requests are for automatic search of keyword synonyms, reverse chronological sorting for real-time chatter, and more advanced search options. Considering this is v1 of post search, their absence is completely understandable, and the fact that it works on a basic level is impressive considering Facebook is trying to index 1 trillion pieces of content.
Today Facebook proved just how much information it holds, and how valuable it all can be. It must make Google mad that it can’t crawl this walled garden. Looking forward, it’s clear Facebook has a ton to gain if it can better understand and categorize semantic content and sentiment.
Mark Zuckerberg considers “Understanding the world” through Facebook’s new artificial intelligence lab one of Facebook’s “three high-level goals over the next ten years”. Earlier this year he said the AI Lab “is trying to build a unified model of how every person [inside] the world is connected to each other. In the near term, our efforts here are in search and News Feed, and will help your network surface more useful information to you.”
Now we know what he was talking about.