Facebook's Graph Search: The Lowdown


Facebook last week announced a new feature: a social search engine called Graph Search, which is being trialled through a limited release for English (US) users.

Essentially, Graph Search intends to make it easier for users to locate information within the social network. Instead of keywords, users will input phrases such as “restaurants my friends like in New York” or “photos of me from June”, receiving a set of personalised results that dynamically change as you type. Currently, the feature includes people, places, photos and interests, but Facebook says it intends to add posts/Open Graph actions: Zuckerberg has stated that this is intended to be something that is developed over a number of years. The video below highlights Facebook’s motivations behind the launch.


As with any new feature, there are many questions to be asked about how this will affect marketers. As We Are Social’s Global Managing Director, Robin Grant told Adweek:

Its immediate impact will be very small. It seems that this is a product that has been built without any reference to user need. At the moment it offers little real utility, and requires a massive change in behavior for people to start using it. As a result it’s likely to flop initially, just like Wolfram Alpha did on launch.

However, there may well be potential for the feature. Robin also told Econsultancy:

If Facebook evolves Graph Search beyond the current limits of ‘People, Photos, Places, and Interests’ and in turn users warm to it, it could succeed and therefore become relevant to brands.

This would mean that Facebook’s sponsored result ads would become more relevant for advertisers, and ultimately it would mean that raw fan numbers would be much more important to brands (or more accurately, the number of fans they have that match the profile of their customer base), as to appear in Graph Search results a brand will need to have a friend of the searcher as a fan.

It would also make it essential for retailers, or any business with physical locations, to maintain Facebook place pages for each of their branches.

So a great deal of the issue rests on how useful Graph Search really can be for Facebook’s audience. TechCrunch highlights just what you can search for; for example, if you’re looking for a hot date one evening, but are fussy about films and don’t want to travel (what a catch!) you can search “singles who live near me and like Indiana Jones”. You can also get local recommendations by requesting the likes of “restaurants in Rome liked by people from Rome”.

A great article by Wired explains the behind-the-scenes development of the tool, referring to it as the platform’s ‘weapon’ to challenge Google. The piece discusses everything from the motivations behind the tool to the attitudes of Zuckerberg and others about its release, painting a picture of a feature that could potentially be hugely powerful. It finishes with a quotation from Zuckerberg who, when asked if he thinks the number of search queries on Facebook could match Google’s, stated:

Hopefully, over time. But we’re building this because we think it’ll be something that people want to use. There’s a lot of things we haven’t built yet. But I think even in the beginning, the experience is going to be, ‘Wow, this is amazing.

So Facebook’s ambitions are large and its approach to search does produce some interesting results. However, there are also a number of problems. A largely critical review by CNN mentions how, in order for Facebook to be a successful recommendation tool, a shift in certain behaviours is required. For example, people are not using the ‘like’ button as often as the social network might hope, meaning that when you search for “restaurants my friends like in Chicago”, you are missing out on all those restaurants that people have not been bothered to actively “like” on Facebook. If Facebook can attain more accurate ways of accessing information, such as obtaining data from posts in which people state their enjoyment of a particular restaurant, the tool could be significantly more useful.

Another challenge the tool faces is the lack of activity on the accounts of many users. The graph below displays the percentage of monthly active users behaving in certain ways.

In fact, Facebook may need many more people to partake in such activities in order for Graph Search to work. Unfortunately for the social network, this is not the way in which the trend is currently heading, as it is largely only in developing non-English language markets that these are on the up.

However, there is a school of thought stating that Graph Search itself may help to change behaviours in the desired manner. The theory behind this is that, in the current environment, ‘sharing’ anything on Facebook is a selfish activity that does not actually benefit those with whom you share and, as such, is not in line with real-world sharing. However, with Graph Search, this could change, as liking or sharing certain content helps create a framework whereby everyone benefits from better and more accurate search results. Yet there is a response to this, which wonders if the inverse may also occur: could users be less inclined to like certain pages if it is seen as a real display of their opinions? Certainly, there are a number of different factors likely to impact how users and marketers are affected by the feature and it will be of real interest to see how these develop.

Finally, a few thoughts about the problems of privacy associated with Graph Search. Whilst there has been no change to what information is private or public, it is now much easier to locate anything that was already available. In fact, Gizmodo have highlighted a number of occasions on which people may wish to be a bit more careful with what they ‘like’. This means it might be sensible to check any pages you currently “like” – if there’s anything in there that you don’t want people to see, make sure it isn’t available to them. In fact, Facebook have now made it impossible for users to opt out of search results. Whilst this makes sense for the platform, considering how important it is for them to have access to as much data as possible, it is a reason for people to be even more careful when deciding what information they make publicly available.