With Tinder, first impressions are often made in 1 or 2 seconds, and even that happens between two other sets of first impressions. Singles are in competition with each other, and the name of the game is to convince the user to slide their finger right with as much confidence as possible.
As a digital media analyst, I can’t help but draw parallels between Tinder and the most popular digital ad platforms.
After all, they work on the same principle: You have show the product (yourself) in the best possible light, while managing an online campaign where you’re targeting audiences based on geolocation (km), demographics (age, sex), and where you share your preferences and specifications, for example height, via a profile.
That makes the first photo in your profile equivalent to the ad copy or creative that will bring about future conversions, and with this type of prospecting campaign, it’s essential to get the user’s attention from the very first second. Your first profile picture is your initial point of contact with future suitors, and plays a major role in the success of your love life on this fiery app.
Last January, during an internal Hack-o-thon, Mike Hall, Tinder’s machine learning lead, developed a concept to help singles on the platform answer just this question. Nine months later, Tinder launched Smart Photos, an option that you can activate under your profile pictures that will automatically show the picture most likely to garner a right swipe, first.
How does it work? Just like with a real online marketing campaign, it is important to constantly test your ads using tried and true A/B testing techniques. Some ads speak to users more than others, and it’s not always the one you prefer that will turn out to be the strongest performer. The same thing applies to your Tinder photos. Although picking your favourite photo of yourself seems logical, more often than not, we are very poorly positioned to determine which photo sells us the best, according to Dr. Jess Carbino, Tinder’s resident sociologist. Singles on the platform choose their first photos for all sorts of personal reasons, but rarely take into consideration the impact of this choice on other users. Carbino asserts that a profile will tend to get more left swipes if in the first photo:
To address this problem, Hall and his team needed to develop an algorithm that would rotate each of your photos into first position every time your profile was viewed, and thereby determine which photo gets the most right swipes. By calculating the Swipe Right Rate (SRR) of each photo, in other words, how many of its total swipes were right swipes, the algorithm is able to determine which photo to display first in order to optimize your matches as much as possible. According to the company, some users saw their matches go up by as much as 12% after activating the Smart Photos option.
The SRR is therefore equivalent to a click through rate (CTR), the most important metric when it comes to getting a good Quality Score for a Google AdWords campaign. The higher the Quality Score, the higher your chances of appearing at the top of search results, which gives you greater visibility, and in turn, more sales or conversions (depending on your objectives).
And this even goes farther, because unlike a search engine, Tinder is a bidirectional platform in the sense that, to get a match, both parties need to have swiped right. Which means that the optimal photo varies by impression, depending on whether your suitor(s) are more interested in your corporate side, with your photo in a business suit, or your adventurous side, with that beach photo in the tropics.
The Swipe Right Rate is an important metric for the calculation of your personal Tinder score, called the Elo score, a term borrowed from chess and used to rank the best players in the world. Each app user has their own Elo score, which is used to rank users and present you with people with the same degree of “desirability” as you. This score isn’t shared with users, probably to avoid triggering profound depression, but it’s still possible to improve your Elo score using a few best practices. An algorithm doesn’t pass judgement, it simply reports the results of your actions using complex mathematical calculations that factor in a number of considerations, including your photos and your SRR, explains Sean Rad, the company’s CEO.
Jonathan Badeen, Senior Vice-President of Product for Tinder makes a comparison to the popular game Warcraft. He explained to Fast Compagny que lorsqu’il jouait avec un joueur ayant un score très élevé, il finissait par gagner plus de points que s’il jouait avec un joueur ayant un score inférithat when he took on a player with a very high score, he ended up gaining a lot more points than when he played with someone with a lower score to his own. He figures that this is a good system for quickly associating and classifying users based on the people they match with. In other words, the more you match with people with higher Elo scores that yours, the higher your own Elo score will be. The reverse also holds true. There’s no question the Tinder algorithm is more nuanced than that, but other people’s opinion of your profile definitely plays a big role in your score. Tinder data analyst Chris Dumler, sees the platform as a “vast voting system” that indicates which people are more desirable than others at different levels of the desirability scale.
So, to succeed on Tinder, as well as with your digital campaigns, you have to adopt the right behaviours. Even though usage behaviour differs between the sexes, certain practices like right swiping everyone, very popular among men, will assuredly penalize users in terms of the quality of profiles they’ll be presented with, their reach, and as a result, the number of matches they get.
So, how is it possible to always show your best photo first, when which one that is depends on so many different factors? Mike Hall approached this optimization problem using the Multi-armed Bandit1: It’s a probability puzzle, where a gambler looking at a row of slot machines has to decide which machines to play, how many times to play each machine and in which order to play them in order to maximize winnings in a sequence of play.
To respond to the Multi-armed Bandit problem, Hall and his team used the Epsilon Greedy algorithm which allowed them to discover which of your photos generates the most right swipes without wasting impressions on your less effective photos. Your photos enter into an exploration phase where each photo is tested, before moving into operational mode where only your best photos are shown first. The team also thought to make a few adjustments to the algorithm, so now, when you add or remove a photo, it’s taken into consideration.
Whatever you think about the app, Tinder’s success is definitely related to its ease of use, with its simplistic design of ranking cards. However, behind this intuitive experience hide major technological challenges. By introducing Smart Photos, Tinder is demonstrating that they’ve taken one more step towards being a more efficient platform.
There’s no formula you can follow that will guarantee you more matches. But still, if you use honest, recent photos, take the time to only “like” the people you’re really interested in, and use the application often, you’ll be giving yourself your best chance of coming across the person you’re looking for.