We use social software to build and maintain relationships. Examples of social software include Customer Relationship Management (CRM) tools, professional forums, dating websites, Facebook and Pinterest.
Social software features are judged on speed and convenience. You get a lot of users quickly when your software makes things so fast and easy that it brings the power to get something done to new groups. In this article, I’m looking at the evolution of social software, and the factors that drive its improvement. How is each generation of social software better than the last? How can we find each other more quickly and easily?
Social software helps people find each other. It helps them build and maintain a relationship that wouldn’t have happened otherwise. That business forum helped you find that business coach. Facebook lets you stay in touch with that old classmate. Match.com helped you fall in love.
Designing social gratification
In the design of social software, the meaning of ‘fast and easy’ becomes vague. For relationships, “instant gratification” is harder to define, so the user interface is more difficult to design. Unlike with tools, there is no goal state to be reached through distinct steps. Social experiences have a larger design space. And unlike pure entertainment software, such as games, the design space is constrained by the messy reality of relationships.
Note that people are always building and maintaining relationships, with or without software. It’s just that social software makes it fast and easy, as opposed to painful.
That is why you pitch social software to the user as a network of attractive strangers he or she might get to know. They measure the value of your software by the ease and speed with which they can build or maintain their next meaningful connection. You have a bunch of strangers on offer, but all they care about is the next relationship. Any user has to walk a path toward any other user for a relationship to develop. This path is the relationship funnel. Your challenge is to design this funnel well. Most likely it is the main factor in the success or failure of your social software application.
Smooth relationship funnels
If you have a social application, you are in the business of creating relationships. There is a marketplace of connections, and you match the strangers together, on autopilot, in the best possible way. Your software is an idealized introduction, modeled after an old- fashioned face-to-face introduction. In the real world, the matchmaker role is a bit random. Being a real estate broker means a lot of running around for lukewarm leads. To many people, events for “networking” feel more like “working”.
That is the opportunity and challenge of social software: you take what is good about face-to-face introductions, but you say goodbye to the limits of cold hard reality. You make meaningful connections the fast and easy way.
Your software’s success depends on your detailed understanding of your target audience’s relationship funnel. You need to know the cold facts that you are protecting your user from, and which bits of information to share about the stranger that they could relate to.
This detailed understanding of a relationship funnel does not fit in a drawing on a napkin. It is hard to come up with in a brainstorming session. What is takes is years of trial and error of matching people in a real network. It’s evolution, not revolution. Successful relationship funnels are discovered over the years. They can be invented in a lab, but most of the plausible designs don’t work out well.
Discovering new relationship funnels
We can see how this discovery process works in the history of dating software. The first generation of dating websites played the role of matchmaker in a rather clumsy way. They were a database of unreliable and irrelevant facts about some of the people in town. The users could send each other a bunch of emails. You could pay by the month, or pay if you wanted to send someone a message.
Not much of a value proposition.
Nonetheless, a few fortunes were made. This first generation of dating websites took a step toward better matches than “real life” dating because these clumsy matchmakers solved an important problem; they empowered a previously unserved group. Similar how e-commerce with software-as-a-service helped small merchants sell online, the early dating sites served a target audience that otherwise wouldn’t get to date as much as they liked.
Merely having a list of some of the women looking for men and vice versa was a valuable feature. The relationship funnel worked because the strangers on the website were all wearing a sticker on their head that said “available!”, and you could contact them without anyone else necessarily knowing you did. It worked around the stigma that would come with facing a public rejection. It made rejection deniable. That way, finding your next relationship became faster and easier.
At the time, people didn’t talk about online dating. It was evident that ordinary and average people with normal relationships didn’t need a dating website. Or rather, for them, it would faster and easier to go hang out in a bar, join a tennis club, or go to a party. Their friends, that is, their real-world network, was a better relationship funnel for dates than the first generation of dating software.
The builders of early dating websites discovered a working business model, and matches happened. They took the first steps toward the ideal case: anyone finding the best matching person in town and building the kind of relationship of their mutual choice.
One of the big successes of this era was PlentyOfFish. The owner never took investment; he bootstrapped with minimal help and won big by having something that worked well enough and had a lot of users. The network effect did the rest.
Narrowing the funnel: getting closer to “right”
The second generation of dating software innovated the relationship funnel by boosting the accuracy of matches. At this point, the new features of a dating website would succeed if they limited the ways in which strangers could contact each other. If you just allow all the men to message any woman, you get a lot of wasted time and effort. The key was to realize that the absolute value of any possible match depends on two factors:
First, each person has attributes that are objective and universal and are therefore graded on a scale, together with “the competition” in town, offline or online.
Second, each person has distinctive attributes that are unique. These leave room for chemistry. Some facts shared about yourself on a website can be irrelevant to most people, but important to some.
Both these factors can make or break a match. Your software needs to take supply and demand into account while still allowing unexpected magic to happen.
The first website who got this right was OkCupid. They experimented with some ways to improve matches, and two things stuck: a matching algorithm, and hiding certain users from others.
They use a matching algorithm that is based on data-driven multiple choice questions, instead of old-fashioned personality tests. “Data-driven” means that the questions were asked in an order that was proven to reveal the most information, and the answers that were desirable for the user were weighed into a match “score.”
For example, some people don’t want to talk to anyone who brushes their teeth “every other day or so.” Or any profile they ever need to see has to be of their religion because their parents would never approve of a relationship otherwise. You can indicate how much someone’s answer to any question weighs in your matching score.
All this data was very revealing of the user base. OkCupid made the news now and then by sharing politically incorrect “facts” that appeared in their data. For example, women are more selective; men lie about their height and income; some races are more popular than others.
The sheer amount of users gave them the hard data that allowed them to draw compelling conclusions. They could use answers to predict behavior. For example, they found that the most informative question that revealed how likely someone was to want to have sex on the first date was, “Do you like the taste of beer?”. Those who answered “yes” were more likely to be more promiscuous.
The second thing that OkCupid got right was to hide “attractive” users secretly from “unattractive” ones. They can do this because they have features that serve other users to you and ask you if you like them. If they liked you as well, you would be informed of your mutual attraction. OkCupid can use this data to make an estimate of someone’s attractiveness and position someone on a somewhat objective attractiveness scale.
They generated some controversy when people started receiving emails that they were now tagged as “attractive” and would, therefore, be able to see profiles of other “attractive” users. Some people were offended by this and wrote angry blog posts about it. It makes perfect sense if you want to help people find the best possible match for their next relationship. No one loses. Fairytales are sweet, but there is no sense in pushing a Quasimodo or Shrek toward a Disney princess and suggesting they go on a date. The princess can still see his profile; if she wants to date Quasimodo because they have a similar taste in movies, then she can just drop him a line.
For the developers of a network funnel, it takes a lot of trial and error to improve the user experience. There were plenty of failed startups in the dating website game. At some point a website called Darwin Dating made headlines; they would only accept good-looking members as judged by a random selection of existing users. They had the same insight as the people at OkCupid, but they couldn’t execute on it well enough. Recognizing a problem doesn’t mean you have the resources to build the solution.
These two features OkCupid, the matching algorithm and secretly hiding attractive users from unattractive ones, were real boosts to the user experience of the website. Social software was now trying to be smart. It gave the user a way to contact people he or she could realistically date, based on what they told about themselves and on how much other people wanted to meet them. “We think this man here will be of interest to you; he wants the same type of relationship as you, and he thinks you’re attractive enough.”
This type of social software was now able to help a larger group of people. Browsing around on a dating site was now fun for ordinary people. They were still bothered now and again by unwanted messages or rejections, but they had a real chance of finding their next relationship among the strangers appearing on screen. Dating had become a little faster and easier. Some of the tedious bits had been removed, and the fun of getting to know exciting strangers was amplified.
As fast as you can: the next evolution
The third generation of dating websites refined the relationship funnel with location and the ease of a mobile interface. The leader of this trend is Tinder. They maximize fast and easy at the cost of accuracy and fairness, just because this isn’t important to most people.
Ease and speed are the essence of Tinder’s interface. Signing up requires no effort; it just imports your Facebook name and profile pictures, and you’re done.
Tinder recognizes that someone’s appearance is the first thing considered and that your location is a constraint. The user simply expresses a gut reaction to a picture, using a thumb or finger to swipe “yes” or “no”. They only see people who are near you, right now, and they can only message someone when they both swipe each other with “yes”.
This functionality emulates real-life flirting. A mutual “yes” swipe resembles good eye contact. You can size up an attractive person and drop them an “innocent” line. You watch their reaction, feel the quality of it, and take it from there. Appearance and location matter most, at first.
Every rejection is plausibly deniable, because if you swiped “yes” and it’s not a match, it could just be that they haven’t logged in yet or have left town. You only ever exchange text with people who swiped “yes” to your pictures.
After a few lines of chatting it is easy to choose to meet face-to-face or not. No questions and answers are needed. Tinder makes the aggressive assumption that all relevant facts are revealed in a few images, a swipe, and a short chat.
These innovations in user interface have made online dating a mainstream activity. Ordinary people can meet on Tinder and not be stigmatized as bad daters; naturally they could have met in a bar or at the tennis court, but they just happened to meet the fast and easy way by swiping right.
Tinder guarantees the luxury of never being bothered by the wrong people. This feature is free and baked into the user interface. It’s like real life, except better.
We see that each new generation of dating software has new features that boost the chance of a match while cutting the amount of time wasted. It makes the dating market more “efficient”. People find each other quickly, with minimal effort.
Tinder just wants to know if you find this person attractive. No? OK, you’ll never see him or her again. Yes? Let’s see if they find you attractive as well. They do. Feel free to send a message; you two should probably hook up.
The other dating sites have improved their relationship funnels by focusing on groups that have a hard time connecting in real life. AshlyMadision.com connects cheaters, and other websites focus on ethnicity (JDate.com), or region (DateInAsia.com). Various websites verify income level and other claims appearing in the profiles.
These examples show how social software evolves. It takes years of trial and error to discover which facts users want to know about each other when. As a user interface designer, you need a detailed view of what fast and easy means to your target market. Only then can you remove the friction from your relationship funnel and match strangers effectively.