Authentic Incentive Systems

How the wrong rewards can ruin a good thing.

Successful online communities make it easier for people to do the things they love to do. The satisfaction that comes from helping someone, creating something, or sharing your expertise might be small compensation, but the internet makes participation so easy that mild rewards are more than enough. Amazon, Wikipedia, Flickr, and countless others have built empires atop this humble foundation.

Like Rome, however, such communities are not built in a day. It can take months to attract a core group of contributors and amass a respectable library of content. New entrants are understandably eager to accelerate the process, but many traditional incentives are bad medicine. They undermine your users' intrinsic motivations.

Lets consider a hypothetical example. You've just launched a new website that allows users to rate and review free software. Having an active core community of reviewers is essential. A few hundred people have created accounts, but the reviews are only trickling in.

You decide to kick start your community by running a contest: Next month's five most prolific reviewers will each receive a cash prize! Nothing motivates like money and, sure enough, activity rises rapidly throughout the month. When the contest ends, your database has grown fat and your activity feeds are hoppin'.

Then, activity drops off a cliff. For a few weeks, it's even lower than it was before the contest. Sure, you have more reviews than you did last month, but many of them are lacking in detail and concern popular software that most consumers already know about. This is more than just a matter of rewarding quantity over quality. It's a matter of motivation.

The Overjustification Effect

The traditional summary is as follows: "External rewards undermine intrinsic motivation." If that's how you want to think about it, you'll do fine, but there's a little more to it.

The real problem in our hypothetical example is that a powerful reward (money) overshadowed weaker rewards (personal satisfaction, the esteem of others, etc.) that were already in place. These kinds of reinforcement schedules produce high rates of behavior that spike just before the reward is given out. Afterwards, activity drops below its original levels and may take time to recover. (For a more detailed discussion, see my article on the Overjustification Effect.)

Incentives can also undermine quality. Intrinsic motivations like altruism and self-esteem promote quality content, because they're all about the audience. If the people consuming your content don't see value in it, what's the point? The internet amplifies these kinds of motivations by facilitating direct communication with your audience (via comments, user ratings, discussion forums, and email).

Contests, on the other hand, are all about winning. They take the focus off the audience and place it squarely on the rules and the prize. Do the rules reward quantity over quality? What are the minimum requirements for entry? How can I game the system? These are not questions you want occupying your core contributors' minds.

There are two kinds of external rewards that don't trigger the overjustification effect: performance feedback and praise. These can form the core of a more authentic incentive system, one that keeps contributors focused on their audience.

Amplifying Intrinsic Motivations

The web runs on metrics. You can track any activity performed on your site, but don't confuse "can" with "should." Identify the actions that mean the most to your community, track those actions, and then expose the resulting metrics to your users. Just letting people know how they're doing can be a powerful incentive.

User profile at Yelp.com

In our example, the most important site activity is writing new reviews. Tell users how many reviews they've written and how long it's been since they wrote their last review. When another user rates or comments on one of their reviews, let them know about it. Provide graphs of their page views and user ratings over time, so they can improve.

Let them compare their performance with those of other reviewers and the community at large. Bullet graphs are great for this. The more engaging your visualizations, the more often they'll get used.

Give top performers the recognition they deserve. Clear some space on the home page to list the most read, highest rated, and/or most prolific contributors. Showcase their best and/or most recent work. Put special badges on their profiles, so other users can easily identify them.

bulletgraph: below average bulletgraph: above average

Bullet Graphs

Always generate such lists based on a rolling average, like the last 30 days. If you just use lifetime totals, your first contributors can cement unbeatable leads. This demoralizes new members and allows the old blood to grow complacent. Highlighting recent data helps maintain stable rates of activity and levels the playing field for all.

Rolling metrics are also more sensitive to new patterns of behavior, which makes them better tools for corrective feedback. If the quality of a contributor's work is declining, past data can obscure the trend. A sudden drop in their average user ratings, on the other hand, is an unambiguous signal to get back on track. (See also my article on Shaping Social Behavior Online.)

Honest performance feedback can produce constant rates of behavior and quality performance without triggering the overjustification effect. Let your top contributors know how they're doing and they'll keep doing well.

Surprise Prizes

If you do feel the need to give out prizes, just don't turn it into a contest. Track the targeted behavior without anyone's knowledge, then surprise the top performers with all the fanfare you can muster. This publicly reinforces the desired behavior without inviting a flood of questionable content or a post-contest drop in activity. The effects may not be as striking, but they will be more stable and longer-lived.

The more unpredictable you can make such rewards, the better. Our software review website could send t-shirts to the highest-rated reviewers one month. Two months later, they could send a cash prize to the user who posted the most comments.

(If you want to get technical, and I do, this approach replaces the fixed-interval reinforcement schedule responsible for overjustification with a less troublesome variable-interval or variable-ratio schedule. See also my article on The Deep Magic of Behaviorism.)

Many of the more traditional tools of motivation are not designed for use on volunteers, especially not volunteers who are engaged with their audience and embedded within a community. Understanding their intrinsic motivations, and amplifying them, is the key to building an incentive system that works.