Step by Step: The Incremental Approach to Crowdsourcing

Do you like electronic music? If so, listen to this new Avicii song, “X You”:

As you may have guessed, crowdsourcing produced this song. Avicii spent the last three months collecting musical samples from his fans, evaluating them, and presenting the best ones back to the crowd for them to pick their favorite. Bit by bit, from bassline to breakdown, his fans created one of the first crowdsourced pieces of music.

I’m calling this method the incremental approach to crowdsourcing; it involves a time-consuming but vigorous process of polling, discussion, and idea submission, with each cycle adding a tiny bit to the final product. It’s not the easiest or most resource-conservative way to crowdsource, but when the planets align and the controlling entity jumps into it with both feet, it can provide truly dazzling results.

Take Wizards of the Coast (WotC), for example, the company that makes the popular trading card game Magic: The Gathering. They recently started the fourth iteration of their “You Make The Card” campaign, which allows fans to help them create a card that will appear in a future release. We’ve also seen this approach utilized by Nissan, who used the help of their Facebook fans to build a limited-release performance racing vehicle. In both campaigns, the company built something great by letting their fans choose what went into it with a multi-step process that took several months.

How It Works

All three of these campaigns followed a shockingly uniform series of steps. The contest begins with the most general parts of the project and moves to the specific. For Avicii, the beginning was the song’s melody; WotC started with the card type, and Nissan first asked for an exhaust system that would do justice for the car’s engine. The crowd voices their opinion on which of several options should go onto the next round, and then the next piece is selected to be added on. More discussion, more voting, more submissions; rinse and repeat until you have a complete product.

It’s important to note here the distinction between incremental and iterative development. Both consist of several periods of discussion, voting, and designing; the main difference is what is produced at each step. Incremental development adds a new thing to the project each time; a new sound effect to a song, or a new ability on a card, or a new set of tires on a car. Iterative development, by contrast, would design the entire product at once, and then make it a little bit better better with each progressive pass. This is sometimes the method used to create Wikipedia pages, in which a heap of information is dumped onto a blank page and slowly pared down, edited, formatted, and given citations in the coming days.

What Makes It Great

The incremental approach isn’t for everyone; you need a ready-built fanbase that is not only numerous, but dedicated. Avicii is one of the most popular DJs in the world, Nissan is a multi-billion-dollar auto-industry leader, and Magic: The Gathering players have been dutifully flipping cardboard for the last 20 years. Because of this, these entities can afford to hold a contest that stretches out over months, where those less popular may find their crowds losing steam after the first few weeks.

Additionally, when designing a product where every piece of it needs to work well with every other piece, the incremental approach ensures that the crowd doesn’t get ahead of itself and the work is easy enough to swallow. If WotC had attacked this project using the iterative approach, they would have faced the tremendously difficult task of sorting through several thousand card submissions, some of which likely being completely unusable. By instead uniting its crowd on each consecutive step, WotC streamlines the conversation surrounding its project and focuses its audience on the task at hand, while not losing sight of the upcoming steps and the end goal.

Compare this to a project where each piece is designed individually. It’s been said (but never concretely attributed) that a camel is a horse designed by committee. This brings to mind a product that is designed simultaneously by several entities that have conflicting interests in the final outcome; some may want it to have functionality or features that the other groups aren’t interested in or actively oppose. In this manner, parallel design by many sub-groups of a crowd can create a product that while unique and novel, doesn’t really accomplish any one goal to a satisfactory degree.

It also helps that all three of the entities I am using for examples show great adhesion to the three tenets of successful crowdsourcing, as I’ve mentioned before with WotC and Nissan. Avicii also hits all the marks of incentive, barrier to entry, and compartmentalization; his fans are more than willing to put in the effort for the chance to hear something they created get blasted on the radio or through nightclub speakers, the submission process for samples was simple and available on social media channels Avicii’s fans frequented, and the incremental approach ensured that each new piece of music was carefully planned and fit with everything before it.

The Juice is Worth The Squeeze

I would love to see more companies use this approach in the future to replace the standard “contest” model of crowdsourcing that really only allows input from one creative mind. The incremental approach is truly the right choice for entities that want to tap into the collective knowledge of their entire crowd, with an added bonus of collecting market research as they go. Even if Nissan doesn’t end up using Exhaust System C, they will still know what their fans think of it after the contest is completed. WotC’s You Make The Card not only gave them the final product, but also inspired entire mechanics that went on to become very well-loved.

Let me know in the comments if you’ve seen a company use this approach to achieve great (or not-so-great) results, or if there’s someone who you think would benefit from changing their approach to this one.

Nissan’s Crowdsourced 370Z Completed, Fans React

 

Finally! Since the earliest days of this blog, I’ve kept a close eye on Nissan’s crowdsourced project to have their fans help them design a car. Now, it’s finally here. After months of polling on Facebook, tweaking, and gathering results, Nissan finally showed off the fan-made build to the world with this video:

Over the weekend, I’ve been passing this video along, browsing car blogs, and generally trying to get a read on the reaction to this final product. It surprised me to find that the announcement isn’t getting much traction; even on the Nissan Performance Facebook page, the very place where fans designed the car, the announcement video barely has any comments (although it’s been shared a few times).

With this being said, here are some of the first impressions of the car:

• Ugly/poor design. The fans don’t waste any time critiquing their own work. People will complain about anything, of course, but the general consensus is that there was a lack of an overall design goal, which is fundamentally true. Nissan could have avoided this by crowdsourcing a design goal as part of the project, but obviously it’s a little too late for that.
• “I’ve already built this car.” A bit of a two-pronged comment. On one hand, this person is disappointed that the car isn’t anything out-of-this-world special. On the other hand, this was Nissan’s goal from the beginning; they wanted to use off-the-shelf parts so enthusiasts at home could build their own fan-inspired 370Z. Take this comment with a grain of salt.
• We want more! By far the most repeated comment for the video was “That’s it?!” This does not represent a disappointment in the car, but a desire for more information. The fans want to see their car do some really impressive moves on a racetrack, and most of all, they want to hear what the engine sounds like. I can’t say I disagree with them; a performance vehicle should be performing if it’s being shown off.
• Still pretty cool. Despite the complaining, reactions are generally positive. The novelty of this project helped a lot; many commenters were unaware that a project like this existed, and think it’s neat that Nissan stuck its neck out to trust its fans. The “you-can-build-your-own” aspect is also a big plus, especially since Nissan has no plans to mass-produce the car.

In summary, this was an excellent first crack at a crowdsourced vehicle, and Nissan is in a very good position if they want to try a project like this again. This 370Z seems to be the “Two and a Half Men” of cars; it’s nothing super-special but it’s generally liked because it doesn’t offend anyone. If Nissan wanted to do Attempt #2, they should start with a fan-chosen design goal and end with photos, videos, articles, statistics, sound bytes, and virtual shows spewing from every pore of the Internet. As a matter of fact, that’s my biggest complaint with this project. You and your fans have made something really cool, Nissan! Show it off and get people talking about it!

 

The Three Tenets of Successful Crowdsourcing

Even in my so-far small experience with crowdsourcing, I’ve noticed that many campaigns succeed or fail based on the presence or absence of three criteria. I’d like to share them today, and get some feedback on how essential my readers believe these principles to actually be.

1. Incentive. I just got done reading a post by fellow blogger VisualBloke about the relations between human behavior and the social web, and it mentioned incentive as being influenced by tones of self-interest. Meaning, make it personally worth it for your participants. Most CS’ing campaigns have this built-in, offering participants the opportunity to contribute to something that would normally be out of their sphere of influence. Nissan’s Project 370Z is a good example of this; how many people have ever had the chance to design a mass-market automobile? Kickstarter offers a more clear-cut incentive system, with different levels of monetary contribution translating to “rewards”, extra influence, or early access to the final project. No matter how you slice it, participants need more incentive than “we need help”.
2. Barrier to Entry. How easy or difficult is it to join your crowdsourced work force? The right answer to this question depends on the level of talent you need from the people comprising your crowd, but the bottom line is to make your barriers low enough as not to deter any sufficiently motivated individual, but high enough to ensure quality. This seems complicated; let me explain. OpenStudy.com, as a resource to all students, wants to set its barrier low for entry because everyone from elementary schoolers to graduate students and beyond needs to easily access the site. They accomplish this with simple design layout and a no-frills registration process, making it simple for any student to get involved. By contrast, the aforementioned Project 370Z demands gearheads and performance experts. Hiding the project on the performance-dedicated subsection of their Facebook page was a move that may have seemed less than prudent at first, but actually succeeds in weeding out the less-than-desirable opinions without resorting to a dedicated moderator, simply by virtue of them being absent.
3. Compartmentalization. Present a crowd with a big problem, and you get a million less-than-helpful answers. This is the principle that rules overwhelmingly in the comments sections of political news sites. Break the problem down into manageable chunks for maximum crowd effectiveness. I hate to pick on Project 370Z again, but damn it, it’s such a good example of successful crowdsourcing. Had they presented their crowd with the simple directive of “help us build a car”, they would have gotten a hodge-podge of vastly different ideas, opinions, and (unfortunately) misinformation. Instead, they break the process down into parts, and then further into individual questions. “Help us build a car” becomes “which of the following three wheels best fits the Metalloy design?” Give people a manageable problem instead of an open-ended question, and you get more focused and relevant answers.

There are probably many more aspects of effective crowdsourcing, and even more that will guarantee failure, but I’ve noticed these three present in every successful attempt I’ve seen. Have you observed any crowdsourcing patterns worth noting?

Nissan’s ‘Project 370z’ Crowdsources a Car

In a move that could be accurately described as “the coolest ever,” Nissan has decided to use social media to crowdsource the development of a new car. The one-off 370Z sports car is Nissan’s way to reaching out to its dedicated and knowledgeable fans to use their input in the best possible way. Project 370Z is also a marketing masterstroke, since parts used to build the car will be off-the-shelf auto parts that any gearhead can pick up at their local shop.

The nerve center is Nissan Performance’s Facebook page, where the following post, made just over 24 hours ago, already has 85 “likes”:

Project 370Z: Stage One is now complete. A few of you noted the desire for larger pipes than we originally offered and we agree. Based on your comments, we’re bumping up the exhaust to a Greddy Turbo Ti-C – a better match for the previously chosen twin turbo kit. Thanks for the input. Later this week we’ll need your help in choosing the right suspension for the Project 370Z car.

The brilliant move on Nissan’s part was specifically targeting the Facebook page. The company notably shunned any sort of formal survey page or webapp, ensuring that the opinions it received were from dedicated, creative, and vocal fans who held car design and performance close to their hearts. This is what crowdsourcing was made for; reaching out to people who are anxious to offer their skills, but would otherwise be unable to.

Nissan will unveil the fan-designed 370Z in May at the ZDayZ auto show in South Carolina.

By Seth W