2 Comments

Summary:

The secret behind Aardvark’s success was acute awareness of how close they were to failure, Aardvark co-founders Max Ventilla and Damon Horowitz said Friday. They detailed a process of rapid idea rejection and extensive testing throughout Aardvark’s short startup history.

Aardvark, by some measures — accumulating many active users or making lots of money, for example — did not succeed. But hey, it did get acquired by Google for a cool $50 million, so it’s worth hearing a little more about how the social search startup succeeded on that front. The real secret, according to Aardvark co-founders Max Ventilla and Damon Horowitz, who were speaking at the Startup Lessons Learned conference in San Francisco on Friday, was acute awareness of how close they were to failure.

Aardvark co-founders Max Ventilla and Damon Horowitz

“I’m probably going to fail, so how can I increase my chances of success but — more importantly — use that failure and make it less painful and minimize the risk of the catastrophic fail, when you’ve run out of money and you actually have to change jobs,” was how Ventilla described their thought process.

“Once you’ve sold your startup, it feels very much fated. I don’t think we felt that,” Ventilla said. That’s despite the fact that he had formerly worked at Google and Horowitz had founded multiple acquired startups. But the truth is that the Aardvark founders had no idea what their product would be or if they’d be able to build it.

Aardvark's abandoned product ideas

In order to find out, they got users to test-drive their ideas. For the first six months Aardvark prototyped ideas, gave them to 100-200 people, and if they saw they weren’t taking off, abandoned them (a slide of five abandoned ideas is pictured on the left). Once they figured out that social Q&A was the ticket, they didn’t pull back on user testing, bringing in 6-12 users a week over the 30-month span of the startup.

But Aardvark didn’t actually build products right away. The service connects people with questions to those in their broader social network with corresponding answers, but for nine months, a human being was involved in every single interaction — a kind of “Wizard of Oz” that classified the query and otherwise managed the conversation from behind the scenes.

In the meantime, Aardvark recruited its core team and raised $7.5 million. In fact, Ventilla said was easier to convince investors and prospective employees that Aardvark could figure out how to automate something that people were already using than to get people to use something new.

Aardvark's long path to product automation

Once Aardvark felt it was on the right track, its next goal was to improve its ability to learn and engineer faster, said Ventilla. The entire company — 30 people at the time of acquisition — reviewed user testing, feedback and metrics, and split testing for new features on a weekly basis. “As long as you don’t run out of money, if your second derivative is good you’ll leave everyone else in the dust,” he argued. Aardvark also made sure that users and advisers alike knew their opinions were being heard and implemented. “Part of your job as an entrepreneur is to be barraged by opposing viewpoints,” said Horowitz. “If you’re not prepared to be grateful, than you shouldn’t do a startup.”

“This is a sinking ship from day one, and that’s why we’re going to do lots of things and be totally uncompromising about when we abandon one ship and try to get on another leaky boat,” said Ventilla of the startup experience. Or as Horowitz put it, “Assume you are wrong, and then at least you’ll be correct in that assumption, if nothing else.”

A video of the talk is embedded below:

http://www.justin.tv/widgets/archive_embed_player.swf
Watch live video from Startup Lessons Learned on Justin.tv

Related content from GigaOM Pro (sub req’d):

Google’s Social Scheme Hinges on Fears, Not Fortunes

By Liz Gannes
  1. [...] Actually, that’s something that startups are sometimes doing and that has been worded by GigaOM writer Liz Gannes as Fake it till you make it in an article on Aardvark. [...]

    Share
  2. I think this is a great concept, but entrepreneurs need to be careful about how they actually implement it. Spinvox became infamous for carrying this idea too far when it was found that they “automated voice recognition” technology was basically call center reps manually typing out voice messages.

    Story for gigaom no less:
    http://gigaom.com/2009/07/23/spinvox-voicemail-transcription-done-by-call-center-staff-not-algorithm/

    Share

Comments have been disabled for this post