Dan Siroker at Startup School 2013

Siroker describes trying to turn his college research in sentiment analysis into a stock market startup, Sentiment Solutions. Working alone, he went without a paying customer, then joined Google to learn how to build products. At Google, he saw how early users and their feedback could help a product improve.

Inspired by Barack Obama’s call for evidence-based government, Siroker joined the 2008 campaign and ran A/B tests on its digital work. He found the tests valuable but slowed by the need for developer help, which became the original inspiration for Optimizely. Before that, he and Pete Koomen tried CarrotSticks, an online math game, and Spread.ly, a service offering discounts for social sharing; both taught them about customer feedback and distribution.

Siroker and Koomen began exploring Optimizely after recognizing that Spread.ly’s model did not work. A former campaign colleague offered to pay about $1,000 a month for an A/B testing tool, before they had written any code. They built a prototype, followed Paul Graham’s advice to pursue the idea, and grew Optimizely to a $100,000 annual run rate in its first year, then $1.2 million the next.

As Optimizely expanded, Siroker applied feedback loops to hiring, using a higher bar for candidates and refining the hiring process. The company reached 42 people and a $7.6 million annual run rate, and later grew to 130 employees with its product available in 10 languages. For its Series A, the founders held mock board meetings with potential investors and chose Peter Fenton of Benchmark for the questions and feedback he brought.

Siroker sums up his approach as getting feedback from what you do and using it to improve. He says founders can fall into an activity trap by focusing only on running existing processes. Instead, they should define how their company works, as Tesla did with direct car sales, and keep refining that approach.