Make Decisions With Data, Not Hunches

Your pride says your hunch is on the money, but be sure the numbers tell the same story before you take a leap.

shutterstock images

As an entrepreneur, nothing's more satisfying than going with your gut and getting it right. Every once in a while your nose leads you to gold, and you get to pat yourself on the back and enjoy the moment.

Most of the time, though, managing on a hunch is an exciting way to get things horribly wrong. When you're guiding a start-up business, making the best moves starts with gathering data to support those great hunches.

And the data is out there. Whatever the industry, you can track down numbers to give you direction. In the online arena especially, there is a constant flow of traffic and information that leads to cold, hard facts.

Over the years, I've learned that it is far more productive to go where the data takes you than "go with your gut." Two years into my co-founding TripAdvisor, I had a gut feeling that there was a burning need for a new feature on TripAdvisor. I was sure the masses wanted a VacationAdvisor service that would guide them through a series of questions and in turn tell them exactly where to go on their next vacation.

After six months of development, TripAdvisor launched VacationAdvisor only to find out that nobody was using it on the site. My initial urge was to double down and spend more time "working out the kinks." But after looking at the cold hard facts and conferring with my co-founder, I realized the project was doomed.

Lessons learned? Gather data early in your start-up lifecycle and go with the hard facts. Had I spent more time surveying our customer base early in the process I would have learned that consumers didn't want the product I thought was a sure hit. A few days of market research could have saved our company six months of development time and much wasted resources.

After that ill-fated experiment, I became a convert in decision making with data. At TripAdvisor and now at CarGurus decisions are only made with supporting data. If the initial data says an idea has no merit it gets killed and we move on - even if the idea is one championed by the CEO (me) of the company. There can be no sacred cows in the world of start-ups.

In fact, at CarGurus we have taken data decision making to what some might see as an extreme end. To test new ideas, we often run what we internally call the "under construction test" to gauge our audience's interest in a new idea. For a small segment (5% or less) of one day's worth of traffic on our site, we put up a link that invites visitors to click through to a new feature. When the consumer clicks on that link they are taken to an "under construction" page that explains that the feature is under construction.

Now, I'm sure this concept offends many product design purists. However, only 5% or less of a day's worth of our site users even see this experiment and as such we minimize the inconvenience to our users. The data we get in return, however, is invaluable. If nobody clicks on the link/call to action for the new feature we know that nobody is interested in that new feature and we take that as evidence that the feature is not worth building saving months of development time and wasted resources.

If you have the expertise to collect data in-house, use it. Even if you are not in the online world, spend some time talking to some of your customers before embarking on a new service or product at your company. If you have vendors that can gather market intelligence, call them. Go to a reliable source to get empirical evidence at your fingertips before you make a leap. Otherwise you're working from opinions, which are great for dinner conversation and bad for business plans. Start with the data, and you'll have a good idea of where you'll end.

Langley Steinert is founder and CEO of CarGurus.com, one of the most frequented auto research and shopping websites. Steinert was previously chairman and co-founder of TripAdvisor, the world's largest travel-information website. @langleysteinert