Brad Feld

Tag: experiment

This post first appeared in the WSJ Accelerators series titled The $100,000 Experiment in response to the question “When should you make a substantive change to one or more parts of your business model?”

During the past few years, the word “pivot” has become one of the most overused words with regard to startups. For some, it means a tiny incremental change in the business. For others, it means killing off whatever path you are on and rebooting the business, creating something entirely different. For others, it means everything in between.

A long time ago, I realized that every successful business was a continuous process of small experiments that operated in the context of a long-term vision. When an experiment worked, you did more of it. When it didn’t, you ended it and moved on.

The magnitude of these experiments are dependent on the stage and resources of the company. If you are a three person startup with very little money in the bank, your experiments are tiny ones. As you get bigger and have more success, your experiments can get larger.

I was once on the board of a company that was cash-flow positive early in its life. The entrepreneur decided to raise more money, even though he didn’t need to. I was perplexed and asked him why he was raising the amount of money he had decided to raise. His answer was that when he had no cash in the bank, he was willing to run $1,000 experiments. When the company was cash-flow positive, he was comfortable running $10,000 experiments. He now wanted to feel comfortable running $100,000 experiments, and this financing enabled him to do this. If he ran a $100,000 experiment and it failed, it wouldn’t tank the business.

When an experiment works, do more of it. So the $10,000 experiment that pays for itself in three days by generating $4,000 of gross margin on a daily basis is worth doubling down on and running at the $20,000 level. If this generates $8,000 of gross margin on a daily basis, double down again.

But if the first $10,000 experiment generates nothing, study the data that results from it. Make sure you measure your experiment. Create a hypothesis about what a successful outcome would be. Try to control as many variables as you can while you are testing something new, so you understand what is actually going on. If you find yourself devolving into a qualitative discussion about all elements of the experiment, you won’t learn much.

If your successful experiments are pushing you in a direction that is different from your long-term vision, or from the existing core business you are running, step back and think hard about what you are learning. Are your experiments conclusive enough to cause you to change your strategy? Do they reveal surface problems in your existing business or strong suggestions about better approaches?

If your successful experiments are doing this, then consider a serious shift in your business. But in the absence of this data, be very careful about defaulting into a mode of constantly and aggressively yanking on the steering wheel of your business. Instead, do small experiments, often.

If you want another perspective on this, go read the WSJ Accelerator article by David Cohen (TechStars CEO) titled Use Your Head, But Trust Your Gut.