Fog Creek's marketing experiments, and the release that beat them

In May 2001 Fog Creek Software earned $6,463 from licenses. Joel Spolsky and Michael Pryor had started the company as a consulting shop, lost that business in the dot-com crash, laid off their two salaried programmers and were left with two products and a rented brownstone. Spolsky later told the story of the next two years in his column for Inc. in 2009, and it reads as a catalog of well-intended distractions.

They chased an eBay seller they took for a pirate, and it turned out the discs held an old shareware program with the same name as theirs. They spent weeks negotiating what they thought was a partnership with a website for webmasters, then discovered the other side was selling them advertising. They built an affiliate program: about two hundred people signed up, forty sold anything, and only three earned more than a thousand dollars. A mailing to existing customers cost three hundred dollars and brought in five hundred. A 72-hour discount inside the trial version took months to build and did little.

Then, late in 2002, they shipped FogBugz 3.0, a much better bug tracker at a higher price. Monthly sales went from around seven thousand dollars to more than thirty thousand. Spolsky’s conclusion was that they had spent years treating a product problem as a marketing problem. The exact sentence above does not appear in that column, which phrases its lesson as making the product more compelling, but the idea is the same and the column is the clearest evidence for it.

It is worth being precise about the conditions. FogBugz was sold to programmers who could download a trial, and the programmers already came to Spolsky through his blog, which by then was one of the most widely read in the field. A better version reached people who were already looking. The line was said by someone whose marketing had been done for him in a different way.

When those conditions are missing, a better product can lose. Palm’s webOS was praised by reviewers when it arrived in 2009, and in August 2011 HP stopped making webOS devices a few weeks after launching its tablet, because too few developers and buyers had followed. Enterprise software sold through procurement departments shows the same pattern: the product with the better sales team often wins against the product with the better features.

So the rule holds for software that people can try on their own and that improves visibly with each release. For a first-time founder that description covers a lot of products, and the Fog Creek years are a useful warning. When sales are flat and every marketing idea seems urgent, the dull option of making the thing better is often the one that works.