Hacking Traction: The Dark Side of Marketing Optimization

lock picking

In 1964, Justice Potter Stewart, who was having difficulty explaining what exactly he meant by “hard-core” pornography, famously said, “I shall not today attempt further to define the kinds of material I understand to be embraced…but I know it when I see it.” When entrepreneurs ask investors what they’re looking for in a startup, the response they get is often something along the lines of “traction.” And when asked to describe traction, most investors channel Justice Stewart, saying only, “I know it when I see it.”

Investors seek to take on the risks they can control and minimize the rest. The biggest unknown variable in the risk equation is usually the market, as in, “Is there even a market for your product?” This is especially challenging when it comes to products aimed at consumers, as you cannot possibly speak with several million people to get their feedback on a product before it’s even launched. So investors want to see enough usage of your product to give them confidence that there is both a big enough market to generate a large exit, and that you have nailed the product/market fit. That is what we mean by traction.

Misunderstanding the objective of traction, a number of Silicon Valley entrepreneurs have mastered the art of driving millions of people to a site or application regardless of whether the product solves a real problem or not. At best, this approach skews the data used to discover if a product is getting the kind of usage that would indicate there’s a big enough market for it. At worst, it’s spam.

Hackers hack everything, not just code

That things would go down this path should have been obvious. I’m not sure who kicked off this hacking traction phenomenon, but a small group of extremely bright entrepreneurs now use a very sophisticated form of multivariate testing. For example, they simultaneously test an extremely large number of variables on a registration page in order to maximize registration (the copy, size, color, font and placement of data fields; the format and text of buttons; the flow of pages), then do the same thing on the “invite friends” page (the text of the body and subject of the invite email), and so on.

This method takes advantage of latent human behaviors to drive registrations and visits — and in terms of those goals, it’s unbelievably effective. While this knowledge is still relatively well guarded, it is slowly making its way through Silicon Valley.

The insight that other pages say more about the pages they point to than such pages say about themselves (PageRank) was a revolutionary idea that led to Google Search. Machine learning has done a great deal to tune search, but machine learning without PageRank would not have resulted in Google.

Similarly, using very sophisticated optimization routines to maximize desired behaviors in a social application is a great way to make a good thing better. However, using these techniques too early can muddy the waters so that you don’t know if what you have is any good or not. If you were going down the wrong path, wouldn’t you want to know that as soon as possible?

Now that these entrepreneurs are returning to investors with “traction,” some are having trouble raising money. They are, understandably, confused. “You asked for traction — here’s your traction. Show me the money!

Too much of a good thing

Multivariate testing and other optimization schemes can be a great way to make a good product even better, and they are underutilized by many companies. But too many startups have begun misusing such traction techniques as a strategy rather than as a tactic, inadvertently destroying the feedback needed to build a great product.

As for me, I shall not today attempt further to define the kinds of businesses I hope to invest in…but I know them when I see them.

Mike Speiser is a Managing Director at Sutter Hill Ventures. His thoughts on technology, economics and entrepreneurship will appear at this time every week.