Statistical Proof that Marketing Web Stuff is Useless

Posted by Andy Singleton on Oct 18, 2007 12:38:00 AM

According to the evidence that I have, the marketing that you do to attract users to Web applications and Web 2.0 sites does not make any difference in the rate of adoption. You can get more people to register, but you can’t get more of them to adopt.

Assembla recently became a sponsor on Techmeme as part of our new (and possibly useless) marketing efforts. How effective is the Techcrunch / Techmeme / Silicon Valley press machine? It generates traffic. Does it change the odds of success?

Here’s an Alexa chart showing Assembla (no marketing or megablog mentions) with two other services that were mentioned in Techcrunch and other blogs. I follow these services because we have “tools” in our workspaces that compete with them – our SVN/Ticket/Alert system competes with Unfuddle, and our new Images tool competes with Conceptshare. We have offer a lot of tools, and we try to learn by matching each of them up against some point solution. These graphs are typical. In the lines for Unfuddle and Conceptshare, you can see the big bumps early on from a Techcrunch mention, followed by a consistent decline in traffic. For Assembla, you see a classic exponential adoption curve.

I brought this up last year in Marketing in a Web 20 World:Why the Best Products Sometimes Win. To summarize: Traditionally, the job of marketing has been to get prospects to come and take a look at your products. If you have to drag them to the showroom, that’s a tough job, and a major contributor to success. On the Web, prospects just have to click a link. It’s easy. Too easy. But, getting them to come back a second time is much harder than it used to be. They have so many more places to go, and all that motion is equally weightless, and adopting a Web service requires time they don’t have. So the field of competition has shifted, from outbound marketing, which only needs to be barely adequate, to the product, which must be good, and the adoption process, which must be fast.

I wrote that article after working with a client to launch new Web 2.0 service for a certain class of professionals.  My client was good at marketing, and he engaged a big industry publisher to promote the service. We were registering a lot of new users, tens of thousands of them. However, these registrations didn’t translate to adoption. Hardly anybody was logging in and using the service on a regular basis. Bad news. When I looked closer at the usage patterns, I could see that there was a hard core of real users, and these users were inviting their colleagues, and the base was growing. Good news. The only problem was that this core was tiny, less than 100 people. They were people who had been engaged as beta testers early in the product development process. It was obvious that there were two different types of users: those who registered because of marketing, who did not adopt, and those who registered because of invitations and referrals, who did adopt.

The adoption of that intensively marketed service matched almost user-for-user with the adoption of, its polar opposite. At that time, a year ago, was a prototype-quality project, and the entire traffic generation strategy consisted of links on two related wikis. The total time invested in outbound marketing was about 30 minutes for posting those links. This halfhearted promotion resulted in an extremely limited stream of visitors, and only 10 to 15 registrations per day. But our rate of repeat usage was about the same as a service generating 100 times as many registrations.

Apparently, registrations grow from marketing, and adoption grows from referrals. Or, to put it another way, a user is much more likely to adopt after a referral or an invitation, than after seeing a link and clicking on it. I can’t explain it, but it is the underlying truth in the numbers.

I’ve seen this pattern several times now. For example, we started making an effort to market a few weeks ago, and we tripled the number of new people coming to the site. Growth in usage has been good. However, growth in usage was also good during September, when we did nothing. If you look at a chart counting people who logged in twice during the week, there isn’t even a blip in October. It’s a smooth curve, uninterrupted by marketing efforts.

Now that I am working on the blog and talking to bloggers, I can see that blogs face an adoption challenge that is similar to applications. It’s pretty easy to bring several thousand people to look at a decent article, if you give it an obnoxious title as demonstrated here. But, only a handful of those people will add an RSS subscription.

Maybe I’m just not a good marketer. My volumes are low. I’m not in Silicon Valley. I’ve never been mentioned in Techcrunch. I might be so far out, Pluto looks like “in” to me. So, look at the behavior of Google. They are the biggest, most successful Web company in the world. They are located in Silicon Valley. They don’t do marketing for their new online services either. They just put the features out there and let people try them. And, according to this report, Google doesn’t advertise, and spends little on marketing. Google grows by referral.

A service that grows by referral has the characteristic mathematical pattern of exponential growth. The growth rate is proportional to the number of users. Exponential growth doesn’t necessarily mean fast. When you have a lot of users, it’s fast, but when you have a small number of users, it’s really slow (see graph above).

You’ll lose your shirt if you bet big on the slow end of an exponential. Fortunately, if you wait long enough for an exponential, it will get to the fast part. Patience and longevity are important.

Our adoption curve is close to exponential because it is mostly from team member invitations – a type of referral. I like to think about it as a having a doubling time. If our growth rate is 21% per month, or .6% per day, then the doubling time is about 4 months, and we grow usage about 10 times during one year. That is almost exactly what happened in the year from August to August.

Is there a way to force the issue? In my observation, yes, but it’s not with marketing. It’s with small adjustments to the product quality and the adoption process. Why does one social network take off, and the other flop? Hard to say. It’s small things that make a difference in the daily exponential. From what I have seen, it’s not features that make the difference. We added and improved a lot of features on, without changing the basic 3 to 4 month doubling time. However, over the summer we made improvements to the design and reliability of At the end of August, our daily increase moved to about 1% per day, which produces a doubling time of about 2 months (we just hit the first doubling after 8 weeks), and an expected usage increase of 30 to 60 times in a year.

Time is Money, and you can use Money to buy Time (the time to reach a certain level of adoption). However, if you spend the money on marketing, you will be spending a lot of money to buy a little time. According to the statistics, your odds are better if you spend money on the product

Topics: assembla, software business

Written by Andy Singleton

Working on Continuous Agile and Accelerating Innovation, Assembla CEO and startup founder

Follow Assembla

Get Started


Subscribe to Email Updates