I’m writing a book in which I claim that “you need to market your science!”  Upon hearing the word “marketing”, a lot of scientists look at me as if I’ve gone over to the dark side.  It’s almost as bad as carrying the bubonic plague.

But marketing has got to be better than some of the stupid mistakes I’ve made in my career (maybe I should call it “career blunders?”).  More than once I developed a great idea and then it suffered in obscurity because of my poor marketing of the idea.

I’m going to illustrate the concept with real-life examples. This is the first installment of “science marketing blunders”.

It starts in 1991. I was a graduate student. I had just joined a new lab, bringing my computer skills bear on analyzing data from the latest generation of instruments for DNA sequencing.

I was skilled in computer science, but not skilled in marketing.

I developed a program, called BaseFinder while I was in Lloyd Smith’s lab.  Its goal was to reduce a complex “trace” signal data into a simple stream of letters conveying the sequence of the DNA… A…C…G…T. This process is termed “base calling”.

Section of DNA Sequence Trace

At the time, the only available base calling program was bundled with the DNA sequencing instruments made by Applied Biosystems, Inc. It worked ok, but it was a closed, proprietary solution. When it made a mistake, the user was forced to just eat it, and pray/hope (or attempt to cajole ABI) that it would be fixed in the next iteration of the software.

My program was developed to fill the gap left by the commercial software, providing an open-source solution that had several benefits:

1. It was more accurate than the commercial software.

2. It provided a numerical measure of confidence for each base position, so that downstream processes could be automated to use the confidence value, flagging “poor” calls for review.

3. Its method of scoring the data was “modular” using (new at the time) object-oriented programming methods. So if a user wanted the program to start accounting for new types of features in the data, they could just add a module, without disrupting the rest of the program.

4. The user could define their own data processing “scripts” that combined a set of analysis modules to optimize for their own experimental conditions. Once a script was defined, it could be applied to automatically process large quantities of data.

BaseFinder Screen Shot c1998

I was working in a well known lab, and the community was hungry for software like this. It was technologically superior software to what had been available.

A handful of people have used it over the years for specialized analysis tasks, but it has been far overshadowed by other programs, primarily Phred, which was developed a bit later in Phil Greene’s lab.

Why didn’t it take off? Because of lousy marketing on my part! It had great features and benefits that would have helped people, but I didn’t do the things needed to promote it effectively.

That is a waste of the big investment in that software (10+ person years of effort).

I was too focused on the technology to care about the marketing. That was my fundamental marketing blunder.

Here’s how it manifested:

1. I chose a computer platform that nobody was using, the NeXT (Steve Jobs’ company after he left Apple in the late 80’s). I chose it because of the great technology it embodied. It had the first object-oriented operating system and programming environment, years before Java existed. I thought that it was way cool and would save lots of programming time to use their special object-oriented programming language and toolkits. But this introduced an instant handicap – only people with NeXT computers could use the software! Those computers were expensive – costing $4,000 – $5,000 or more (equivalent to a $10k or more computer with today’s dollar). Also, people didn’t want to spend the time to set up and maintain a new computer platform in their labs. Because of that, the adoption rate of our great new software was almost zero.

2. Once it was working, BaseFinder was old and boring to me. I didn’t talk about it much. I didn’t submit abstracts to conferences about it. While we kept working hard at developing it, there was no concomitant effort to promote it. When I did go to conferences, I was always talking about the latest algorithm, rather than talking about what people really would have benefitted from, which is the software platform itself.

3. User documentation was poor. For the few people who did pick it up and try it out, we had little in the way of good user documentation, so it was a frustrating experience for them. It was a sophisticated and complicated program, that could do many things, but it wasn’t intuitive. Producing intuitive software is “marketing” in its greatest form. The user interface facilitates the user getting value out of the underlying algorithms. Conveying value is a key point of marketing.

4. Our confidence values were arbitrarily scaled and not probabilistic. Even though we were the first ones to introduce this kind of confidence value, our approach was superseded by Phred, which had probabilistic confidence values. Phred’s confidence values weren’t necessarily more accurate, but they were more understandable to humans, because they were framed in terms of something many scientists are familiar with (probability).

These problems could have been remedied with a mind for “Marketing” our work.

For example, I could programmed BaseFinder in a common language like C that could have been readily ported to other environments. I could have skipped the fancy user interface at the start, and then added the user interfaces later for distinct platforms (Phred’s development took that approach).

I could have spent just a few days writing up user documentation to help people get going with the software.

I could have submitted abstracts to at least two major sequencing conferences every year, talking about BaseFinder and highlighting what was new and how it solved important problems. Phil Green did that with Phred.

Instead, I was always too busy with the “technology.” I always thought my problem was that I hadn’t done a good enough job with my software. I thought that if I only worked harder and smarter, and made the software better, that people would discover it. Even after 10 years of doing that, and people still not discovering it in droves, I eventually gave up on it.

Paying attention to the concept of Marketing would have made all the difference in taking the technology that I poured my blood, sweat, and tears into (and the Government poured a lot of money into), and getting it used.  Marketing is all about conveying the value of some thing that you have to offer. In this example, I had a potentially great thing, but I failed at conveying it’s value.

Should marketing really be such a dirty word? Perhaps if used for nefarious purposes.  But using it to promote useful software is far from being nefarious.

—————

I have a few more good marketing blunder stories that I’ll share here in the future.

What about you? I want to hear about your science marketing blunders. Let me know about that time you should have gotten the Nobel prize, but you didn’t because of lousy marketing on your part, or someone who you were working for.

And, if you’re interested in preview of my upcoming book on how to avoid these kinds of marketing blunders in science, just sign up here.