Wednesday, January 07, 2015

Don't fail so fast

I am sure you are all familiar with the "launch now" strategy, touted by the tech industry for over a decade now.  It goes by many names.  Fail fast. Permanent beta.  These strategies are based on the idea that you shouldn’t spend too much time researching your design in advance. Instead, release your rough draft and use the market as your testing environment.  Ask your early adopters what they think and make changes in Version 2.0 accordingly.  The advantage is that you get on the market before any potential competitors.  If can build up your subscriber base, you get some network externalities and switching costs that blocks others out.  The downside of course is that your original design might really suck. 

My Take

But here is where I have a problem with this approach (besides early products sucking).  If you do some careful and controlled research, you can develop generalizable rules.  Generalizable means that you know in advance what attributes work with what kinds of products, what kinds of customers, and what kinds of use cases.  In advance!!  After a few rounds of controlled testing, you don’t need to test a particular product very much in advance to know how it will perform in the market.  You can release a very good product the FIRST TIME !!  Yes, I know all caps is yelling – I am yelling on purpose. 

Of course, there will be nuances that are different for each product that you won’t know for sure from the controlled testing design principles.  You are welcome to launch first and test nuances in the market with the intention to fix them in Version 2.0.  That is where A/B testing has its greatest strength.  But there is a big difference between using the market to test your fundamentals and using the market to test some nuances. 

Another benefit is that design principles will show you some global maximums in design whereas field testing locks you into the basic architecture and you wind up only finding local maximums.  If you know anything about optimization you will understand how important this can be.  It can make your design an order of magnitude better. 

Many of my students come from companies that have the launch early and fail fast mindset.  They don’t want to learn research methods and they prefer recipe style design guidelines.  Bake at 400 degrees for 15 minutes and out pops a decent meal.  If your guests say it needs salt, hand them a salt shaker.  But to me, that just isn’t good enough.  I want my user to rush out dying to tell all of her friends how great her experience was.  How unique and unexpectedly satisfying from start to finish. The proverbial “customer delight.”

Your Turn

Which do you prefer?  As a designer, would you like to have some generalizable design principles that you can turn to for each new design?  Or do you want to start with a blank slate and use the market to test each idea? 

As a user, do you want a half-baked product now and upgrade when the time comes?  Or do you want companies to do their homework in advance?