You might have tried it before: as you incrementally tweak your web application with small changes, you can’t seem to convert your users any better. It seems as if you’ve hit a dead end. A large number of A/B tests and multivariate testing seem to have found the best working solution. Is the essence of my application just not good enough to convert better?
You might have tried it before: as you incrementally tweak your web application with small changes, you can't seem to convert your users any better.
It might not be, however, there is a good chance that you have hit the local maximum. The local maximum is a point in which you have reached the limits of what you can get out of your current design. Even if you make a thousand tweaks, it is not going to get any better in its current structural form. You cannot optimize your way into better conversion. You need to radically change the fundamentals of the design to move forward – a big overhaul that rethinks communication, flow, graphic design, and combination of elements in general. You need to innovate.
Reaching the local maximum occurs frequently. It is most often seen when you rely on metric driven tests when making improvements such as A/B and multivariate testing. Making small incremental improvements while testing their performance represents in many ways a better, maturer, and more controlled design process. It can however also be a tedious and slow process as every small change needs to be tested. Improving your design in this way will lead in one way only: to the local maximum.
A good design is measurable
A good design only makes sense if it is measurable. In order to make the success measurable, you need to lay down what metrics you are trying to improve. Find out how.
If you were to redo your design from scratch, would you do everything the same way? Would you try a radical different solution? How many fundamentally different ways can you solve your current design problem? You initially chose to go for one particular of them and optimized it to perform the best it can. But what if a radically different take on the design problem would actually perform a whole lot better? What if there were a local maximum in another solution that would perform better – that would be a better design?
How do you know if there is such better design? You don't. The only thing you _can_ do, is to once in a while try a radically different take on your design problem. *This is when you innovate*
How do you know if there is such better design? You don’t. The only thing you can do, is to once in a while try a radically different take on your design problem. This is when you innovate – when you come up with a solution that does not have much resemblance to the prior.
Think of each local maxima solution as if it was a mountain to climb. The higher the mountain, the better the design. There is a way to get to the higher mountain from where you are now, but it takes changing the design in a more radical way. The goal is to maximize the potential improvement reached by optimizations. You can either climb the mountain you are currently on, that is to optimize your current solution or you can jump from one mountain to the other – to innovate into a new solution. When you jump to a new mountain, you never know how high it is, or how high up you landed – how big the potential is for optimization. So either you landed at the bottom, in the middle, or on the top of the mountain. In the bottom, there is much room for improvement through optimization, where there is none at the top – at the local maximum.
You might have arrived at a new solution that has huge potential in being optimized, why it could lead to better performance than where you came from. It could also be that you arrived at a solution, that was as optimized as can be from the start.
There is a difference between optimizing and innovating
You can improve your design solution either through optimization or through innovation. There are two phases of design: ideation and make changes – innovation and optimization. Once you’ve optimized your current solution to its local maximum, you need to think out of the box again, and make the leap to new radically different solutions.
You optimize through a data-driven approach that thinks of design as a logic problem where the best solution is found relying on data and tests of every design choice made. When optimizing, we rely on data and tests as our instruments for deciding upon what is the best design.
Design optimization asks: What works best in the current model?
Optimize through a data-driven approach that thinks of design as a logic problem solved by constant tweaking and feedback through data and tests of every design choice made.
The goal of innovation is to maximize the potential improvement through optimization.
You innovate through intuition. When we use our intuition, we make best guesses and rely on our previous experience. We study what others are doing and use best practices. An integral part of innovation is thinking experience, aesthetics, and flow as a whole. When innovating, we rely on designers, best guesses and discussion as instruments for deciding upon what is the best design. The goal of innovation is to maximize the potential improvement through optimization.
Design innovation asks: What is the best possible model?
How to optimize
You need to measure in order to optimize
You don’t know if you are actually optimizing your design unless you measure. Depending on your business and your goals, there is a number of different thinks you can optimize toward. These are some of the most common metrics designed for.
- Acquisition: How much does it cost to acquire a new user? Cost Per Acquisition (CPA), life time value (LTV) are common metrics. Your CPA should be lower than your LTV.
- Conversion funnel: Where in your flow do your users fall off as they move through a funnel.
- Prevention: Do you want to prevent users from deactivating their account? Design a way to decrease or prevent people deactivating their account and measure how well you’re doing.
- Activation: Do you want to keep your users at your site? Then design a way to constantly get your users to find new and interesting stuff. Common metrics are pageviews per user and time on site.
There is no fixed list for what metrics you can use. You can invent your own or go with the standards. Find one and pick one that fits the goal of your web design.
Read more: A good design is measureable
Use metrics that will work for all mountains
Choose metrics that are so general, that they will work for radically different solutions. Is your general goal to get more sign ups, then measure how well you convert incoming users.