Several years ago I studied artificial intelligence with the Open University and I was introduced to Genetic Algorithms. Put simply, a genetic algorithm uses techniques inspired by natural selection such as inheritance, mutation, selection and crossover to find solutions to problems. Initially the algorithm will start with a number of randomly generated individuals scattered around the solution space, and the fitness of each individual will be assessed. Much like real life the fittest individuals will pair off and reproduce to create a new generation, while the weaker individuals will never reproduce. Along the way there will be the odd random mutation which will most likely result in a candidate being disadvantaged, but will very occasionally give an individual an unexpected advantage.
This got me thinking, websites evolve in a very similar way. They will often start with a weak design and little content, but as their fitness is assessed through Analytics and user feedback, designers will modify content, experiment with layouts and use ideas from more successful websites.
Take a look at the image below and imagine that this graph represents all of the possible versions of your website and the higher the point on the graph, the more conversions that version of the site generates.
Now imagine your website is currently represented by the pink star. The website is at a local maximum, so every time you make a small change to your site you will lower the conversion rate. Your only hope of increasing conversions is to be brave and make a dramatic change that may just take you over to the larger hill where small changes can result in more conversions.
The video below shows how the n-Coders site has changed over the past two years (this was created by rendering every git commit change to the home page).
It is interesting to see this video generated by gsource. This shows the internal activity on the website over the two year period.
Whilst we are constantly monitoring the performance of the n-Coders website and using this data to refine the site (and it is certainly performing a lot better than two years ago) there have probably not been any changes that were significant enough to result in a leap to a higher mountain. Unfortunately it is very expensive and time consuming to make significant changes unless you are a conversion optimisation specialist, but the starting point for improvement is always being able to measure current performance.