Like Me? Follow Me.
Businesses have always sought to improve their websites based on the knowledge they learn from the consumers, striving to reach a personalised service. From my past experiences working in an off licence I could quickly get the super strength ready for the local alcoholic at opening time. When I worked in the travel industry I quickly became aware that hardly anyone will fly directly into Barcelona, preferring the low cost flights into Girona or Reus. This information comes to the frontline support/sales with each customer interaction.
Scenario time: you have an online business having 10,000 visits a day. How much are you learning when you’re not interacting with the customer? Your front line customer support is the website itself. The website like any front line support should learn and adapt from its knowledge. So, how do we get a website to adapt?
Data Collection
It’s a very simple concept: a customer visits and you store whatever information you can. If a user came into my shop I’d be taking in what information I can. Is the customer male? From past experience it may be a bit weird for me to start calling him babe and sexy. Is the customer over 60? From past experience I’ve found they don’t want to buy glow sticks. The customer visits the fruit and veg aisle before they go to the sweet counter.
Your website should collect this information also:
1. Collect Demographical information: Facebook applications can take in pretty much a user's life story with permission. Everything from age and gender, to favourite bands or how many siblings they have.
2. Collect Geographical Information: If the user is on their smart phone you can also pin point location using web geolocation and/or probably a Google API, or server stats.
3. Identify The path they follow: What pages does a user visit? How long do they spend on that page?
4. What actions they took: What products they purchased?
Now I have recorded...
Name: John Smith
Age: 47
Location: Bolton
Likes: Hungry Hippos
Bought: Transformers 3, Giant inflatable beer can
You then acquire a record for every customer that visits the site.
Analysis
It’s pointless to have that raw data and do nothing with it! The point of having a website is to reduce man hours and overheads, it’s not worth looking at all this and trying to figure out what is going on when we have a perfectly good server that specialises in analysing data. You should analyse each part in clusters; even the most bizarre information can sometimes correlate to your conversion rates.
Clustering
The system can cluster data together, putting people into groups of similar characteristics. I put all the people who like Hungry Hippos together and use this to decide what action would be best presented to the user.
People who like hungry hippos:
80% bought how to win at hungry hippos book
40% furiously clicked on pictures of beads before signing up
20% phoned the call centre
1% bought Transformers 3
0% bought tacos
So instead of showing the user endless pictures of tacos the best thing the website could do is prioritise the page to view the book ‘how to win at hungry hippos’, maybe make some room for pictures of beads and possibly show the phone number.
Real Life Usage of the Adaptive Web (non hungry Hippo usage)
A/B Testing
AB testing is one of the simplest form of testing. It is simply what it says, do people prefer version A or version B.
Recommender Systems
Recommender systems are some of the most prominent and powerful systems on the web. When Amazon released its recommender system its revenue soared as it was able to lead customers into an endless buying cycle by knowing what they liked based on past purchases.
E.g People who bought `Star Wars: The Complete Saga [Blu-ray]` also bought ‘The Lord of the Rings Trilogy [Blu-ray]’
Content Discovery (dynamic navigation)
This is seen all the time in content discovery websites (reddit,digg) . Showing information the users find most popular. Another example of this is dynamic navigations that order/change based on user behaviours. Constantly trying to evolve and change to get the user to the action you want them to perform.
Let Your Site Learn From Experience
Your website is your front line. Like in the traditional market place the best customer facing employees are the ones who learn from experience and adapt to the individual customer giving them a more personalised service. Sometimes it may be questionable; sometimes it may just seem plain stupid, but once you allow your site to learn, it will. The end result will be incomparable conversion rates and constant improvement with minimum overhead.
If you want to allow your site to move forward without constant supervision, get in touch with our web development team. We have the ability to offer machine learning algorithms to different aspects of your site and trust in the information that the site displays. For more information, contact I-COM on 0161 4023170.
