How Rubik’s Cubes & Layer Cakes can help you personalise your website

Personalisation is a hot topic at the moment and has been for a while now. 

Quite often it gets made out to be something more complex than it needs to be but if you follow some basic concepts, you can make personalised experiences simple.

Ingredients you will need: 

?? Some degree of a Data Layer. Not sure about this one? We can have a chat if you'd like, it’s worth understanding.

?? An understanding of Segmentation. Groups of people who fit into distinct clusters which are not able to overlap. Traditional segmentation means people in segments shouldn’t be in more than one segment. If they are, you’ve done your segmentation wrong.

?? An outcome you want to affect. Why are you personalising anything? Is it to spend money and waste time or is it to achieve better results?

?? Insight Fragments. These are clues of data that suggest you might be onto something.

?? Spicy Ideas. Get creative with your ideas. If you’re not sure if something will work test it. Sell in the test. That’s called AB testing and you might find your crazy idea changes the game.

Here are three examples of segment and personalisation ideas I’ve pulled out of Google Analytics using the data layers and data that are already available. Some of these metrics have used low volume sample data - They are only for demonstration. I've also focused on the ?? Insight Fragments & ?? Spicy Ideas for now.

#1: Segment: Age + Gender

One of the simplest ways, age and gender can often yield disparate results thanks to the differing core values and interests of your audience. Personalising the experience for each cluster can be a quick way to achieving better results. Using parents as an example, the values of a father may differ from the values of a mother. And it gets more complex when you explore how old their child is and how many children they have.

?? Insight Fragments:

The two most popular products over the past 12 months, in terms of website visitation, have been Product A and Product B. These two products have very different audiences. Product A’s audience was 60% male. Overall, 40% of the audience was in the 25-34 age range. Product B was 65% female and 37% in the 25-34 age bracket.

 ?? Spicy Idea:

Based on the age and gender of the visitor, perhaps highlighting a particular product most relevant to their segment may have improved the product purchase rate, assuming purchase rate is the metric we are wanting to affect.

#2: Segment: Geo

This one is super simple and it makes logical sense. Proximity is often a significant factor of importance to customers. Have you seen ads where the ad has been taken from another market, dubbed, and then run? This is an example of ads not being localised. You would normally tailor your ads based on geo, so why wouldn’t you do the same to the website and continue the experience beyond just the advertising? It is a laziness trap to assume all Australians are a homogenised group of people with interests that don’t differ even between states. 

?? Insight Fragment:

Without going into the why behind this, the data suggests customers in Queensland love their Product C & Product D more than NSW. NSW + VIC are all about that Product B. 

?? Spicy Idea:

Maybe customers from different states should see different products in different environments relevant to them when they visit. Maybe the environment around the product varies too. For example, in Queensland, all the Product C's are surrounded with Barbecues, Thongs, Surfing, and other things I stereotypically associate with Queenslanders (do some research first here so you avoid biases). 

#3: Segment: Time of day

Another simple way to personalise. All this would need is the viewers time zone and the time of day.  

?? Insight Fragment:

Turns out that between 9 am and 3 pm are the key ‘Action’ times. On surface value, the data suggests that there may be a pattern relating to school times as it picks up most significantly from 9-10 and begins to decline from 3. This is an insight fragment that might be worth exploring deeper to validate or disprove.

?? Spicy Idea:

Perhaps changing some of the CTAs on the homepage to be focused on completing this action during these times may yield a better action rate. 

BONUS: 

WTF do Rubik’s Cubes & Layered Cakes have to do with segmentation & personalisation?

How I like to think of Segmentation is a bit like a Rubik’s cube. It might seem impossible until you know the formula. Here is the formula I use for interpreting segmentation.

Imagine a cube with each dimension you want to segment by as an axis. This way you should be able to create a niche segment where you can start personalising. Just swap out Age, Gender, and Geo to be the different dimensions you want to cut it up by. 

In this example, let’s imagine Age is Young and Old, Gender is Male and Female, and Geo is The City or The Sticks. 

?? Segment A may be OLD + MALE + THE STICKS vs ?? Segment F which might be YOUNG + FEMALE + THE CITY. Literally the complete opposite. Notice how one person can’t be in two segments here.

So if you want to get interesting with your personalisation, you’d combine all three. This gives you a triple layer cake here, time of day + age & gender + geo to really push that experience.

So here’s the layer cake for the last example. 

Time of Day + Age & Gender + Geo

???????? Age + Gender by showing products that index highly against segments.

?????? Geo by showing an image of your visually recognisable local store. 

???? Time of day by tailoring the CTAs within the timeframe to be variations of ‘Action’.

Food for thought I suppose.

Note - It is important to review these as a proper cause and effect, these examples are some quick nuggets I found when looking this morning.