Five ways to excel with personalisation
Personalisation is the process of creating contextual relevance for a user to engage them and build trust, ideally leading to a purchase, subscription, donation or link – whatever it is you want the end result to be.
The biggest problem with personalisation is that the term is so broad it covers a wide range of different practices. When most people talk about personalisation, they generally mean one-to-one facilitated by machine learning. What’s rarely mentioned is that personalisation has different dimensions. True one-to-one is complex and requires tech, processes and expertise.
Brands such as Netflix and Facebook have personalisation programs so advanced that organisations at the beginner level need to understand it will take time and experience to achieve that level of sophistication. Personalisation is a process, not a tool you can turn on and off.
So, where to start? I like to think there are five key steps to personalisation and each becomes more complex as you advance. We advise our clients to start with the basics, establishing the simple steps to set up their program for the future.
A very basic personalisation method is symmetric messaging. This is about matching what’s shown on the media channel to your landing page. For example, an ad on Facebook might have an image with some copy on it and, when the person clicks through, the landing page has a matching image with matching copy. It’s not sophisticated but it reassures visitors that they’re getting what they came for in order to reduce bounce rate. This is super simple to set up and gives every user coming from different channels a basic level of personalised experience.
The next level is single dimension personalisation, or ‘cosmetic personalisation’. This is where you personalise based on a single dimension, for example location, weather, first time versus returning etc. Most standard experimentation tools provide this style of personalisation as part of their basic package.
We once ran an experiment with a clothing client where we personalised items using the weather as a dimension. On a rainy day we pushed items like long-sleeved tops and raincoats to the front page.
Cosmetic personalisation is simple to set up but needs a level of strategy to think through the experience design, the size of audience, the content experience and so on. It’s a little more complex than symmetric messaging, and powerful when done well.
Focused on how someone behaves on your site, this style of personalisation uses behavioural information to target a contextual experience to this user type.
For example, you may notice that users exit your store on a specific page, such as the category page. You might consider targeting these users at the point of exit with a limited time offer if they purchase within a specific timeframe. Another common example of behavioural personalisation is a cart reminder for a returning user.
This style of targeting is super powerful if you see specific behaviours on your site.
One-to one personalisation is a lot more complex. It requires you to stitch together data points from several sources to attain a more holistic picture of the customer. This might include a phone call to call centres, an in-person visit and/or a web chat as well as the webpage data collated to find the next thing the user will be interested in.
Say the user is looking for a home loan. They go onto a bank’s website and search for a home loan specialist and input their details to make an appointment. Later, they realise they can’t make the time so they call in to change the appointment. A good one-to-one practice will, when they return to the website, reflect the updated time with a confirmation of the changed time.
This kind of experience requires cross-functional teams, heavy data analysis, technology and a lot of work.
A crucial step
One of the biggest mistakes organisations make is spending a long time planning for personalisation before switching it on. But personalisation isn’t a project that has an endpoint; it’s an ongoing practice. Building layers contributes to process, and once process is embedded, you can create better infrastructure and then increase tempo. It’s about learning to walk before you run.
Start with some of the simpler personalisation steps first so you can understand the implications of what’s needed. It will also provide you with data to leverage for further experimentation, for example testing the effectiveness of different offers.
Courting customers takes work, and personalisation should begin with a series of small steps, simple but effective methods that contribute to the wider practice of personalisation. There’s a whole lot more I want to say about the role of behaviour in personalisation but I’m going to save the heavy-hitting stuff for another post. Stay tuned!
And if you would like to know more about personalisation and conversion rate optimisation (CRO) please tune into the New Republique Podcast, Australia’s first podcast dedicated to all things CRO (conversion rate optimisation), experimentation and personalisation. You can listen and rate us via iTunes or Spotify