The P in PC stands for personal.
The reason we’re so attached to our screens is surely to do with the limitless sense of connection they bring, but also, to some extent, our relationship with connectivity is bound up with our sense of personal identity.
Consider the feeling we get when asked to operate someone else’s phone or computer.
Being greeted by an unfamiliar start screen just feels wrong. It’s way too, impersonal.
No one can deny, we have an intimate relationship with technology, much as we may have had growing up with Meccano, Panini stickers or Cabbage Patch dolls (have I dated myself? I don’t think so). When we boot up and log in, we want to feel like Norm bowling into Cheers (now, I have).
And for all that fearful talk of data privacy - let's wise up - it's a total moot argument. We choose to leave our anonymity at the door - happy to give up our personal data in the hope of a pitch perfect web page.
The web design industry is well into middle age. In all that time we’ve seen many fads and phases come and pass, but one thing has been steady. The holy grail. To create personalised user journeys. The kind that magically adapt to us based our preferences, actions and demography.
Yet precious few brands offer much more than lip service to the job of making websites feel particularly unique to us, in spite of what we all know to be true, that, with regret, our attention spans are in rapid decline. We flee any website that frustrates us, even if in seemingly trivial ways.
It’s hard to see how this pandemic of diminishing attention might ever hit reverse. The genie is out the bottle, in an Uber, heading for Heathrow and - don’t we all agree - it's only getting worse? The kids aren’t suddenly going to become more patient, forgiving or alert to the nuances of our laboured brand storytelling.
However, as we all know, for every change in societal behaviour (such as shorter attention spans - keep up) technology always responds in kind.
Artificial intelligence services make these topics especially consequential right now. Urgent, acute thinking is required from brands in order to ride the wave of waves that’s building on the now visible horizon. The metaverse is already here. A significant paradigm shift is brewing.
We all sense this. We also sense, I believe, that the winners will be those who have invested in a personalisation strategy smart enough to make everyone of our customers feel warm and fuzzy in a world of unlimited distraction.
Of course, a big job lies ahead. Does anyone have all the answers? Probably not. And given the near infinite variables tabled before us, it's little wonder we might suffer some analysis paralysis.
So, where to start? The principle of Occam’s razor is useful here. By clearing away unnecessary information and addressing only those things that present the fewest assumptions, dependencies and complexities, we’ll make headway.
With this in mind, here are some pointers we may find useful as we set off.
The basics
Web personalisation means using algorithms to anticipate our customers and display our most useful piece of content, right now, in real-time.
When it's done right, it blows our mind.
How many times have we entertained the conspiratorial thought that Facebook simply has to be listening to us? How else could Zuck have known we’re considering a trip to the Maldives and tempt us with just such a holiday advert?
Coincidence? Well… Conspiracy theories aside, Facebook is in a league of two (Jeffrey Bezos, please stand up) when it comes to pin-point personalisation. With thousands of coders, testing millions of combinations of algorithm every day, we could say Facebook and Amazon invented modern personalisation.
Well done BeZuck! However, it’s probably safe to assume that most of us aren’t attempting wholesale digital colonialism. Indeed our needs are far more modest and less nefarious (we hope).
The reason we want to apply personalisation to our website is to increase the conversion rate, right? We want to get more folks to submit expressions of interest, to buy from us, to subscribe to us, to hit that link — and hit it again.
To be even more straight-talking we want to convert anonymous website visitors into happy loyal customers, the kind of customers we’re on first name terms with, fleshed out in the sort of detail that helps us go above and beyond. After all, exceeding expectations, as trite as it sounds, is where it’s at.
Personalisation (combined with a great product and thoughtful services inculcated in a willing culture of experimentation) is the secret sauce of loyal custom, or as as Kevin Kelly wrote in his 2008 essay, leads us to ‘True Fans’.
Obviously, we can’t personalise without at least some information. So the first step in our strategy (which is also the most obvious, and gratifyingly, the easiest thing to do) is to start asking very politely for users' email addresses. It’s surprising how many websites seem reluctant to do this. But it’s crucial, so let's not overlook it.
Our websites are very capable of working out if we think it’s a user’s first time on our site and if that’s the case, we should be ready with a simple set of logical rules that let us do two things really well: firstly, sign-post popular content pieces (or commonly bought products), and secondly, trigger respectfully gentle calls to action — something like:
‘To say thanks for visiting our website, let us send you an invite to an upcoming webinar…’
You get the idea, and of course, many of us are doing just these sorts of things already.
But, when our visitor reacts to our call to action and clicks, ‘submit’, where does that email address go? What happens next? These are the crucial questions that deserve thoughtful attention.
In the future, prediction will be the priority.
Customer Data Platforms (CDPs) are first and foremost web analytics tools. They monitor how customers use our websites. But they’re different from something like Google Analytics because they use traditional and machine learned algorithms to predict a customer’s future intention. That difference, makes all the difference.
In other words, the AI says:
‘...I’ve observed this person, they’ve done a variety of things, I think they have a very high probability of converting, so let’s run a campaign for this individual to capitalise on their enthusiasm…’
And, of course, at the other end of the scale, the Data Platform can detect those who appear to be at risk of abandoning us in favour of our competitors. Products like Optimizley’s Data Platform will then organise these visitors into customer lifecycle categories such as loyal customers, regular repeat customers and those at risk of churn or not converting. Not only this, but the software allows us to drill for micro-detailed views of each user, letting us see exactly what they’ve done, with every interaction time stamped — which is pretty amazing.
What does this mean? Well, it certainly means that the days of patchy marketing lists and generic email blasts should be banished to the same mental archive where we might find floppy disks and Ourprice. Data Platforms allow us to develop predictive models in much the same way as Facebook and Amazon do. If you weren’t interested before, you really should be now! So to answer the question, where does that email go? It should go to our Data Platform, because that’s where the AI can do its thing. We need a CDP in our marketing technology set up, because (without wishing to be unnecessarily apocalyptic) if the AI isn’t working for us, we can be pretty sure it’ll be set against us.
The wise marketer makes little distinction between website and email. Let’s be honest, customers who have not yet bought from us (or not yet worked with us) are unlikely to convert during their first three minute browse of our site. They’ll need a nudge. Probably a few. Probably quite often. Email is of course, great for this and the more customised that nudge is, the more likely it’ll bear fruit. Even once we’ve convinced a prospect client to commit, more nudges will be needed if we’re to develop the sort of long-term loving relationship we hope for.
In our example where we ask for the email address of a first time visitor — their data will be processed by our CDP. Which in turn will tell our email system to send the right kind of message, designed explicitly to bring that person back to our site where the next level of personalisation will begin all over again — setting in motion a cycle of increasingly specific promotions. With each visit, our CDP gets smarter and starts making better predictions. That's an exciting thought, right?
And so it makes sense that personalisation truly comes alive when its thought of as way to improve the total lifetime value of our customers - sweeping up passive prospects on a journey toward becoming die-hard fans.
The end in mind.
Too often we’re encumbered (rather than empowered) by our existing technology. Too often we’re entrenched in biases associated with the way things have been done before. How much of our decision making is unfairly influenced by sunk investment? It is important, sometimes, to allow ourselves to consider what would we do if we were a nimble greenfield start-up. In practice, this is nearly always impossible to do, but that shouldn’t stop us from conducting this healthy open-minded thought experiment. Because, whatever stage or type of business we may be, its almost certainly true to say that our mission (meaning, the work we’re doing in the present) is to deliver the best possible experience for our customers by embracing every techno-advantage available to us. Our vision (meaning, what we hope to see in our future) is that our customers believe we are so tuned in to them, we can predict the next thing they’ll want from us, and it’ll feel just like magic - whether we’re sat at a keyboard, scrolling our phone or wearing funky augmented reality glasses.
It may sound facetious, but it’s helpful to visualise our vision. To do so, even in simplistic ways, makes the mission clearer and the future feel far more attainable. Take a look at this concept of a Homepage. A concept, to help us visualise a future where our homepage is entirely dynamic - uniquely personalised for every individual user.
Some among us may ask: where is the full bleed hero banner? OK. Apologies for being deliberately divisive, its not shown in the schematic above, there’s probably a place for it but please keep in mind that as we improve the sophistication of our website content marketing, we should be alive to the fact that if our hero spot isn’t well personalised - well, we’re relying on hope rather than design and we may be wasting an opportunity to capture interest in an ever-decreasing window of attention. Take a look at Amazon’s home page - it has surely the hardest working hero spot on the internet, no vanity artwork - a true best in show.
Train your AI
We’re crunching millions of customer clicks with one clear purpose in mind, to queue up the next thing we think our customer will want from us. In order to create great recommendations we have to help our AI understand our products well enough to accurately match them to our customers’ preferences. We have to teach the AI as much as we can about our business. And so, of course, we have to speak a little of the language of the algorithm. What makes sense to an algorithm? Attributes. Objective, quantifiable, descriptive attributes. And even if we don’t sell what would traditionally be considered a ‘product’ it's still useful to think about our service or value proposition as a catalogue of products - each with their own set of attributes. It may seem complicated, and it's no small task, but our products need to be logically arranged into nested tables of attributes so that our AI systems can scan, find and fetch them.
Lets look at a simple example of product attributes.
Every product belongs to a Category. Every product is assigned a number of descriptive Parameters, and each Parameter will have a Value.
The nature of our catalogue and the way we think about defining attributes will be different for each of us. A useful way to think about this, is as follows:
As customers, we tend to browse by Category, typically done using a website’s navigation menu.
Then we may decide to filter by a Parameter (e.g. by product types, or by colour, or by some other detail).
Finally we may choose to fine tune our search by sorting by Value, the most common being things like sort by price order, high to low, or sort by a particular colour range.
Of course, the way we decide to slice and dice our product catalogue will depend on the business we run - whatever that may be, its an important activity, it can’t be neglected because it’s fundamental to our personalisation strategy.
Stop thinking about customers in terms of marketing segments.
The old advertising adage, ‘if you don’t get it, it's probably not for you’ - refers to the idea that we each belong to certain demographics. Yet, because of the huge computing power AI gives us, it could be argued that that’s fast becoming obsolete thinking.
We can now think less about marketing segments and moreabout customers as individuals. After all, traditionally, the purpose of marketing segmentation is to help focus the effectiveness of our campaigns.
If we’re serious about our vision, where unique curated experiences appear like magic for every web site visitor, why do we need segments? In effect every customer becomes a ‘segment of one’ and, tantalisingly, we have the technology to treat them as such.
A ‘segment of one’ calls on three sets of data, behavioural, demographic and product, as shown in the diagram below.
To embrace the brave new world of beautiful personalisation, we — as teams of people committed to placing AI at the heart of customer experience — should begin thinking more like computer scientists, and, dare we suggest, more like algorithms.
If that's a dreadful thought, fear not. Like anything new, the learning curve is admittedly steep at first, but levels out fast. Let's look at it in its most simple form. To develop content strategies fit for the purpose of personalisation we must start thinking in terms of ‘if / then’ statements based on, and in conjunction with what we know about our customers. Looking at what our customers buy. Looking at the preferences they often show. On this basis we can start to write plain English algorithms and test them. For example:
If a customer shows an interest in our blue products.
Then next time they return to our website, let's promote our products that complement blue.
Let’s test that, see if that works. Let’s refine that, until it does work.
A lifetime of value awaits
Don’t make us think. Don’t make us work too hard, to work out how to do business together. Because, we just won’t bother.
In a fast and complicated world, isn’t it comforting to once in a while have our needs anticipated, met and perhaps exceeded? When it works, it's wonderful. We like it. It makes us feel special and we forget the fact that it's all just code and cookies underneath millions of pixels - because anything that feels personal, feels human.
Perhaps the reason more of us haven’t truly embraced personalisation yet is because the technology is still relatively new and to be perfectly honest, it is quite intimidating. So, a note of caution - a lifetime of value does indeed await, but certainly requires a steady, patient, rigourous attitude and a faith in gradual improvement. Let's avoid the mistake of trying to do too much too soon.
Get the basics in place. Visualise the end game. Organise our thoughts and off we go.
Comments