The advent of Big Data has supposedly enabled any brand out there to boil their data lakes in an attempt to understand how to provide a one-one experience with their customers.
Or so it goes in theory.
To do this properly is actually rather difficult, potentially involves large amounts of digital information and is computationally intensive. But stuff like this is happening and I regard it as ‘proper’ Big Data. Not just a bit of data mining and segmentation.
The intent of course is to sell more stuff but in a way which improves the customer experience. Let’s not confuse this with Analytics, which is giving insight to the business.
It led me to think… is this the future of personalisation? It’s barely even happening yet but is this as good as it gets? Multiple, siloed copies of my interests and preferences?
Of course not.
There are two key issues with the Big Data Personalisation model kicking around today.
- Every brand is building its own version of my personalisation profile.
- No amount of processing of my interactions is ever going to determine what my true intent is now.
The first issue has three solutions; I own it, you own it or we own it. I’ll come back to that.
The second one is a little more insidious.
Sometimes personalisation works really well with human beings. Those trusted people who know you and what you like. The man in your local corner shop for example. Before the internet I used to know this lady called Nikki at a travel shop, she knew my preferences for skiing holidays.
When the human experience works well it works really well but is still flawed. Once the individuals who know you move on you’re back to the beginning.
Similarly, my tailor knows my exact size for shirts but that isn’t much help when I’m not having them made by him.
So even human interactions suffer gaps, which brings us back to that second issue.
Your corner shop man - “I’ve got this lovely new bread which you’ll love”.
Sorry, I just went Gluten-free. Oops.
How do we overcome this today in the human world?
Easy - we have a conversation and re-set expectation and updated intent in that person’s mind.
True laser-guided personalisation is always going to involve me providing some up to date information. Otherwise anything I get back is out of date, an inferred guess, or both.
So what’s the answer?
Well interestingly this is where going forward technology can potentially improve the experience beyond even human interactions, bar a life-long personal assistant on your payroll.
First off, I want to own my profile that is an accurate and up to date reflection of me. Not 400 versions which are all partially flawed, just one.
And this should exist as a complex but single entity, perhaps in the cloud. Somewhere that I can maintain it and others can interrogate it.
Whilst my online digital footprint and buying history can inform it, ideally I would interact with this by voice. Enter next generation Siri, Cortana and perhaps more interestingly Alexa, when she arrives the UK courtesy of Amazon.
Over time I create a mirror of me.
A single, learning agent acting on my behalf that knows my wardrobe, likes and dislikes, birthday calendar, how my house is furnished. Everything to inform my personal shopping experience, except personal data which I would rather not share.
Think ‘Through the Keyhole’ - you can know a lot to help personalise suggestions without knowing exactly who that person is.
I maintain my profile and intent simply having a conversation with my assistant. I keep it up to date. I tell it my likes and dislikes. It knows what I bought. It knows when i’m buying presents. when I want something I can let it loose on the available options out there. I’m in the market for something. Give me options. Totally turning the model upside down.
The more I invest in this the better it gets and the more I get out of it.
I can allow brands to interrogate my profile (if I choose) to target *really* relevant offers to me and my assistant can act as the gatekeeper for that.
Perhaps they even pay me for the privilege.
Sure, by all means boil the ocean if you must to work out if something is interesting to me but don’t waste all those CPU cycles trying to work out who I am first. I can tell you that, really really accurately.
It feels like this model is coming someday.
And who knows. It might actually work.