A few years ago, simple personalisation was enough to create an outstanding customer experience.
A small tweak here and there, the customer’s name at the top of your marketing emails, a curated homepage that shows products they’ve viewed recently.
But as we all know, years of increasingly personalised interactions have raised the bar for customer communications. Today, customers are looking for something more: an experience that’s tailored to their individual needs and preferences, instead of the needs of their customer segment.
But as it turns out doing real, authentic, impactful 1:1 personalisation at scale is really hard. It requires very granular insights, crystal clear strategy, the ability to analyse data at scale – and, crucially, the customer understanding needed to act on the insights that data analysis delivers.
As we’ll see below, AI is a powerful tool to tackle challenges like this. But first, let’s dive deeper on hyper-personalisation itself.
What is hyper personalisation?
Where ‘personalisation’ is typically focused on targeting communications to customer micro-segments, ‘hyper-personalisation’ is the art of tailoring interactions to the needs of individual customers.
For the customer, it’s a bit like having a personal shopping assistant. They know what each customer is likely to be looking for, when they’re going to be interested in splashing out and which hidden gems they will love (even if they haven’t seen them yet).
You know what your customers want – sometimes even before they do
When you predict what customers want to see online and when they want to see it, you can start to make the kinds of careful tweaks to their shopping experience that will delight them. By using historical data on everything from website interactions to sales data, marketing teams can predict what individual customers will want to see at certain times, on certain channels.
Take Levi’s: they’ve just added a handy new tool named Grid Sort to their website. By the time a visitor lands on a product page, Grid Sort has already sifted through all the real-time data about the visitor to work out what kind of products they’ll be interested in. The visitor gets shown the clothing they’ll probably end up browsing for – before they’ve had to go through dozens of denim jeans.
Powered by AI, Grid Sort can read and understand images, predict the order which will be more interesting to the visitor and inform shopping decisions for the customer. This not only makes the customer more likely to convert there and then, but it also makes for a quicker, easier shopping experience that might just translate to brand loyalty.
Build a true understanding of your customers
With hyper-personalisation, a business must take its customers’ interactions over time into consideration. They’ll know that you’re more likely to prefer a message on your mobile during the week but like to browse a physical brochure at the weekend. They’ll know that in the summer, you’ll spend more on swimwear except for two weeks over August when you’re likely to be shopping for jewellery for your Mum’s birthday. And they know that you’re a sucker for a well-placed discount email, but only tend to redeem them in the first couple of months after Christmas.
This nuanced understanding of the customer, spanning data types, channels and time periods, is the essence of hyper-personalisation.
In fact, that’s exactly what we did for one of America’s leading fashion retailers. We pinpointed where their marketing was influencing demand and helped them drive higher sales in areas of greater opportunity. We ranked the customers more likely to convert from catalogues, which meant their marketing team could reduce spending on catalogues to customers that weren’t likely to convert and focus their attention on sending out more personalised catalogues to the customers that would.
AI and hyper personalisation
AI solutions are the perfect fit for hyper-personalising large-scale marketing campaigns for several reasons:
• AI can analyse and automate at a scale that’s impossible for human beings.
• By drawing links between thousands of different data points, AI can uncover patterns and relationships that simple rules-based approaches cannot.
• AI can learn the recommendations, content and messaging that are optimal for each individual customer, removing guesswork from the equation.
How can AI be used for hyper personalisation?
Artificial intelligence, combined with automation and data engineering, gives marketers the ability to provide truly hyper-personalised service to customers: every experience an individual has with the brand can be adjusted and optimised to fit their preferences.
What does good hyper-personalisation AI look like?
Marketers have long recognised that AI is a powerful personalisation tool – companies from Netflix (one of the first companies to offer a high-degree of personalisation in media) to Amazon (who revolutionised recommendations with “customers who bought X have also bought Y” feature) have been using AI to tailor what their customers see for years.
But a few years of exposure to clunky personalisation – from poor product recommendations to insistent chatbots and overly-invasive messaging – has put customers on their guard. As personalisation becomes hyper-personalisation, the stakes are continuing to rise.
So how do you use AI to hyper-personalise experiences for customers without alienating them?
Hyper-personalisation enhances the customer experience
It’s all-too-common for marketers to look at customer behaviour patterns in channel silos: their email cadence is determined by the brand’s email software, the webpage they’re served is decided by the brand’s website algorithm and their customer service experience by recommendations from their CRM system. This is because retailers are often organised this way internally – it’s what we call Conway’s Law.
But humans don’t make decisions in silos. The way they interact across different channels reveals important information about their behaviour and can help to inform their likely customer lifetime value. Making decisions about personalisation in silos means missing out on large parts of the picture.
Marketers that do hyper-personalisation well use customer data to build a comprehensive view of their needs and behaviours. They set up a programmatic approach to personalisation whereby every interaction, on every channel is informed by analysis of all available data that is relevant to that customer and their experience. Bridging these gaps between siloed data to achieve the elusive omnichannel marketing experience requires cross-team data integration and a sturdy platform that unifies the data.
This approach improves customer experience by:
Driving consistency – Data moves freely back and forth between the customer-facing channels and the brand’s AI system, so the interactions on every channel are informed by the same detailed picture of the customer.
Personalising for each individual – Customers aren’t just seen as email-openers or link-clickers. Instead, machine learning can be used to spot patterns across all available customer and channel data, helping to understand individual customer’s needs, preferences, and quirks, which in turn can be used to tailor every interaction and improve their experience.
How do you hyper-personalise with ever-changing customer preferences?
Hyper-personalisation is all about building strong long-term customer relationships, so AI-powered solutions need to be designed and configured to work effectively in the long term instead of over-prioritising short-term gains.
Say a brand’s AI identifies that a customer is more likely to buy if they receive an email on a Tuesday, so it sends product recommendations or new offers to that customer every Tuesday. This works for a month or so, before the customer starts to get tired of being constantly nudged towards the digital sale rail.
It turns out, they in fact are more likely to respond on a Tuesday, but only if they haven’t received an email for a couple of weeks. We typically turn to historical data to fuel AI. But in this case, the past isn’t too helpful at telling us how the customer’s buying behaviour will evolve.
An AI solution needs to be able to learn these niche preferences, or that the short term revenue increase will quickly turn into long term damage to the customer relationship.
Running controlled experiments at scale is key to understanding and optimising customer behaviour
No AI can expect to understand and anticipate these kinds of quirks and intricacies immediately, but it’s vital that marketers use AI solutions that can learn from dynamic customer data. To do hyper-personalisation well, marketers must regularly run small, controlled experiments to understand customer preferences, and use these results to retrain AI solutions with the updated data to inform the next best action in the customer journey.
This, however, comes in-tow with a new challenge: incrementality. How can we measure whether the changes presented by the experiments wouldn’t have happened regardless? How can we be sure it’s worth investing in paid search if organic traffic could’ve been directing customers to sales, instead?
AI can provide context to each consumer that interacts with your brand by directly modelling the incremental impact of a specific marketing action on an individual.
Without experimentation, models will learn biased views of customer behaviour. Running controlled experiments at scale is how marketers can optimise behaviour and measure the performance of those models objectively.
The quality and quantity of your customer data is crucial for effective hyper-personalisation
Hyper-personalisation is a powerful approach for optimising marketing communications, deepening customer relationships, and ultimately boosting ROI, but it does require large amounts of customer data to run effectively.
Marketers need to recognise this as they build their hyper-personalisation strategy: to deliver a more personalised customer experience, you need to capture more, high quality information about them, their intentions, and their needs. This can be done explicitly (e.g. “click on the products you like”) or implicitly (e.g. observing dwell-times, counting add-to-baskets, etc.). As Merkle’s 2021 Consumer Experience Sentiment Report shows, many customers are willing to share their data with companies, but only in exchange for improved customer experience.
Integrating AI with operational systems across the business, not just marketing and sales channels, is hugely beneficial. Using the full suite of your customer data, you can break out of greedy optimisation for clicks and instead focus on providing the best customer experience.
Unlock hyper-personalisation at scale with Faculty Customer Intelligence
We cannot whittle down customer interactions to a 1-on-1 basis. But with so many opportunities for AI to help customise the customer experience, we can build systems that leverage the technology to reach the lowest level of personalisation possible for use across a range of use cases. Faculty Customer Intelligence is designed to help marketers build a true understanding of their customers – including detailed pictures of when, where and how they interact with a brand.
Uses internal and external data across touchpoints ensuring that personalisation is driven by all you know about a given customer’s preferences
Targeted approach for control trials and experimentation allows you to learn by trialling different interventions and strategies at minimal opportunity cost
Causal AI algorithms that outperform all open source models learn the incremental impact of interventions, enabling targeting of investment to where it’s most needed
Explainable, transparent and trustworthy AI solutions give users explanations as to what is driving the system’s recommendations
Customer Intelligence brings together customer data from every touchpoint, helping you predict how individual customers will respond to marketing so you can build hyper-personalised campaigns that span the entire customer journey.
Find out how Customer Intelligence can help you build truly personal, authentic experiences with your customers.