Build it, and they will come; how building a personalisation engine leads to transformative insights in every area of your business…
In the first blog of our three-part series we explore the topic of personalisation. We discuss why it matters, and crucially, why it’s more than simply delivering relevant content to consumers browsing a website.
Personalisation matters
Most businesses understand that consumer experience is the new competitive battleground to attract and retain customers. Simply offering the cheapest products is not enough if the customer experience is slow or frustrating. Businesses that make the customer journey easy, frictionless, rewarding and fun will reap the rewards of customer loyalty. Customers also have incredibly high expectations for their experience, something we all have Amazon to thank for raising the bar on.
Personalisation represents a huge opportunity to deliver compelling customer journeys, by highlighting relevant products, offers and services. It can imbue journeys with that sense of reward, wonder, fun and exploration, making you come back time and time again.
“Personalisation – the action of designing or producing something to meet someone’s individual requirements.”
Personalisation is an attempt to capture the differences in human taste and deservedly treat everyone as an individual. Our aim in effective personalisation is to create the most rich and vivid picture of the consumer, described in a variety of actionable ways. Our goal is to optimise every touch point we have with the customer and use insight about other customers “like them” to inform their experience.
The benefits for customers and businesses are numerous. Effective personalisation can influence all commercial levers, increasing sales, reducing costs, and improving margins. Study after study backs up why we should embrace effective personalisation.
For example, Accenture report that “91% of consumers are more likely to shop with brands who recognise them by name, remember their preferences, and provide them with relevant offers and recommendations.”
And McKinsey state that “our research shows that effective personalisation can increase store revenues by up to 30 percent, and several next-gen technologies can improve store productivity by at least 10 to 20 percent.”
So why isn’t everyone doing it already?
Most long-established businesses realise they need to do something to push forwards their efforts, but struggle to land personalisation effectively because it’s a multi-faceted challenge bringing together technology, data science, marketing, customer experience and merchandising.
Newer businesses have the advantage of building personalisation into the fabric of their organisation from its inception, creating that common language and implementing the necessary technology and data tool-set.
The challenge for many businesses can be leaving personalisation in a silo, potentially building a product recommendation engine, and calling it a job well done.
What if we told you that’s the tip of the iceberg?
We believe that personalisation can become a transformational force within a business, impacting merchandising, pricing, website design, customer experience, CRM, and supply chain.
How much benefit is being left on the table by not using this goldmine of customer intelligence to power insights in other business units?
Both large and small businesses should expand their definition of personalisation and look to other less obvious areas that personalisation can impact, to break down silos and build strong cross divisional connections.
Getting started: building a product recommendation engine
A great place to start along the personalisation journey is to build a product recommendation engine. It might feature as part of your website recommending products that complement something a customer has already put in their basket or, if you’re a known shopper with historic purchases, the recommendations can be based on the entire set of transactions they’ve had with you.
A technique called collaborative filtering can be used to profile each customer, and identify other customers who are “like” them in terms of their historic purchases, then make recommendations for products that are frequently bought by people like them.
This is now all around us in our daily lives. Say you’ve decided to start running to improve your health and take your newfound enthusiasm to Amazon to purchase some running shorts that will make all the difference in your running career.
Now once you’ve made that purchase Amazon starts to tell you that people who purchase those shorts, people like you, also buy a water bottle and some wireless headphones, and read “Born to Run”… before you know it you’ve bought a heart rate monitor! This is collaborative filtering in action.
The data you need to plumb into this recommendation engine might simply be your product level transactional database with a unique customer identifier or web browsing session data from your website and some information about the products.
It’s a simple concept but incredibly powerful. It can drive additional revenue from recommending complementary purchases, or drive inspiration for customers looking for something new and exciting that’s related to their interests. The customer wins and the business wins.
Learning from the masters
Netflix are the masters of personalisation, and it’s easy to see why they’ve placed such a high value on getting it right. As far back as 2016 they stated that “the combined effect of personalisation and recommendations save us more than $1bn per year.”
Their business model is reliant on customers being “sticky”. They need you to fall in love with the platform and have a friction-less and rewarding experience. The only way to do that is to understand you in great detail, tailor your experience in every possible way and reduce the overwhelming “endless aisle” of films, TV shows and documentaries that they could put in front of you.
Every interaction you have with one of Netflix’s “products”, be that watching Stranger Things, searching for Making a Murderer, or giving Tiger King a thumbs up, is filed away in their database.
Their farm of servers, the size of a small English village, whir away mining the millions of user interactions with their service to create meaningful relationships between customers and content and identify customers who look like others.
So, what aspects of their service are personalised? Well for starters each scrolling category bar, like “Films Based on Real Life” is selected from a set of genres that would likely appeal to “people like us”. Each title within that category would be handpicked to stand out and engage us. Using techniques such as collaborative filtering, Netflix can easily surface the next best show put in-front of you, for example because people like you watch Ozark.
Netflix have taken personalisation to the extreme and personalised the artwork for each title depending on which we’re most likely to engage with, perhaps based on lead character. They know which version of the front cover artwork to present you to entice you to watch Pulp Fiction, because they’ve modelled your affinity towards Uma Thurman and John Travolta…yes you read that correctly, you have a “propensity to like John Travolta” attribute in the Netflix database.
Now we can see how creating simple relationships between customers and products can lead to a whole raft of opportunities to create better experiences, but don’t stop there!
Think laterally to create more value from personalisation
Creating a product recommendation engine helps you understand an incredible amount about each product, each customer, how products relate to each other and how customers relate to each other. It creates a data goldmine!
This dataset might be the richest source of customer level information in your entire business, and it has far reaching applications. Let’s look at an example.
This article details how Airbnb set out to create a simple product recommendation engine to power a website section called “More places to stay”, which gives alternatives to the property you are viewing based on similar listings. Again, this is powered by collaborative filtering type machine learning techniques.
“Together, Search Ranking and Similar Listings drive 99% of our booking conversions.”
This was easy to implement and drove incredible value for Airbnb. They could have stopped there, but they didn’t. They used the similarity of different properties to create clusters or “demand aggregations” linking together “consideration sets” of properties that serve a distinct customer need, be that “High End Luxury Skiing Resort in Boreal” or “Budget Villa in Lagos Portugal”. The idea is that because they serve a customer need, these properties are competing against each other and need to be priced accordingly.
Property owners set their own prices on Airbnb, which can sometimes be out of line with competition, meaning a low chance of being occupied. As an owner the answer might be to lower your prices, but this trial and error approach takes time and never tells you the optimal price.
Lots of unoccupied properties or super cheap prices don’t work out for the owner or Airbnb, so their solution was to create guide prices for the owners balancing likelihood to be occupied and margins. These guide prices take into consideration the lead time to the customer’s preferred dates, but also the “consideration set” that the product falls into, automatically making the property pricing competitive with similar listings.
Airbnb didn’t let their goldmine of data sit in a silo and they reaped the rewards accordingly.
However, it takes some level of lateral thinking and creativity to see how the personalisation can transform other non-obvious areas of your business. At HyperFinity, we’re working hard to expand the definition of personalisation in the minds of businesses leaders.
In the next blog of this series we will take a deep dive into the language and building blocks of effective personalisation. See you next time…
Visit HyperFinity Get In Touch
Join the discussion 2 Comments