Over the years I’ve been in ecommerce there have been many themes that for a period of time have dominated the digital airwaves. Big data, internet of things, apps, the single view of the customer, cross-channel/multichannel/omnichannel/no channel (delete where appropriate) and, most recently (and scarily), the metaverse have all had their time in the sun. Whilst some of these remain relevant, none have had the continuous presence as personalisation. For years now, it’s been a mainstay of ecommerce experience and marketing and there are literally thousands of tech providers promising to deliver it to your customers and achieve great results.
To define personalisation, it’s broadly the presentation of relevant products, services or content to a prospective customer based on data that has been collected about that individual. When I started in retail you basically had one offer and one message for everybody. Almost nothing was tailored and it was a case of ‘take it or leave it’. Now there’s almost infinite flexibility. With this breadth it’s tempting to try to do everything, but while some things are definitely worth doing, others aren’t.
So, what are the factors that determine why it works in some circumstances but not in others? I believe there are three principal dimensions that determine whether and how personalisation is going to be effective.
Intent
Building an understanding of customer intent is truly the holy grail of retail. If you understand what a customer wants, it can help drive pretty much every aspect of your interaction with them. There are two principles that I find helpful when thinking of intent, namely drive and specificity.
Drive describes the level of customer intent. Is the customer determined to buy or relaxed about it (need vs want)? Are they in a hurry or do they have time to browse (now vs whenever)? Have they decided to buy from you or are they shopping around (you vs whoever)? How far have they gone in the buying process, e.g. looking at delivery options, adding to basket?
Specificity is related but has its own characteristics. Does the customer know what they’re looking for or are they looking for ideas or inspiration? Have they got all the information they need or do they need more to proceed? Have they searched for something generic, eg. ‘walking shoes’ or for something very specific eg. ‘Salomon X Ultra GTX’.
Clearly, the more drive and the more specificity you can glean, the better and the more valuable that consumer is to you and the more specific your personalisation should be. The more that you have to assume, eg. based on historic buying behaviour or looking at profile characteristics, the less valuable and the less you should tailor.
Preferences
The idea is that the better you understand what someone does and doesn’t like, the better the recommendations that you present to them. It makes a load of sense, but it’s an area where the claims of technologists have always bothered me. At heart it’s because I don’t like the idea that my behaviour can be predicted by algorithms – surely I’m an independent-minded and spontaneous individual who cannot be tied down by AI! Well, clearly there’s something in between the two extremes and it may be instructive to think through where it is.
Looking at product categories may be helpful. There are some categories where preferences are integral to the purchase process, others where they’re less integral but still relevant and some where they’re incidental or irrelevant.
Let’s take grocery shopping first. This is an area where preferences are integral and where observed buying is a pretty good guide to predicted buying. It’s no surprise that CRM, big data and hyper personalisation emerged from this sector. Toiletries and cosmetics also fit this model of high repeat purchase and relevant brand or product preferences.
Fashion strikes me as being somewhere in the middle. The key preference here seems to be a brand preference which dictates the brands I choose to shop with. If you can present me with relevant alternative brands before I’ve started shopping perhaps that might tempt me to explore. This marketing arena seems the best application for personalisation. Once I’ve started my purchase journey I’m not sure there’s much value, other than reflecting my gender or remembering that I only really ever buy dull-coloured polo shirts (a scurrilous myth perpetuated by my wife).
At the incidental end is perhaps my most recent sector, furniture. For starters, frequency of purchase is low and repeat purchase almost unheard of, so it’s hard to make predictions on a customer’s next purchase. Added to which, it’s difficult to gauge preferences. What someone deems important in a sofa they might see as irrelevant in a dining table or pendant light. Perhaps at Heal’s we were missing a trick, but we really struggled to get much out of preference-based personalisation. Instead we focused on intent-based programmes.
Whatever sector you’re in, have a think about whether customer preferences are integral, relevant or incidental to purchases. And of course, whether you have enough data points to glean these preferences. If you assume too much it may do more harm than good.
Channel
The final dimension to consider is channel. Personalisation tech can be applied at some level in every aspect of digital marketing and web experience, but some will be more valuable than others.
All customer acquisition activity benefits from personalisation or at least profiling. Being able to target the right message to the right person at the right time is one of the beauties of digital marketing. So this is a no-brainer. Another no-brainer is where a high level of intent has been shown. In these cases you’d be foolish not to follow up with personalised messages.
Where you choose to apply across other areas of customer experience and how much effort and resource you put into each is worth thinking about. Let’s take email as an example. If you’re a business for whom preferences are integral then you really should be personalising emails, or at the very least segmenting them. If you’re not, it may not be worth the effort. Similarly with onsite experiences. You may be better off using standard algorithms to show alternative or complementary products, rather than imagining you can personalise them to what you assume appeals to the specific customer’s preferences.
Having looked at technology and its applications, I’ll finish with a word on what we found to be the most effective personalisation tool at Heal’s. It turned out to be good old-fashioned person-to-person interaction. After all, what could be more personalised than actually talking to a real person about your needs, your preferences and your questions. Our online chat feature, which connects customers to in-store team members, delivered more than 5% of total company sales in 2021. Our telesales operation (basically one expert person at the end of a phone line) was even bigger. It just goes to show that tech is important, but it ain’t everything.
David Kohn is a retail veteran whose experience most recently includes customer and ecommerce director at Heal’s, and previously commercial and digital roles at WHSmith, Waterstones, Borders and Snow+Rock Group. Now acting as an independent advisor, David remains on the look-out for interesting and effective innovations.
You are in: Home » Guest Comment » GUEST COLUMN Personalisation – a personal perspective
GUEST COLUMN Personalisation – a personal perspective
David Kohn
Over the years I’ve been in ecommerce there have been many themes that for a period of time have dominated the digital airwaves. Big data, internet of things, apps, the single view of the customer, cross-channel/multichannel/omnichannel/no channel (delete where appropriate) and, most recently (and scarily), the metaverse have all had their time in the sun. Whilst some of these remain relevant, none have had the continuous presence as personalisation. For years now, it’s been a mainstay of ecommerce experience and marketing and there are literally thousands of tech providers promising to deliver it to your customers and achieve great results.
To define personalisation, it’s broadly the presentation of relevant products, services or content to a prospective customer based on data that has been collected about that individual. When I started in retail you basically had one offer and one message for everybody. Almost nothing was tailored and it was a case of ‘take it or leave it’. Now there’s almost infinite flexibility. With this breadth it’s tempting to try to do everything, but while some things are definitely worth doing, others aren’t.
So, what are the factors that determine why it works in some circumstances but not in others? I believe there are three principal dimensions that determine whether and how personalisation is going to be effective.
Intent
Building an understanding of customer intent is truly the holy grail of retail. If you understand what a customer wants, it can help drive pretty much every aspect of your interaction with them. There are two principles that I find helpful when thinking of intent, namely drive and specificity.
Drive describes the level of customer intent. Is the customer determined to buy or relaxed about it (need vs want)? Are they in a hurry or do they have time to browse (now vs whenever)? Have they decided to buy from you or are they shopping around (you vs whoever)? How far have they gone in the buying process, e.g. looking at delivery options, adding to basket?
Specificity is related but has its own characteristics. Does the customer know what they’re looking for or are they looking for ideas or inspiration? Have they got all the information they need or do they need more to proceed? Have they searched for something generic, eg. ‘walking shoes’ or for something very specific eg. ‘Salomon X Ultra GTX’.
Clearly, the more drive and the more specificity you can glean, the better and the more valuable that consumer is to you and the more specific your personalisation should be. The more that you have to assume, eg. based on historic buying behaviour or looking at profile characteristics, the less valuable and the less you should tailor.
Preferences
The idea is that the better you understand what someone does and doesn’t like, the better the recommendations that you present to them. It makes a load of sense, but it’s an area where the claims of technologists have always bothered me. At heart it’s because I don’t like the idea that my behaviour can be predicted by algorithms – surely I’m an independent-minded and spontaneous individual who cannot be tied down by AI! Well, clearly there’s something in between the two extremes and it may be instructive to think through where it is.
Looking at product categories may be helpful. There are some categories where preferences are integral to the purchase process, others where they’re less integral but still relevant and some where they’re incidental or irrelevant.
Let’s take grocery shopping first. This is an area where preferences are integral and where observed buying is a pretty good guide to predicted buying. It’s no surprise that CRM, big data and hyper personalisation emerged from this sector. Toiletries and cosmetics also fit this model of high repeat purchase and relevant brand or product preferences.
Fashion strikes me as being somewhere in the middle. The key preference here seems to be a brand preference which dictates the brands I choose to shop with. If you can present me with relevant alternative brands before I’ve started shopping perhaps that might tempt me to explore. This marketing arena seems the best application for personalisation. Once I’ve started my purchase journey I’m not sure there’s much value, other than reflecting my gender or remembering that I only really ever buy dull-coloured polo shirts (a scurrilous myth perpetuated by my wife).
At the incidental end is perhaps my most recent sector, furniture. For starters, frequency of purchase is low and repeat purchase almost unheard of, so it’s hard to make predictions on a customer’s next purchase. Added to which, it’s difficult to gauge preferences. What someone deems important in a sofa they might see as irrelevant in a dining table or pendant light. Perhaps at Heal’s we were missing a trick, but we really struggled to get much out of preference-based personalisation. Instead we focused on intent-based programmes.
Whatever sector you’re in, have a think about whether customer preferences are integral, relevant or incidental to purchases. And of course, whether you have enough data points to glean these preferences. If you assume too much it may do more harm than good.
Channel
The final dimension to consider is channel. Personalisation tech can be applied at some level in every aspect of digital marketing and web experience, but some will be more valuable than others.
All customer acquisition activity benefits from personalisation or at least profiling. Being able to target the right message to the right person at the right time is one of the beauties of digital marketing. So this is a no-brainer. Another no-brainer is where a high level of intent has been shown. In these cases you’d be foolish not to follow up with personalised messages.
Where you choose to apply across other areas of customer experience and how much effort and resource you put into each is worth thinking about. Let’s take email as an example. If you’re a business for whom preferences are integral then you really should be personalising emails, or at the very least segmenting them. If you’re not, it may not be worth the effort. Similarly with onsite experiences. You may be better off using standard algorithms to show alternative or complementary products, rather than imagining you can personalise them to what you assume appeals to the specific customer’s preferences.
Having looked at technology and its applications, I’ll finish with a word on what we found to be the most effective personalisation tool at Heal’s. It turned out to be good old-fashioned person-to-person interaction. After all, what could be more personalised than actually talking to a real person about your needs, your preferences and your questions. Our online chat feature, which connects customers to in-store team members, delivered more than 5% of total company sales in 2021. Our telesales operation (basically one expert person at the end of a phone line) was even bigger. It just goes to show that tech is important, but it ain’t everything.
David Kohn is a retail veteran whose experience most recently includes customer and ecommerce director at Heal’s, and previously commercial and digital roles at WHSmith, Waterstones, Borders and Snow+Rock Group. Now acting as an independent advisor, David remains on the look-out for interesting and effective innovations.
Read More
You may also like
Register for Newsletter
Receive 3 newsletters per week
Gain access to all Top500 research
Personalise your experience on IR.net