Automation can save time when it comes to personalisation, but as brands such as Adidas and Pinterest have found, it can backfire in embarrassing ways. Mike Austin, CEO & co-founder, Fresh Relevance explores the significance of human oversight when running personalisation at scale and the importance of thinking outside of the box as to what could go wrong with personalisation initiatives at their outset.
Most marketers recognise the critical need to personalise their marketing in order to resonate with their customers. If done manually, this can be a tedious – probably even impossible – task. This is where automation comes in. It allows brands to deliver tailored campaigns at scale by analysing behavioural customer data and delivering the content that is most likely to resonate with each individual.
Automated, personalised campaigns eliminate a lot of repetitive and time-consuming manual tasks and help marketers deliver ROI in a predictable way. But automation has its pitfalls if approached with a set-and-forget mentality by the marketer. It requires human oversight to ensure that the intricate marketing machine ticking away in the background is able to deliver the company’s goals.
Pinterest’s wedding gaffe is a point in case. The social network sent users, who had pinned wedding content to their boards, an email congratulating them for tying the knot and recommending related pins and boards. Pinterest’s automated marketing had picked up on the viewed wedding content and produced ‘personalised’ marketing messaging to match. However, just because a user curates a wedding-themed board, this doesn’t mean that they’re getting married soon.
Of course, this doesn’t mean that automation has no place in marketing — it definitely does. In order to prevent such personalisation blunders, however, marketers need to consider some of the key scenarios and fine-tune campaigns if needed.
Basing marketing content on behavioural data, such as browsing and purchasing information, allows brands to engage with their customers based on their demonstrated interests. Yet, marketers need to be careful around gift-giving occasions, such as Christmas or Valentine’s Day, when a strong focus on past behaviour can lead to inaccurate personalisation.
Take the example of a thirty-something shopping for a Christmas gift for his mother. His historic browse and purchase data might be stereotypically masculine and focused on grooming products. If he suddenly starts browsing women’s perfume in early December, this is a strong indicator that he’s gift shopping. Yet, as personalisation engines tend to personalise based on the shopper’s past purchases and behaviour, marketers will likely miss the opportunity to recommend products for gifts they may be searching for.
Therefore, product recommendations should be adapted before seasonal events and provide inspirational, crowd-based suggestions, such as trending products or bestsellers, to help shoppers find that perfect gift. And after the holiday, product recommendations should again focus on longer-term behavioural data to avoid suggesting items based on gift purchases
In today’s data-rich age, marketers have access to an unprecedented amount of consumer information as well as the tools to analyse it and act on it. This opens up innumerable opportunities for personalisation. However, just because marketers can personalise almost everything doesn’t mean that they should. It’s important to strike the right balance, as showing that a brand tracks and analyses a customer’s every website interaction, purchase, etc. can come across as stalkerish and creepy if approached too aggressively.
Target is a classic example why the sensitive use of behavioural information is important. The US discount retailer identified purchase patterns for a number of products, such as cotton wool and unscented lotion, that strongly indicated that the customer was expecting a baby and even gave clues when the baby was likely to be due. However, given the sensitive topic, it made customers feel uncomfortable when the retailer proactively sent them targeted vouchers for baby products. Here, it can make sense to use behavioural information in more subtle ways, for example by mixing personalised offers with social proof, such as best-rated or trending products, in order to be helpful without being invasive.
Many marketers segment customers based on their purchasing behaviour, but often forget to factor in returns. This can paint an incomplete and even misleading picture of your customers. For example, if a shopper purchases a floral print shirt, this could be indicative of them liking this style of clothing. On the back of this information, the product recommendations engine may well start suggesting all kinds of flowery garments. But if the customer returned the item because they didn’t like it, these recommendations are unlikely to resonate and may even annoy the customer. Therefore, it is vital to update a customer’s purchase history with the items they returned to ensure subsequent marketing is based on relevant information.
As ASOS’s updated returns policy highlighted earlier this year, high-volume buyers are not necessarily your most profitable customers — if they are actually serial returners, they may even be a serious threat for your profit margin.
Personalisation is an incredibly useful tactic for marketers. By understanding what the customer likes and needs, it helps drive sales and long-term customer loyalty. However, as with all forms of automation, marketers cannot afford to take a completely hands-off approach. Human oversight is needed to ensure that brands send the right message to the right customer, at every single interaction.
By Mike Austin
CEO & co-founder