An interview with Joe Browns founder, Simon Brown in which he discussed their journey as a business and views of the market today.
Personalisation is an industry buzz-term at the moment (you wouldn’t be seeing this article if it wasn’t), and there isn’t a retailer or B2B online seller out there who hasn’t been regaled with stories of how implementing the magic of Machine Learning will instantly catapult their business into the sales stratosphere.
Indeed, Dynamic Yield have recently released the results of their research, which indicates that a huge majority (92%) of marketers and executives see the value of personalisation in the digital environment.
It may be some surprise then, to discover that 55% of people surveyed by Researchscape felt that marketers weren’t getting personalisation right, and that on average, personalisation is only adding around 10% to the effectiveness of conversion rates compared to video which can add up to a whopping 80%.
Personalisation is very much seen by some retailers to be a magic bullet – something to fire and forget whilst the machines optimise shopping recommendations for customers and increase conversion rates through complex algorithms and “black box” magic.
The reality is that introducing personalisation into your digital team shouldn’t decrease the amount of manual merchandising that happens on site; it should just be targeted in different areas and based upon intelligent insight into your customers’ behaviour and requirements.
Knowing When To Use Personalisation
The key for many retailers is to understand the nuances of when and where to use personalisation, the degree to which personalisation should be used to target an individual, and when it should use the behaviour of crowds to influence buyers.
For example, when working with a well-known electrical retailer, I saw merchandisers desperate to identify what type of goods (washing machine, TV, coffee maker) a visitor was looking for, and to target them exclusively based on the first one or two items they viewed, even in subsequent visits to the site.
This is a dangerous strategy, as there is every chance that your potential customer has already purchased the item they were looking at the last time they visited, or they may be investigating peripheral items to support their purchase.
Beware of jumping to conclusions based on a few scant points of behavioural data, and don’t personalise to the detriment of everything else your business can provide – people still want choice, and they don’t want to think they’re being forced into a decision online any more than they do when they interact with you in store.
Even the biggest retailers get personalisation wrong some of the time. Take Twitter for example – as a fan of Tottenham Hotspur, I follow their official account on Twitter. However, Twitter constantly recommends that I follow the official Chelsea, West Ham and even Arsenal account, as they are “similar accounts I might like”, when the complete opposite is true – they are in fact the least likely of any accounts on the social media platform that I would follow.
This is a great example of machine learning and generic behaviour rules failing to understand the nuances of this particular customer/product relationship – I will choose to follow this “product” (my football club) in complete exclusion of any others, and in fact recommending an alternative isn’t just an unsuccessful targeting attempt, but could turn me away from the platform provider altogether, should I be of a sensitive nature about this kind of thing!
The secret to effective personalisation is to create a perfect balance of relevance and flexibility. Be ready for your customer to completely change their behaviour or requirement on a whim (consider how many people shop in a large shopping centre, moving from one store to another before completing their purchases – often for items they didn’t come in for).
Always understand that there are external influences you cannot measure – Are there two people browsing together? Does the customer have multiple tabs open? Is the customer buying for a friend/relative when they usually buy for themselves?
You can (and should) control some of these factors – for example “buying for a friend” has permanently, and negatively, impacted my experience with a number of online retailers. All it takes is the purchase of one birthday present for a 5 year-old niece and suddenly all my recommendations are for Peppa Pig themed toys, princesses and Lego (OK, the last one might be relevant, but that’s down to luck rather than judgement!).
However, I clearly indicated when buying the item that it was a gift (I had it wrapped, and sent to an alternative address), so why wasn’t this information used when deciding what products to recommend to me on subsequent visits, or at least discounted from the factors used when generating my recommendations?
As a retailer, it is very important to understand the impact of inappropriate recommendations – at best they are a distraction, but at worst they can cause offence or upset.
I have seen one of the major online pharmacies using product recommendations and “people who bought this also bought” rules on products where recommendations really shouldn’t appear – just because one or two of your customers purchased haemorrhoid cream with their aftershave doesn’t mean that everyone else wants to buy those products – and the fragrance brand definitely doesn’t want its potential customers to associate the two things.
Work out how to set up exclusions, stop recommending stuff that you know isn’t relevant (or is totally counter-intuitive) and above all, merchandise your products, using personalisation as the additional cherry on the cake, as opposed to the foundation of your merchandising. Machines are dumb, people are smart – make the most of that fact.
Key points for effective personalisation
- Establish a balance between relevance and flexibility
- Factor in customers changing their behaviour/requirements.
- Remember that there are external influences that you cannot measure.
- Reduce inappropriate recommendations
- Use personalisation as a tool in a holistic endeavour in providing omnichannel excellence, not as a magic fix.
In summary, personalisation is a hugely powerful tool, but it can also be a blunt one. There is no substitute for the human factor in many scenarios. Whilst machine learning can be great for managing the long tail, or for differentiating similar products, relying on it wholesale (or at the very least not understanding where it needs to be supported by human input) can have a negative impact on customer experience and on conversion.
This article was originally published on IMRG, March 2018: https://www.imrg.org/blog/reality-of-personalisation-to-online-retail/
When should you discount? We reveal what to avoid and the key strategies for discounting success.
Download our easy-read checklist to help you decide whether you’re ready to take your online business to the next level.