GUEST COMMENT Machine learning can address retail’s customer service failures

Recent research revealed that the retail sector received more customer complaints than any other industry in 2016. This poor level of customer service had a very real impact on the bottom lines of many business, costing retail firms more than £37bn, according to a report from Ombudsman Services.

Some online-only retailers may consider this more of an issue for their bricks-and-mortar cousins. But providing a quality ecommerce experience has never been more important.

Today’s online marketplace is incredibly crowded, and time-poor consumers won’t think twice about going elsewhere if their experience isn’t up to scratch, let alone if it causes them to complain. Where online does have the upper hand, however, is being able to harness huge amounts of customer data to not only react to customer behaviour in real time, but also preempt issues and complaints.

Machine learning isn’t just chatbots

Data alone isn’t useful. Being able to analyse and derive useful meaning from it automatically and within seconds, using technology like artificial intelligence, is what makes data a game changer. But while a good chunk of the headlines regarding the use of AI in retail focus on the “sexy” application of this technology – everything from chatbots to driverless delivery vehicles – it’s the potential impact on the customer’s ‘everyday’ engagement with a brand which is really exciting.

Machine learning, a form of AI, can identify opportunities that may not have been noticed by the human eye and segment these customers so that you can adapt how you engage with them and be more personalised and relevant.

But machine learning’s potential doesn’t stop at creating experiences which turn browsers into buyers. It also provides the opportunity to identify niggling issues and address them well before they become customer service nightmares.

Remove the fog

Our own research found that confusion around product choice, available sizes and unclear pricing all featured in the top five complaints made by customers of online fashion retailers. This reflects a broad frustration concerning a lack of transparency, which annoyed online shoppers and overcomplicated or obstructed their path to purchase.

Of course there are some quick fixes such as clearer price displays or sizing suggestions based on previous purchase history, but machine learning can take retailers a step further. Through advanced segmentation you can learn more about what your customers are doing and feeling, who they are, what they want, and tailor their online experience accordingly.

For example, you could uncover a large segment of British expats buying items in France, separate them from native French speakers, and ensure they get prices converted to pounds, UK sizing and offers which reflect their preferences. This eliminates the unnecessary time they would normally spend browsing through stock targeted more towards French tastes and ensure potential frustrations are avoided.

Machine learning can also highlight customers who abandon baskets due to hidden costs, such as shipping charges. This allows these customers to be targeted with specific offers, whether that’s reserving free delivery for shoppers in China or a 10% discount for returning consumers.

Listen to your customer

Although retailers can harvest a wealth of customer data from browsing history and previous shopping behaviour, we also shouldn’t forget the importance of actively collecting feedback. Extensive forms aren’t necessary, though – a quick comment box will suffice – but even then you may be faced with wading through high volumes of suggestions (as well as complaints).

Employing machine learning algorithms to process feedback will allow you to draw quick conclusions and trends from the data so you can start making improvements right away.

For example, an abnormally large amount of feedback about slow loading times could point to a need to quickly improve your website efficiency. Alternatively, if many are complaining that they’re unable to find a particular feature, you may want to reconsider your web design. And both the analysis and suggested improvements can be flagged to you in real time.

Many may see feedback boxes as old hat, but combining them with the latest technology allows retailers to continuously adapt to changing customer expectations without the need for an enormous team to analyse and react to all the complaints and suggestions you receive. It’s also possible to create actionable segments based on customer feedback. For example, individual customers may tell you that they favour jackets over shirts – with machine learning, this insight can be actioned immediately and these shoppers can be placed into their own segments.

Investment in machine learning is expected to boom over the next three years – marketers are due to invest $2bn by 2020 according to a recent report we authored with IDC. But it’s important to understand that machine learning isn’t a panacea that instantly solves longstanding customer service problems.

The businesses that will triumph will be those that continue to have their customers at the core of their business, are willing to adapt for them and are already obsessed with delivering a superb service. Machine learning can simply supercharge this strategy online, ensuring complaints are kept to a minimum and sales continue to soar. It’s set to become a crucial tool – one that our bricks and mortar cousins can only dream of having.

Geri Tuneva is head of marketing, EMEA at Qubit