Three ways we are helping brands harness the power of AI in Customer Experience innovation

06 October 2023

There may not yet have been a ‘big bang’ AI revolution, but its potential to revolutionise the way brands engage with customers is undeniable. 

For now, however, many businesses are still working out what to do with it. Still more remain fearful of its potential to disrupt - not just businesses and business models, but entire markets. In fact, while 73% of businesses understand Gen AI and 29% are using it in some way, more than half (52%) expect it to have a ‘meaningful, tangible, disruptive impact on their businesses over the next five years.’[1]

There is some basis for that fear if only the sheer pace at which AI technologies are developing. For instance, patenting activity in Gen AI – always a bellwether for innovation - has accelerated hugely in recent years. From a base of 1,450 patents five years ago, we now stand at more than 6000, a near five-fold increase.[1]

Small wonder then, that so many brands are looking closely at AI. 


Customer experience innovation: How we are helping clients to use AI

The question is ‘where to start?’. There is, of course, no single answer, but the key is to start with a use case – not simply with an ambition to ‘adopt AI’ – and to focus on areas where AI can make a tangible difference right now. More often than not customer experience design, is still about using AI’s ability to take on the heavy lifting – time-consuming, repetitive tasks that are nevertheless essential to human-driven innovation.

This is precisely where we are helping brands to harness the power of AI - drawing on its strengths to drive efficiencies and enable innovation. So, here are three examples of ways we are helping brands to harness AI in customer experience innovation:


1.     Furniture retailer: Using AI to reduce resource burnout and improve product information.

Our client, a large, multi-category furniture retailer wanted to respond to unmet customer needs by making furniture dimension information more accessible and easier to use.

Clearly, when it comes to large furniture items, such as sofas, dimensions are crucial to buying decisions. However, previously, the dimensions of individual items were only available as an image on PDPs. As a result, the information was hard to use, all but invisible to customers with accessibility needs and could not be used as a search filtering facet.

However, fixing the issue across a catalogue running to many thousands of products was proving a hugely resource-hungry, time-consuming task. Carried out manually across a relatively small sofa range, the process of manually extracting size information into spreadsheets took 25 hours and was subject to an error rate of 10%.

How AI helped: We used Google Vision API to take over the heavy lifting. After around five hours of developing bespoke scripts to apply Vision API to this specific use case, the tool delivered a full set of sizing information in just five minutes, initially with an error rate of between 0% and 2%. Further tweaking delivered consistent error-free performance.

The time and resource-saving is obvious - and even more significant when scaled up across the retailer’s entire catalogue. But just as importantly, the retailer is now able to display furniture sizing information in a much more usable, accessible way and is already working on adding dimensions such as height, width and depth to product and search filtering.


2.     Crafting and hobby retailer: Using AI to unlock more customer experience tests

The nature of our clients' product range demands that PDPs feature quite long, text-heavy product descriptions. These may be useful for customers new to crafting but can be cumbersome and unnecessary for more experienced customers.

Our client wanted to develop and test ways to solve this issue and ensure that PDPs were just as usable for customers of all types. However, extracting the information required to run tests – essentially drawing key product details from dense text across tens of thousands of PDPs – had always rendered innovation in this area prohibitively expensive.

How AI helped: We used OpenAI API to gather this detail from every product in its catalogue in a matter of seconds, by asking it to scan product descriptions and summarise the six main benefits of every product.

This required some work to shape and contextualise our question for OpenAI – simply because contextualising and constraining Gen AI input is crucial to gathering useful outputs. For instance, in some categories, we stipulated that the six benefits must include dimensions, weight and guarantee, while other categories required slightly different framing (for example “This will be read by an advanced user”).

However, even with that additional effort, applying AI to the problem opened the door to previously out-of-reach tests by making the preparatory work quick, easy and low-cost.


3.     Apparel brand: Using AI to identify innovation opportunities. 

Our client, a mass-market fashion brand, was seeking opportunities to improve product detail page content. The initial plan was to use promotional ‘badges’ conveying sales messaging – such as ‘Lightweight. Perfect for Summer' - as a means to drive conversions.

However, it was concerned about customer reactions to badges that can be perceived as a ‘hard sell’. 

In response, we arrived at the idea of using common phrases in positive customer reviews to populate dynamic PDP badges with regularly, and automatically, updated content. However, this approach would normally require time-consuming and expensive preparation. Even if the test was applied to a single product category, it would demand many hours of trawling through hundreds of reviews across 40 or 50 products.

How AI helped: We tasked OpenAI to search through all positive customer reviews across a test product category, summarising common phrases in four to six words – a task it completed in a matter of seconds, pulling out phrases such as ‘Perfect fit. Unbeatable price’.

This allowed us to test the hypothesis on a selection of product pages – a test that would otherwise have been out of reach. The results were positive and ‘user generated’ badges are now in place across its digital commerce site. 

What’s more, because we used the OpenAI API to gather the content, the client is now able to use AI to dynamically generate new badge content as reviews are added to the site – with a layer of human QA in between, badges can now be easily, almost automatically, changed to reflect the latest customer reviews.


What we learned

The main lesson from all this has been a simple one: It is possible to generate significant benefits from AI in customer experience innovation now, without spending months or years developing a ‘killer application.’

Every brand with a focus on constantly improving the customer experience can harness AI to accelerate innovation, reduce costs and free CX teams to focus on high-value creative activities.

The key is to remain abreast of developments in AI, but always with potential use cases in mind – because AI is not (yet) a magic bullet in its own right. It still needs to be used in a way that is planned, relevant and has clear, tangible benefits - and, in most cases, human input will still be vital. 

In addition, as with all customer experience innovation, testing to validate everything from customer response and accessibility to commercial impact will be crucial to the effective use of AI - as will be the ability to gather and interrogate data from AI-led solutions.

A good starting point is to think about business goals and where AI can help – where could it drive efficiency, deliver time savings, or enable innovation - and how can effectively test potential applications.


If your brand is starting out on its own Gen AI journey, seeking expert guidance and support - from identifying use cases and developing applications to devising tests - is likely to save significant time and resource, so feel free to get in touch for more information.