We look at the trending use cases for AI & machine learning, as a means to optimising eCommerce websites and operations.
Gartner predicts that by 2020 over 80% of all customer interactions will be run by Artificial Intelligence.
In the offline world of bricks and mortar stores, brands have the opportunity to properly engage with and understand their customers, providing a personalised service.
But emulating that customer experience has been a perennial challenge for digital marketing & eCommerce managers looking to better understand not only who their customers are, but what they care about, and what motivates them to buy (or indeed leave their website altogether without buying anything as so often happens).
The key to building relationships with customers and improving conversion lies within the rapidly evolving area of AI and machine learning.
Now, we’re fully aware that AI & ML are buzzword topics but given their potential to impact the industry, it’s for good reason. First, let’s recap on the distinction between the two.
Artificial Intelligence: Machines that are able to execute a series of specific actions tasks by emulating a real life human.
Machine Learning: A subset of artificial intelligence. Data science that’s facilitates adaptation of AI over time. It crunches data, identifies patterns, it optimises itself. It’s what can enable AI to pivot as it’s directly learning from experience and making data-informed decisions and predictions, rather than simply executing responses based on some predetermined rules.
We’ve spoken to a lot of brands over the last 12 months on where they plan to put their investment of time, effort and budget – and whilst there is a lot of interest in it as a subject matter, for a number of the SME’s in particular this tends to be on the to-do list for ‘Phase 2’ or ‘Yes we should look at that’ – but the reality is many merchants are still trying to get the basics right and are rightfully conscious of not wanting to run before they can walk.
They’re not all brands with the resources of say Amazon or ASOS – companies that have entire departments dedicated to AI and data science. The behemoths invest to better understand their customers, developing an offering that’s personalised to each customer. They are investing in personalisation at scale.
That said, it’s worth understanding some of the contextual applications for these technologies as they can ultimately help improve your bottom line, whether you’re looking to implement something now or in the not-too-distant future.
Personalised Product recommendations & Content suggestions
These aren’t new ideas. And these shouldn’t be in the ‘nice-to-have’ list for your next website redevelopment project.
These are fundamental to meeting customer expectation now, cut and dried.
SalesForce found 62% of consumers expect companies to send personalised offers or discounts based on items they’ve already purchased.
One of the most widely recognised opportunities is around employing machine learning to tailor offers, communications and recommended products to both customer segments and individuals – automatically.
There are already various AI-driven product recommendation extensions already on the market for eCommerce platforms, including but not limited to Nosto, Barilliance, SLI Systems and solutions like Logic Hop that can help serve targeted content.
Whilst you can define your own criteria, AI & machine learning can do a lot of the heavy lifting – it can personalise interactions at scale and critically, improve the approach over time based on actual data which, let’s be honest, not all of us have the time to make sense of.
You can base suggested product & content on various factors, including the obvious content / pages browsed, purchase history, frequency of purchase, geographic location, time of day / week / month, how users arrived (i.e. which traffic channel brought them to the site). Even the weather can be taken into account and could be used to make umbrella products more prominent on a retailer’s homepage on a rainy day.
It can also serve to enhance the ranking and grouping of site search results, deal with serving results in spite of misspellings, and offer relevancy based on past user behaviour.It all sounds great doesn’t it? The ultimate goal is to not only to increase your AOV and improve your conversion rate – but to deliver a better customer experience that works out for everyone – the personal touch will support improving loyalty and retention. It’s not just immediate revenue growth right now, the bigger picture to pay attention to here is Customer Lifetime Value.
Research conducted by Segment found that 44% of consumers say that they will likely become repeat purchasers after having a personalised shopping experience served to them.
On the flip side, Bazaarvoice survey from concluded that 38% of US digital shoppers said they would stop shopping at a retailer that made poor product recommendations – a sentiment that applied to several product categories.
Essentially, you can’t afford NOT to personalise the customer experience and AI can not facilitate automation in this regard, but optimise itself over time.
For a brand that’s competing aggressively on price, or even trying to merchandise product effectively with excess of inventory in mind, you should investigate using an AI-driven solution for dynamic pricing, adjusting for higher conversion rate based on customer behaviour and preferences, overall demand and time of day.
This technology can help brands automate presenting optimal prices and display real-time discounts, taking into account warehouse stock levels to help maximise sales and shift inventory that’s not been selling as expected.
A/B testing & its potential for CRO
Here’s the thing about A/B testing. Imagine you are running a Conversion Rate Optimisation experiment with an A/B test of the main hero image on a product detail page on your eCommerce site.
Let’s say 20% of total conversions are happening with the test’s control, which we’ll refer to as page Variant A – and that 80% of conversions are resulting from a new variation that’ll refer to as Variant B.
The paradigm of A/B testing would lead most to concur that ‘B’ has outperformed and therefore should be designated the new control for future tests.
Therein lies a problem with the logic. There’s categorically no conclusive guarantee that people who converted on variant A would have also converted on variant B.
That’s because a rudimentary test like this isn’t catering for (and isolating) specific customer segments, it’s a blanket approach.
Machine learning can take into account various data sets relating to customer behaviour and more intuitively serve an experience for segments and individual behaviour through real-time data-driven optimisation. This is where the technology is fundamentally going to scale personalisation and make an impact.
It can analyse, learn and predict behaviour, helping to serve the right content & product, to the right people, at the right time.
Mike Williams, Business Development Director
The Personal shopping assistant who works 24/7
Yep, an application of the tech that you will already be familiar with. We’re talking chatbots/virtual assistants. Chatbots can bridge the gap between online and offline customer experiences. Numerous big name brands have delved into the employment of chatbots as personal stylists, from to Tommy Hilfiger, H&M, Sephora and Burberry amongst others. The best examples serve to make personalised product recommendations based on an intuitive line of questioning – helping to improve conversion, not just dealing with wider customer FAQs.
Whilst rule-based solutions are prevalent on the market, a number of established payment providers, like MasterCard for instance, have adopted AI to learn from experience and to iteratively detect and prevent the ever-evolving approaches to fraudulent activity. Developed models can scrutinise each and every action a cardholder takes, taking into account not only the typical variables such as size of transaction size, location, time, device, etc – but also whether the transaction at the very moment is characteristic of that particular user.
Visual search technology uses artificial intelligence to analyse content and context of images to return a set of related results. Algorithms can detect patterns, colors and cuts – and pull products that match similar attributes of products viewed by customers. According to Gartner, by 2021, early adopter brands that redesign their digital products to support visual and voice search will increase eCommerce revenue by 30%. Like AR, it lends itself to some categories more than others obviously – in particular it has huge potential for categories such as fashion and home decor. ASOS rolled out it’s Style Match visual search tool last year and we’re keen to see what engagement and performance it’s had for the brand. Forever 21 embraced the technology and increased AOV by 20%. At Amazon’s re:MARS 2019 conference held in June, the company announced a new in-app tool called ‘StyleSnap’, which will enable users to take a photo or upload a screenshot of any item of clothing and the system will find similar product matches, based on image recognition. According to research conducted by ViSenze, 62% of millennials want visual search over any other new technology. Stats aside, ultimately any tools that make it easier for users to find the products they want – faster, in a smart and easy to use manner like this are only going to help provide a more friction-free and effective experience that improves sales.
Focus on getting the fundamentals right and then look at how various personalisation, search and merchandising options can enhance what you’re doing. If you’re an SME you don’t necessarily need to start hiring data scientists in house and developing new algorithms and models from scratch either. Most machine learning solutions are based on solving problems that bigger businesses have already tapped into. Identify your biggest pain-points within the business, bring us challenges and we can help brands identify suitable options to consider.