Winning in AI Ecommerce
AI is having a massive impact on ecommerce and retail in general… and has been for 20 years. When Linnworks invited me to write a blog on AI in ecommerce I was honoured as automation, and at its heart AI is another automation technology, matters to us here at Patchworks as it drives efficiency and therefore profits for our customers. In this blog I aim to provide actionable insights to help you drive more sales and increase ecommerce operational efficiency through the use of AI and automation technologies.
My bold opening statement “has been for 20 years” probably means we should define AI and machine learning and common AI tools in ecommerce. This is a short blog so will be a high level overview of common applications of AI – it’s not an exhaustive list.
Product Discovery
The earliest form of AI introduced to ecommerce platforms was “learning” search and product discovery. The ability for a customer to search for a product and for the search engine to include machine learning to optimise the conversion to cart based on the data it had to hand (product fit, other customers that had searched etc.). An example would be Amazon which was a pioneer in this area. Back in the early 2000s, they launched the Personalized Recommendation Engine, leveraging collaborative filtering algorithms to better serve their customers. For instance, if a customer searched for a book, Amazon’s AI-powered system would recommend other books frequently bought together or those that aligned with the customer’s browsing history.
Product Recommendations
These machine learning algorithms (people who clicked on this also bought this) then became available as product recommendation engines, aimed at increasing the average order value. AI-powered personalization in ecommerce became important to ecommerce managers to drive more sales. Once again, Amazon helped pioneer this though of course in those early days we all complained “I’ve just bought a TV, I don’t need another television”. Improvements in the technology have helped personalise product recommendations further based on product type and customer type leading to…
Customer Profiling
Applying machine learning to customer data led to the rise of Customer Data Platforms (CDP). A CDP platform unifies customer data from various sources to create a single, comprehensive view of each customer, enabling personalized marketing, email campaigns etc. aimed at increasing the site visits, particularly of loyal customers.
Generative AI
The previous technologies have been in place for years, generative AI, and how easy it is to have AI write campaign pages or talk to customers directly in a chat bot, is the reason for all the hype right now, and it is transforming the ability of retailers to create personalized campaigns (I know of a major brand that creates 4 million localized product images a month for instance – something not possible before GenAI)
AI is also being applied in inventory management, script building, you name it. It all gets a bit confusing, so how do we apply it to our stores?
Quick Wins
You’ll have noticed that I mentioned key ecommerce terms in my preamble – conversion rate, average order value, site visits. Before embarking on a “throw it all in” approach it’s worth analysing your site analytics to see if there are any obvious areas of improvement and concentrating on those, and creating a test framework to measure the impact of the AI technology being put in place.
User experience is absolutely key in ecommerce and the implementation of a CDP, linked to product discovery and recommendations, can make a massive difference to a retailer’s top line. This can positively impact all 3 of the metrics mentioned above with better written, more personalised email campaigns, and better product recommendations on-site when the customer visits making them more likely to convert.
To improve conversion rate further, a negotiation chat bot like Nibble has demonstrable benefits, and live chat can also help when integrated into inventory management particularly for the click-and-collect (BOPIS) model. Better inventory management using tools like our partners Linnworks can help in general offering the best products to the most likely customers.
Avoiding AI Pitfalls
At Patchworks a common pitfall we see is the “do everything” approach. AI is now so accessible, with free-trials, apps etc., it’s tempting to throw a lot at a site and see what sticks. Coming back to the metrics however, it’s better to define a fortnightly cadence, add AI and measure the change based on that technology. Of course it’s also important to keep on top of the cost, most AI tools allow you to define spending limits and with an automated integration make sure you’re checking daily on your spend, or you’ve set up alerts to ensure you have a good return on your investment.
Generative AI is allowing retailers to create dedicated campaign pages or outbound email campaigns more quickly and easily than ever, but it’s important to include a quality control step to ensure that the campaign makes sense and isn’t “obviously” AI (overly verbose, using “Americanised” language and repeating itself). GenAI is a brilliant accelerator but must be checked to avoid customers ignoring the page or hitting the dreaded “Report Spam” button in their email client.
Connectivity and Integration as AI enablers
Patchworks is seeing customers large and small adopting AI across the board. Integrating AI can be as simple as installing the vendor’s “app” and this is a perfectly valid integration method but at Patchworks, our vision is to make integration of technology as easy as possible via our strategic integration layer. This gives our customers the ability and optionality around different technologies including connecting to things like OpenAI, Google Gemini etc. if needed, and to connect through to Linnworks for inventory planning, order management, marketplace integration etc.
A good example is our Mint Velvet case study .As well as automating orders and inventory to the warehouse and accounting software, Patchworks is used to integrate bloomreach’s CDP and Attraqt’s product discovery AI, ensuring customer data and product data is always up to date. Another use of AI that Linnworks is pioneering is inventory planning and management – making sure the right stock is in the right place as demonstrated in the launch of OG Kicks, a project where Patchworks automates the integration of Linnworks, StockX and Lightspeed to ensure their demographic, “sneaker heads”, get the shoe they want at the right time.
Next Steps
When breaking new ground or implementing something new it’s hard to know if it’s working or where to get advice. For next steps, try and find an expert in an agency who has been using AI for the past 2 years, nothing beats experience, and ask them for help to do an AI review on where it can help your business. In addition, work on a measurement framework to ensure you are making positive impacts and make changes 1 thing at a time, don’t throw everything at the wall at once and hope it sticks – incremental changes compound over a time; a 10% improvement in site visits, 10% improvement in conversion rate and 10% improvement in average order value could compound into a 40-80% improvement in revenue!
Closing Remarks
One of the things that AI has brought into sharp focus for me is the acceleration of technology and its impact on retail. Everything is changing, more quickly, all the time. With the only constant being change, it’s important to have a tech stack and business agility to pivot toward something that starts to dominate your industry, whether it’s marketplaces like TikTok or the impact of AI on SEO and paid media.
Creating an agile tech stack that can quickly integrate and incorporate these technologies is therefore key to success in the modern world. Patchworks and Linnworks are examples of platforms that exist with this vision to help our retailers respond to this ever changing world and, as always the real secret to success, put their customers first by making it easier than ever to buy the products they want.
