Karl Hampson, Chief Technology Officer, Data & AI, Kin + Carta tells how their work with Not On The High Street (NOTHS) and groundbreaking work on its search functionality resulted in a 10% increase in search conversion rates YoY when Retail Search went live to all users. Not only that, but NOTHS achieved a 2.2% uplift in revenue per user during A/B testing of the solution and initial testing results were so impressive that Retail Search was rolled out across the entire search function after just one week:
There’s a paradigm shift happening in product discovery. Ecommerce retailers who are leveraging the latest AI tools are able to bring customers closer to their brand via more relevant search, browse and recommendation results. This, much like a personal shopper, can quickly return the exact products that users are seeking.
Whether it’s a black cocktail dress that’s cut to the knee with a splash of pink or a search for inspiration for a gift for your mum, AI is removing shopper’s decision fatigue and the problem that many online users face when having to browse through 100’s of products to find what they want. Instead, Generative AI allows fully natural language queries to be used to find their desired product.
With this new high bar in product discovery, this prospect of growth and advocacy amongst customers marks an exciting opportunity for retailers looking to drive revenue whilst building a closer relationship with their customers.
Not finding what you want
When it comes to product discovery, most retailers are still following the same model; augmenting their search technology with lots of manual rules for merchandising but also to make up for deficiencies such as manually curated synonyms and the dreaded ‘no results found’. It means that, despite significant investment and time, the end result for customers is still lacking. True 1-1 personalisation is uncommon in search and browse but is now completely possible with the latest AI technology.
Search technology isn’t new and the paradigm many retailers use has been around for two decades. Revamping an entire search platform can be daunting. And often costly. When budgets are tight, the back end of retailer’s websites can be left forgotten.
So how can retailers improve on this in a way that’s seamless and isn’t breaking the bank? By implementing the use of AI within their search, browse and recommendation functions or as Google has put it, implementing Vertex AI Search for Retail.
Because who wouldn’t want Google quality search relevancy and ranking embedded into their online stores, trawling through their products in a more sophisticated, thorough and customer-bespoke way?
Think of Vertex AI Search for Retail like this – it’s your trusted librarian who knows exactly where to find the book you’re looking for in a vast library, because you can describe what you want in natural language, rather than the legacy approach of only matching keywords. So, “I’m looking for a book that is about some hobbits that need to destroy a magical ring.” That librarian is going to tell you where The Lord of The Rings is.
This AI-driven, sophisticated search function will give personalised, and perhaps most importantly, correct and relevant results for customers searching or browsing your site. And it will, in turn, lead to greater chances of purchase, reducing churn and increasing the prospect of repeat visits by that customer. The outcome for the retailer? Revenue uplift and happier customers.
But let’s take it one step further. What about marketplaces, which have vast amounts of products, and variable quality product data added from numerous retailers and individuals? This is where Vertex AI Search for Retail can be really put through its paces.
And there are not many bigger marketplaces than Not On The High Streets (NOTHS).
Filling the gaps
NOTHS’s curated online marketplace currently hosts 5,500 small creative businesses from across the UK, ranging from home furnishing to fashion. In total, there are 350,000+ products currently sitting on its platform. The business prides itself on championing creative entrepreneurs while empowering consumers to discover special finds to gift or for themselves.
The operative word there being ‘discover.’
Similarly to other retailers, the NOTHS website’s search function was not performing at the desired level. Which in turn meant that the customer’s experience was also falling short. It needed improving.
As a Google partner, Kin + Carta and its team of data and AI experts with experience working with some of the biggest names in the sector, was brought in by NOTHs to evaluate its current search model, and test it against an implementation of Vertex AI Search for Retail. The marketplace wanted to:
- Boost visibility for 350,000+ products, preventing repetitive displays to shoppers.
- Enhance data and product tagging consistency across numerous small businesses, reducing manual tasks.
- Intensify relevance with advanced AI search, delivering accurate and personalised results to drive conversions.
With these goals in mind, work began.
Implementing AI
The first step was to perform an offline search audit to identify the use case for Vertex AI Search for Retail on the NOTHS website. In two weeks, 350,000 products and 26 million events were ingested, and a sample of 90 search scenarios were identified to be evaluated.
Post-audit, the next step was to implement a production-ready solution in just under 12 weeks. A split of 50% of NOTHS website traffic to the current search tool and 50% to the new solution was conducted, measuring key metrics like search click-through rate, conversion rate, and revenue per user.
The results spoke for themselves:
- Revenue growth – achieving a 2.2% uplift in revenue per user during A/B testing of the solution.
- Boosted conversion rates – experiencing a 10% increase in search conversion rates year over year when Vertex AI Search for Retail went live to all users.
As a bonus, the entire audit and testing process was swift, allowing NOTHS to start experiencing meaningful value quickly. A/B testing finished in June 2023, with the solution scaled to 100% of users just a week later because of the strength of the results delivered by Vertex AI Search for Retail. It’s clear that implementing Google’s AI system was a boon in every sense of the word.
Retailers – what are they missing out on?
The implementation and success of Google’s AI within the NOTHS website search function is a clear example of the need for retailers to up their search game. So what are they missing out on if they don’t?
State-of-the-art search relevance and ranking
The latest Large Language Models (LLMs) are game changers in powering query understanding, identifying highly relevant products from the retailer’s catalogue, regardless of the actual language used.
Personalised customer experience
AI-powered product discovery allows retailers to automatically understand customer preferences, behaviour, search and purchase history to deliver personalised results. Without AI, retailers may struggle to tailor the shopping experience to individual needs and preferences, potentially leading to lower customer satisfaction and loyalty.
More opportunities to convert
AI-powered ranking models, which are automatically created for each retailer, ensure that the products most likely to be of interest to the customer are shown first, increasing the likelihood of conversion.
Whilst all of this is giving the customer a great experience, at the same time the technology is always optimising revenue for the retailer, giving them an edge. In today’s competitive landscape, retailers that embrace AI-driven technologies gain a significant advantage over those that don’t. AI enables retailers to stay agile, adapt to changing consumer trends, evolving product catalogues and innovate in response to changing market dynamics. Without AI, retailers run the risk of falling behind competitors and losing market share.
We now live in an age in which many technological solutions to retailing woes are readily available and evolving quickly with new capabilities. Google Vertex AI Search for Retail is a single platform for all product discovery needs including Google Vision AI and Voice Search. We are moving to a world where customers will demand the latest features such as multi-modal approaches to discovering products. With Google this can be achieved without having to perform complex technology integration from multiple vendors.
Now, more than ever it’s up to retail businesses to take those necessary steps toward implementing AI. They can challenge legacy approaches and ask their CFO ‘What is stopping us from using AI for search?’
It would be fiscally irresponsible not to.
