Carwow has deployed an algorithm-powered recommendation engine to provide real value through relevance and discovery, according to Dr Gareth Menton, head of product- retail media at the online marketplace.

Built by data scientists, the engine looks at what hundreds of users have “built and priced” over the previous year and recommends cars that “naturally cluster together” in buyers’ minds.

Drawing on “behaviour rather than editorial opinion”, Menton highlights that it reflects “what real shoppers actually consider”. He adds that the model has safeguards to ensure “only well-evidenced pairings appear.

The second method the engine uses, compares specifications of the cars, looking at things like price band, fuel type and body style.

He explains that a shopper looking at a particular hatchback would be shown other cars that are similar in specification.

He says: “This drives recommendations for logged-in users in their personal showroom and in some marketing emails.”

The third approach factors in models which can be brought nearby by looking at local dealer stock.

He says: “For the buyer, the value is relevance and discovery. Car-buying is high-stakes and unfamiliar — most people do it rarely and don’t know the full range of options.

“The engine widens their consideration set intelligently, putting genuinely comparable cars in front of them so they can make a more informed choice, often surfacing alternatives they’d never have thought to search for. Because the suggestions are grounded in what similar buyers did and in cars that are actually available locally, the recommendations tend to be practical as well as relevant.”

Menton highlights that the recommendation engine is useful to dealers and manufacturers.

He explains: “It channels buyer attention toward models that are genuinely in stock and toward dealers who serve customers well, converting browsing into qualified enquiries.

“It also creates a clear commercial incentive: manufacturers who put their dealers and stock on the platform gain greater visibility for their models in front of high-intent buyers. In short, the engine helps match real demand to real supply, which benefits everyone in the chain.

He says: We’re also starting to use it on advertising within the site, so that if the person is logged in, they’re configured, and we’ll use that recommendation engine to be able to display something else.

“We’re using it now in future products as well. When we are building new retail media products, we will use the recommendation engine to be able to advise on what type of vehicles we want to advertise to advertisers, so that’s the essence of the recommendation engine, it’s backed by data about our past users.

He highlights that the engine helps shoppers to discover new products by remaining relevant.

Menton explains: “We have to keep a personalised experience in today’s fast-moving world. We really try and keep things personalised, so the user, when they go into their apps, will see something that’s personal to them.

“It’s all about creating a really good customer experience.”

The engine also aims to improve the shopping experience because it shows recommendations that are personal to you.

Menton says: “At Carwow, we are the middle person, so we are the company that engages with the users and we are the route for them [advertisers] to be able to engage with those users. This allows them to be able to show products to relevant users.”

However, users may be looking at one listing and adverts pop up for competitiors.

He comments: “After a user configures a vehicle, we’ll show them five offers. If they configured an Audi A3, we will show them five Audi A3 offers.

“What we’re going to start doing is, in position three, we’re going to be putting a competing site, which means we’ll maybe show a BMW or Volvo to advertise there against all the Audi’s that are going to be shown there.

He added: “I don’t it’s too much of a problem, because those five offers that we show, it’s not one specific dealerships offer. There’s five offers there, so yes, they might not be happy that there’s a different brand on there, but they can go and do the same to the other brand.”

Click here to sign up to Retail Gazette‘s free daily email newsletter