Improving purchase intent: With 2.4 billion behaviour transactions per quarter, carsales.com.au turned to AI

2.4 billion behaviour interactions every quarter are recorded by carsales.com.au. The online car marketplace explains how it used AI to improve use of that data.

Like most publishers, carsales.com.au wants to know more about its audience. And, obviously, for the digital auto marketplace, audience insights have a very direct relationship to sales. AI has just turbocharged those efforts.

“It’s about really identifying where they are in that buying journey. Are they ready to buy?” Stephen Kyefulumya told Mediaweek. He is carsales.com.au’s General Manager of Media – Product and Technology and is very passionate about the opportunities AI is affording his team.

The team at carsales mediahouse, the media arm of carsales.com.au, were able to use Adobe’s Customer AI to determine customer intent, better refine its audience segmentation, and deliver to site visitors smarter targeted advertising.

It put the new Adobe’s AI functionality to work within Adobe Real Time CDP, delivering some impressive data insights. Adobe’s Customer Data Platform (CDP) is built on the Adobe Experience Platform, using known and pseudonymous customer data to create real-time, actionable B2B and B2C customer profiles.

With this greater audience data, carsales mediahouse were able to achieve:

• 20% improvement in identifying purchase intent
• 23% increase in push notification engagement
• AI-powered audience segmentation using real-time behavioural data

Stephen Kyefulumya profile photo

Stephen Kyefulumya – Carsales.com.au General Manager of Media – Product and Technology

There’s a very short ‘ready to buy’ window

“You come on Carsales.com.au to do multiple things, to look around, to assess prices, to do a lot of research,” Kyefulumya said.

“Most people will come to carsales.com.au to do a level of research, but irrespective of whether they eventually submit a lead within car sales, they’ll do research on the ground.

“That 20% was more around identifying how close are people to really buying within that two to eight week period.”

Car buying journey infographic

The carsales mediahouse team already had great first-party data on its audience. There are 2.4 billion behaviour interactions every quarter recorded by carsales.com.au. Every interaction is seen by the team as a signal surrounding user intent. A customer may search for a car, then search for specific models, and go deeper on information as they search for a price. Each of those steps is registered by the system as a signal.

What the team needed from the AI functionality is to take all of those 2.4 billion behaviour transactions and apply that to customer journeys.

That information could then be applied to the broader marketing eco-system that carsales.com.au has developed.

“When we overlaid the buying journeys to the marketing messaging, we saw a 23% uplift in engagement,” Kyefulumya explained. “What that tells you is we delivered the right message to the right person at the right time.”

Buyer journey flowchart

Why Adobe Customer AI?

carsales mediahouse has been using Adobe CDP for around four years. It began with a consolidation project. “That was the first part – is we made a CDP investment,” he said.

“We made the investment so that we could have our first party data in a single well-governed platform where we do have a sense of that single customer view, consolidating all those signals.”

Once that was in place, the team was in a better position to take advantage of the AI benefits Adobe offered.

“AI is about data. AI is about how good your data is,” Kyefulumya said.

“Using their customer AI feature on top of the data that we’ve already inputted was a case of the right tool at the right time to give us speed to value.”

Advice for others looking to use AI

When asked if he had any advice for others looking to take advantage of AI to better make use of the data that the business already has, Kyefulumya kept coming back to data quality and integrity:

“Obviously it changes industry-to-industry and it’s very broad. But there are some consistent themes.

“The quality of your data is a consistent theme. Without good quality data, good quality first-party data coming in, you’re not going to be able to do AI. You need data, you need it well-governed, and you need to be able to create scale out of that.

“And I think once you have that, sieving through the noise is also, I think, picking the right enablers. Pick enablers that are going to give you speed to value. Then iterate from there. would be my advice” he said.

To Top