For an event that has been specifically for CDOs and equivalents, it was an exciting change to hear from the top and have a CEO explain why the data and AI function means so much to their ambitions. Rightmove has been strengthening its data foundations while exploring how AI can create greater value from its extensive view of the UK property market. At the DataIQ Congress, CEO Johan Svanstrom and CDO Steve Pimblett explained how executive sponsorship, a focus on business outcomes, rapid experimentation, and tight control of AI economics are helping turn that ambition into products and capabilities.
The conversation revealed a pragmatic approach to AI transformation that starts with business value, moves quickly enough to learn, measures what happens, and accepts that leaders will need to invest before the outcome is clear.
Start with business value
One principle behind Rightmove’s approach is that its data and AI strategy does not exist independently of its business strategy. Steve described four broad areas where data, analytics, and AI can create value:
- Improving consumer experiences through personalisation and reduced friction.
- Expanding value for business partners.
- Monetising data and analytics (including automated valuation models used by mortgage providers).
- Improving the efficiency of the wider network.
As Steve put it: “We underpin the business strategy with data analytics and AI capability, not the other way around.” Starting with an AI capability and searching for somewhere to deploy it risks producing technically impressive solutions with limited commercial relevance. To avoid this, Rightmove works backwards from where the business creates value.
The company reinforces that discipline through a model connecting data and AI activity to value creation. Johan explained that it continually brings conversations back to “is it helping you create or not?”
Data and AI leaders need to give every data and AI initiative an explicit route to a business outcome by explaining who benefits, what changes, and how that change will eventually be measured.
Data leadership where business decisions are made
When Johan joined Rightmove, the data team was small and positioned several organisational levels below the leadership group. He saw an opportunity to increase both its scale and strategic importance. Today, the CDO sits on the executive team.
For Steve, that positioning provides sponsorship and it sends a message to the organisation about the importance placed on data and AI. This is important for when investment needs to precede demonstrable returns as data infrastructure, skills, and organisational capabilities may need sustained backing before individual use cases can prove their value.
Leaders need to treat executive sponsorship as an operating mechanism; access the discussions where investment, customer experience, product strategy, and commercial priorities are decided if they are expected to deliver against them.
Move fast and build in measurement
Faced with the possibility that conversational AI could change how consumers discover properties, Rightmove chose not to try to predict how search behaviour would evolve. Instead, it developed its own conversational search experience using its existing brand, property expertise, and proprietary data.
The first version of Ask Rightmove was developed and deployed in six weeks. The team considered A/B testing and model selection from the outset rather than looking at measurement as something to add after launch.
Users of the assistant spent 40% more time on the site, saved twice as many properties, and sent twice as many leads to estate agents. Johan acknowledged that the feature was still being used by only a small proportion of consumers, so there is plenty of opportunity.
That combination of speed and restraint is important and a key lesson for data and AI leaders. Early indicators can justify further experimentation without being mistaken for proof that a technology will achieve mass adoption.
Leaders should design experiments around learning and measurement and decide the behavioural or commercial signal that would justify further investment before launching a use case.
Mind the tokens
Successful pilots create economic problems. At Rightmove’s scale, even an AI pilot serving 1% of traffic can represent substantial volume and substantial economic considerations. That prompted an internal phrase championed by Johan and Steve: “MTT: mind the tokens”.
The principle is that teams should consider unit economics while designing AI products, rather than discovering the cost implications once adoption accelerates. Rightmove is applying FinOps and AIOps thinking to this challenge, drawing parallels with the disciplines businesses developed as cloud infrastructure scaled.
For data leaders, this changes what constitutes a successful use case. A prototype that performs impressively at low volume has not demonstrated that it can deliver the same value economically across thousands or millions of interactions.
Teams need to add unit economics to AI experimentation from day one. Leaders must ensure their teams track model choice, consumption, cost per interaction, volumes, and expected business value alongside conventional accuracy and adoption measures.
Invest through uncertainty
A serious leadership challenge facing organisations across the industry is deciding how much to invest while the future remains unclear. Johan described the tension: investors can question whether AI threatens the long-term relevance of an established digital platform and then, when the company invests in responding to that threat, ask for immediate evidence of ROI.
Data and AI leaders cannot know exactly how their niches will evolve over the next five years. To be able to handle uncertainty, the strategy should be to make sure the organisation remains able to adapt as customer behaviours and needs change.
Johan highlighted that there should not be unconstrained experimentation to combat uncertainty. Instead, leaders must combine long-term conviction with short-term discipline. Leaders need to be investing where proprietary data, customer relationships, and specialist knowledge provide an advantage, while accepting that not every bet will succeed.
AI strategy cannot depend on perfectly predicting what comes next. The practical objective is to build the capabilities, economics, and organisational backing needed to respond as the answer becomes clearer.
Register for the DataIQ 100 Summit North America.
