From experiment to impact: What results can generative AI products deliver for publishers?

Will a chatbot do more than just look cool in a product demo? Can generative search help you grow loyal readers? It depends on your approach to experimentation. It's easy for product teams to get stuck in endless back-and-forth discussions, mulling over ‘Who is this product for?’, ‘How much will this cost us?’ and of course, ‘What do we get out of this as a business?’. These questions are important, but this linear approach to development will ensure the idea never makes it past the whiteboard. You do not need to have the precise answers on the target audience or managing costs at the prototype stage. The essential thing is to launch something and adapt it on an ongoing basis as usage reveals deeper insights about its product-market fit.  

Leaning on the FT’s experience, Ask FT’s success has been in no small part due to its phased rollout and development. For the unaware, Ask FT is the first customer-facing Generative AI-powered FT product; there was no telling how it would go down with the audience. But before being launched to customers, the product was rigorously tested by members of editorial and product development teams at the FT to ensure a safer, more accurate rollout.

The first phase of customer testing saw the tool rolled out to a few hundred FT Professional (B2B) subscribers, who make up the bulk of FT’s readers, after which usage data and feedback were used to improve the tool for the second phase. This second phase was where Ask FT was positioned, based on feedback, as a value-add for larger client accounts across a range of businesses, not just finance professionals. This allowed for FT’s commercial teams to take a ‘one-to-one’ approach in order to learn how to optimise the wider ‘one-to-many’ rollout happening now.

Timeline: development of Ask FT feature

By listening closely to early user feedback, the team introduced meaningful updates, like improved answer quality, clearer source referencing, and smarter handling of complex queries, that made the tool more useful and trustworthy. 

We asked a handful of users to try out Ask FT for the first time live on calls with us and listened as they shared if it was working the way we all internally hoped it would. Getting this feedback was essential for us to improve the tool and make it more useful and trustworthy.

 

Samarth Soni

Senior User Researcher, Financial Times

These refinements helped maximise the likelihood of driving business impact in later development stages, where Ask FT has since reinforced the FT’s brand as a credible source of business intelligence. Ask FT provides tangible value to FT Professional subscribers by helping users gather information to prepare for meetings, conduct research and identify expert opinions. Positive repeat-visit indicators suggest that Ask FT can form part of a regular workflow for users.

Ask FT has boosted the likelihood of retention of key accounts by linking to the full text of FT articles in its output, encouraging deeper engagement with the FT content library. 

This increased article consumption can shift some users into becoming Actual Core Readers (ACR). ACR is a key renewal metric; it's the number of unique users on an FT Professional license who have read nine or more pieces of content in any one rolling 30-day period since the start of the contract. In this way, we think Ask FT can play a role in boosting the retention of key accounts. More qualitatively, clients view Ask FT as an innovative add-on, and the feature is helping to differentiate the FT from other publishing competitors as well as generic AI tools.

But what about those pesky costs? For any LLM-based tool, costs scale with usage and managing costs requires trade-offs between speed and accuracy. However, the Ask FT team notes that the relationship between speed, accuracy, and costs is not always linear. Consequently, the right balance of all three will be different for each publisher. At FT Strategies, we have become generative AI experts, helping publishers launch well over a dozen customer-facing and internal generative AI tools in the last year alone. We help publishers launch and scale both customer-facing and internal tools without drowning in cloud computing bills. 

In the end, launching a chatbot or any other customer-facing generative AI product isn’t about chasing hype; it’s about identifying small, strategic bets that can unlock compounding value over time. The Ask FT story is a reminder that you don’t need all the answers upfront. What matters is starting small, learning fast, and staying close to real user needs. At FT Strategies, we’ve seen firsthand how the right mix of experimentation, audience insight, and responsible AI design can lead to real business outcomes like stronger retention and brand differentiation. If you're a publisher wondering whether a chatbot is worth it, the better question might be: how quickly can you learn what works?

To learn more about how your organisation can utilise Ask FT, please visit professional.ft.com/ask-ft.