The problem
An aviation data business had put a lot of work and a lot of pre-sales effort into its emissions data, but had no view of how the portfolio had evolved. The data science team’s work sat in Jira as tickets under a single epic called “Emissions”. There was no roadmap, so nobody could say what had been built, what customers were really asking for, or where interest was heading.
What I did
I took 12 months of transcripts from customer conversations and used AI tooling, Dovetail and Copilot, to see how those conversations had changed over the year. From that I built a view of the themes and epics that had come up. I did the same for the customers we were speaking to today, to see what they were asking about, how it lined up with the work we had already done, and which themes were drawing more interest.
AI can group things wrongly, so I tested the themes with the people closest to customers. In monthly calls with the teams who spoke to customers, and with sales, we asked whether each theme was a fair description of what we had delivered, and they confirmed or rejected them. Before this, finding out who had spoken to whom and what had happened meant asking around. Now it was a look at the insights in Dovetail, which by my estimate cut the synthesis time by more than half.
The result
Nobody had done this before. It gave the data team the first view of how the work had unfolded and what the future epics would look like. What it showed most clearly was how far the datasets had come: emissions had seen a lot of work and a lot of pre-sales pitches, but none of it had yet turned into revenue. So my recommendation was to pause further development and first convince customers to buy what we had already built.