Speaking
What I talk
about.
Three arguments I actually have, from running a data and AI function rather than advising on one. The record underneath them is public, and every line links to somebody else’s page.
The talks
Three, and
I mean them.
Each of these comes out of work I was accountable for rather than a topic I found interesting. They shorten and lengthen, and the third one is better with an audience willing to interrupt.
Nobody finishes the transformation they started
Monarch butterflies leave Canada each autumn and reach a handful of forests in Michoacán having never been there before, and it takes several generations to do it. The one that arrives is not the one that set out. Large programmes work the same way: the sponsor changes, the team changes, the budget cycle changes, and the only thing that can survive all of it is the direction. I talk about how to write that direction down so it holds after you have gone, why programmes lose the heading in the handover rather than in the delivery, and what to ask for in the first month so the answer still makes sense in the third year.
Everyone wants the AI. Nobody wants the foundations.
Most AI pilots go well in the room and then quietly stop, and it is almost never the model that failed. It is that two systems disagree about what a customer is, nobody owns the definition, and the people who would have to trust the answer were never in the conversation. I talk about the unglamorous work that makes the clever part possible: governed layers the business actually recognises, a definition somebody owns by name, and the difference between data people can reach and data people believe. It is duller than the demonstration, and it is the reason the demonstration never ships.
The order you do it in is the strategy
Most transformation plans are a list of correct things in the wrong order, which is why they stall a few months in with everyone still being polite about it. Accountability has to be settled before structure, or the reorganisation only moves the argument somewhere else. Demand has to be managed before capacity is added, or the new team is consumed on arrival. Ways of working are announced before the date they take effect, not after it. And job descriptions get rewritten to the outcomes you actually need, because titles are how a team quietly stays the same. I walk through my own sequence, including the two moves almost everyone runs the wrong way round, and I am happy to be argued with about it.
The rooms
Where I
have said it.
Every line here links to the organiser’s own page rather than to mine. Where a source names my employer rather than me, the line says so.
2026
Google Cloud Next
Las Vegas, on what is new with data agents, which is Google’s own title for the session.
Anaplan Connect London
The main customer keynote.
Women in Data Flagship
The Empowerment Stage, with Virgin Media O2 as a mainstage partner, and 3,500 people at the event.
The c-suite podcast
Recorded at Google Cloud Next in Las Vegas, in partnership with Kyndryl.
2025
Dreamforce
The breakout Salesforce titled Deliver Insights Across the Enterprise with Tableau, alongside the global chief information officer of Box.
Gartner Data & Analytics Summit EMEA
ExCeL London, on turning data governance into a business enabler, with the co-founder of Atlan.
Tableau and Agentforce innovation preview
Featured speaker on Salesforce’s innovation preview, with Tableau’s chief executive.
Where I am useful, and where I am not
Rooms I am useful in
- Chief executive peer groups, where the room is a dozen people who each carry the whole thing.
- Private equity portfolio days, where the same problem is sitting in eight companies at once.
- Mid-market conferences whose audience is the executive team rather than the practitioners.
- Client events run by accountants, law firms and other professional firms.
- Vendor customer conferences, where the room is cross-sector.
Rooms I am not the right speaker for
- A technical deep dive or a product demonstration. I am not the engineer in that room and somebody else will do it better.
- A motivational keynote. I talk about work, and there are people who will send an audience out feeling changed far better than I would.
- Anything in telecoms, connectivity or media, which I am contractually out of until the end of 2026.
If you are putting a programme together
The most useful email tells me the date, the room, who is in it, and what you would want me to talk about. I will come back to you on whether I am the right person for it, and when I am not, I usually know who is.