A conversation with a Head of Finance who built his own reporting layer, hit its ceiling, and learned what has to come before the tools.
Open LinkedIn on any given morning and AI is about to transform finance. Forecasts will write themselves. Insight will be instant. The month end will disappear.
For this year’s FP&A research I am sitting down with finance leaders and asking what is actually happening inside their functions. One recent conversation has stayed with me, because it came from someone who has done more with data than most finance leaders I meet, entirely self-taught, and his message about AI was the opposite of the feed.
He is the Head of Finance at an owner-led UK recruitment business that has grown from a start-up into an established SME. What he has built, and where it stopped working, says more about AI readiness than any vendor demo.
The self-build
He joined at the end of 2019, brought in to put structure around a business that was feeling the strain of its own growth. The pandemic arrived the month he started, so year one was survival. Then the rebuilding began: accounts payable automation, expense automation, and eventually the move from an entry-level accounting package to a proper ERP.
The CRM was old, with poor reporting, and the vendor offered no way forward. So he taught himself Power BI. Every morning, sales activity data now extracts automatically from the CRM into SharePoint and combines into Power BI, giving the business daily visual activity tracking. Before that, an administrator ran team reports in Excel and emailed them around.
Once a week at best. Nobody was looking at it.
Head of Finance, UK recruitment business
A weekly report nobody read became a daily habit the business actually uses. That is a genuine transformation, built by one person inside the finance function. And then it hit a ceiling.
The ceiling
The ceiling was not Power BI. It was everything underneath it. The business runs an old CRM and an old pay and bill system that are meant to talk to each other; run the same report in each and you get different answers. His daily pipeline relies on a single morning export, which means an interview recorded a day late never enters the data at all. And planning is stitched together from disconnected sources.
If you want to change anything, you have to rebuild pretty much the entire model.
Head of Finance, UK recruitment business
The dashboards were never the problem. The architecture underneath them was. Which is exactly why his answer to my next question matters.
What to fix before AI
I asked what a finance leader should fix before investing in AI or a planning platform. His answer took two words: data quality. Bad data, he told me, just gives you faster meaningless reporting, “or wrong reporting, which is probably even worse.”
Then he said the line I have been repeating ever since.
System implementation is a very small part of transformation, and it’s probably the one that should come at the end.
Head of Finance, UK recruitment business
Data first, he said, and then culture. Put in an impressive new platform without training people and enforcing how data gets created, and “you’re going to have a shiny new tool which is going to produce empty reports.”
Incentives create data
Here is the part most AI conversations skip entirely: data is created by people, and people respond to incentives. In his world, a recruiter logs one interview in the system when five actually happened. At that point every ratio the business runs on that data is meaningless.
You need to have the right incentive for people to produce the right data.
Head of Finance, UK recruitment business
Pay commission only on what the system shows and behaviour changes in days, not quarters. Put a Monday dashboard on the office screen naming the week’s most active performers and competitive people compete. He has watched “even the simple thing like seeing their face on the dashboard in the office” shift how a sales floor records its work. As he put it: “If you can’t hide, then there is extra motivation.”
None of that is a technology project. It is design: processes, incentives and behaviour, deliberately arranged so that good data gets created at source. He was equally clear about whose job that is: “It’s on us as finance to go and meet them where they are.” Put a P&L in front of a sales leader and you lose them in five seconds. Put the same story in two or three visual KPIs and they act on it.
What AI actually does
He is not an AI sceptic. He uses it heavily at his own level, and his description of it is the most precise I have heard: “I know the inputs I give the AI is good. When I get it, it is absolutely amazing.” The moment the underlying data breaks, even slightly, the quality of the output nosedives straight away.
He had a warning about polished demonstrations too: “You will never achieve that demo data in a real company.” Real businesses have friction, silos and people who are good and bad with systems. AI is leverage on your data layer. It multiplies whatever is there, clarity or chaos.
The invisible backbone
I finished by asking what would make an external FP&A partner credible to someone in his position. Not dashboards, and not software. The value, he said, is in building the structure nobody sees.
What if my sales increased by 10%? What does that mean? There is a lot in the background that needs to happen to be able to have that end product.
Head of Finance, UK recruitment business
That background is what I call the design of FP&A: ways of working, processes, incentives and data flows, arranged deliberately, with systems implemented last, on foundations that can carry them. Skip the design, and in his words, “you’re going to be wasting your money.”
Where does your function stand?
If AI is anywhere on your roadmap, the real question is whether your data layer is ready to be amplified. The Board Readiness Assessment takes ten questions and about four minutes, and gives you an instant score. If you would rather talk it through, book a Board Readiness Review: thirty minutes with me, confidential, no pitch.
Interview from the 2026 FP&A research programme. Interviewees are anonymised by role and sector. The 2025 FP&A Report is available now.
