The Report Is Getting Cheap. Decision Support Is Not.

In July I argued on the Finatical blog that finance work should run in four steps: report, analyze, recommend, decide. I also noted that many organizations stop after the first or second step. Before AI, that was a missed opportunity. Now it is a pricing problem, because AI is getting good at exactly those two steps.

I trained as a chartered accountant. Later I ran Excel4apps, a software company built on the fact that getting from the ledger to a usable report was slow, skilled work. Today I’m building Finatical, which sells to the accountants and fractional CFOs this piece is mostly about.

What AI takes, and what it leaves

AI can now pull the ledger, lay out the variances and draft the commentary, often in less time than it used to take to find the right export. The report still has to be right, and checking it still takes someone who knows the books. But producing it is no longer slow or scarce.

AI can tell you that gross margin slipped and which product lines caused it. It cannot settle on its own whether the owner should raise prices, drop the product or ride it out. That depends on things that are not in the ledger: what the owner wants the business to become, how much risk they can carry, what the bank will tolerate.

In 2021 I wrote on this site that automation should move finance time from completing the close to interpreting the results. In my view, AI goes further. It does not just speed up the report. It makes the report too cheap to sell on its own.

The pricing has already moved

For accounting firms, I think the squeeze shows up first in pricing. The 2024 client advisory services benchmark survey from CPA.com and the AICPA covered 206 client advisory practices at US firms. Only 10% still used hourly billing as their main pricing method. Median growth was 17%. The release says growth in key metrics likely came from:

• more client advisory engagements, instead of tax cleanup and annual project work;
• standardized processes;
• fixed fees billed on a recurring basis.

A fixed monthly fee looks like good news when AI arrives. The work gets cheaper, the price stays put and the margin goes up. That lasts until a competitor offers the same package for less, or the client notices how fast the reports now arrive and asks what the fee is for. A price that rests on production is only as safe as the production is hard.

The same survey points to where the more valuable work sits. Firms earning significant revenue from CFO-level or higher business advisory reported more than 30% higher monthly recurring revenue. I don’t read that as advice being fashionable. Advice rests on knowing the client, not just the books, and that is much harder to copy.

Finance teams inside companies face the same pressure, with less visible pricing. I expect a team whose main value to the CEO is the board pack to find that value harder to explain every quarter.

A recommendation has to be said out loud

In June I wrote that “reports don’t drive decisions. Stories do.” The story that matters is usually an uncomfortable one, told to someone who may not want to hear it.

Telling an owner that their largest customer loses money, that the planned hire should wait, or that the price increase they keep putting off is overdue is a different skill from building the analysis. It takes context: the owner’s plans, the history of the business, the decisions they already regret. It takes the confidence to hold a position when they push back. And it takes nerve, because the cost of staying quiet never shows up on a report. The business keeps making the bad decision, and the report that could have stopped it gets filed.

AI makes this part of the job matter more. An owner can now get a decent variance analysis from an AI assistant, sometimes before their accountant has called. The tool will even recommend an option and give its reasons. What it cannot be is the person the owner holds to that answer next quarter: someone who knew what was riding on it and said it anyway. Once the analysis is available to everyone, I think that person is most of what is left to pay for.

Advice can’t stay a partner’s job

Most firms and finance teams, as I see them, are still organized like a production line. Bookkeeping feeds the close, the close feeds reporting, reporting feeds review, and at the end a partner or CFO has the conversation with the client or the CEO. That made sense while production was the hard part.

If AI runs most of the line, the structure should follow the decisions. Each client, or each part of the business, needs someone who owns the questions the owner will bring next: whether to open a second location, why cash is tight in a good quarter, which customers are worth keeping. That person does not have to be the most senior one in the building. If advice stays a partner’s job, a firm can only advise as many clients as its partners have hours.

Where people learn to make those calls, once they no longer spend years preparing reports, is a fair question. Wouter Born takes it up in who trains the next controller. Firms that answer it can advise every client. The rest will ration advice to the clients who pay most.

Measure what changed, not what shipped

What a team measures tells you what it thinks its job is. The classic finance measures all count production:

• days to close;
• reports delivered on time;
• accuracy;
• in firms, fees or clients per professional.

They should stay, as a floor. A late or wrong report still loses a client. But a team can meet every one of them and never change a decision.

The harder measure is whether finance changed what the business did. Did the CEO come to finance before a decision, or after it? Was a recommendation made and followed? Was the cash squeeze seen in time? None of that fits neatly on a dashboard. But a team that never asks has told its people that the report is the job, and they will believe it.

For firms, I expect the fee to follow the measure. A practice that counts only reports delivered will be priced as if reports are all it sells.

Chris Meyer is a chartered accountant and a partner on Born Capital’s Investment Committee. He was CEO of Excel4apps when insightsoftware acquired it in 2018.