The Doorman Fallacy and the CFO

The man at the door

Rory Sutherland, the British advertising executive, tells a story in his book Alchemy that every business owner should hear before signing their next software contract. A consultant walks into a hotel and studies the doorman. On paper, the job is opening the door. An automatic door does that for a fraction of the cost. Cut the role, book the savings, move on.

What the spreadsheet missed is everything else the doorman did. He greeted regulars by name. He hailed cabs in the rain. He noticed who didn’t belong in the lobby. He made the hotel feel like a place worth paying for. Remove him and the door still opens, but room rates and repeat guests quietly slide, and nobody connects the drop to the cost cut that caused it.

Sutherland called this the doorman fallacy: defining a role by its most visible task, automating that task, and losing everything you never measured. A recent piece by Gediminas Lipnickas of the University of South Australia shows it playing out in AI adoption right now. Commonwealth Bank of Australia replaced 45 customer service staff with a voice bot, then reversed course and admitted the roles “were not redundant.” Taco Bell’s own technology chief has said AI-only drive-throughs may not make sense. And one survey found up to 55% of companies that replaced people with AI now say they moved too fast.

I think finance is next in line for this mistake.

Financial statements are the opened door

Ask a consultant to describe what a CFO does and the answer will sound a lot like the doorman’s job description: produce the monthly financial statements, a budget, maybe a cash forecast. Visible, measurable, and increasingly automatable.

And it should be automated. A clean close, a P&L, a balance sheet, a first pass at variances: AI tools can now produce those faster and more consistently than a person keying entries. If your finance function spends most of its hours assembling reports, you’re overpaying for that time. I tell clients exactly that.

But a financial statement without context is an opened door. It’s the visible output of the job, not the job. The numbers can tell you gross margin fell three points. They can’t tell you it fell because your largest customer is renegotiating freight terms, that your banker will notice at the next covenant test, and that the fix is a conversation with your logistics vendor before the quarter closes. Someone has to read the numbers, know the business well enough to see when things are going off script, and then do something about it.

That someone is the job.

The CFO is the general contractor

The best analogy I have for a CFO is a general contractor. A GC doesn’t pour the foundation or wire the house. They know who should, in what order, and what goes wrong when the electrician shows up before the framing inspection. Their value lives in relationships and judgment, not in the hammer.

A good fractional CFO plays that role across the administrative side of the business. They know your banker and how to frame a line of credit request so it gets approved. They know your insurance broker and when to push back on a renewal. They manage the payroll provider, the software vendors, the tax preparer, and the auditor if you have one. Inside the company, they lead and develop the people doing the daily accounting work. And they sit with leadership as an advocate for the business when the decisions get hard.

None of that shows up on a monthly report. All of it shows up in your results. It’s the doorman hailing the cab and knowing the regulars by name. AI can draft the email to your banker. It can’t be the person your banker trusts.

The rollup trap

This matters now because of who is buying accounting firms. Private equity has been acquiring advisory CPA firms at a steady pace. A common playbook splits off the audit practice, which is regulated and has to stay under CPA ownership, from everything else. The non-audit side gets standardized, offshored, or automated as far as it will go, then sold back to small businesses at a low monthly price.

Not every PE-backed firm runs this way, but the economics push hard in that direction. And on paper, it looks like the automatic door. The statements still arrive. They may even arrive faster. What’s gone is the person who knew your business, noticed the margin slip, and called your banker before your banker called you.

The cost doesn’t show up right away, which is exactly why the fallacy is so persistent. It shows up two years later as a credit line that wasn’t ready when you needed it, a pricing problem nobody flagged, or an acquisition offer you weren’t prepared to evaluate. Lost options and slower growth compound. Over time they outrun whatever you saved on a cheap subscription, usually by a wide margin.

Where the job actually begins

None of this is an argument against AI. It’s the opposite. The answer isn’t a CFO who ignores the tools. It’s a CFO who is fluent in them.

When a fractional CFO uses AI well, the analytical work (the close, the reconciliations, the reporting, the first read on variances) gets done quickly and accurately. That doesn’t shrink the role. It moves the hours. Less time assembling, more time interpreting, advising, and working the relationships that move the business. It also means one experienced CFO can support more companies, and larger ones, than before. That makes real CFO judgment affordable for businesses that used to go without it.

There is one prerequisite, and it isn’t optional: data quality comes first. AI on a messy chart of accounts produces confident nonsense at high speed. Clean data, consistent processes, and a reliable close have to be in place before the tools pay off. With the right person leading it, that groundwork takes months, not years, and the companies that do it now will be the ones ready to use what comes next.

Once the analytics are handled, that’s where the CFO’s job begins. Not where it ends.

Don’t settle for the door

Small and midsize businesses sell themselves short when they assume a cheap automated service delivers what a CFO does. It delivers the door. When you evaluate finance help, ask three questions. Who will read my numbers and tell me what they mean? Who picks up the phone when my banker, insurer, or payroll provider calls? And how are you using AI to spend less time producing reports and more time on my business?

That’s how we’ve built Primetrics. We put AI to work on the analytics so our CFOs can spend their time on the part of the job that was never about opening the door. AI to get your team back, never to replace it.

If you’re wondering whether your current setup gives you the doorman or just the door, I’m glad to talk it through.

Sources

• Gediminas Lipnickas, “The doorman fallacy: why careless adoption of AI backfires so easily”, world.edu, November 2025 (cites Orgvue, McKinsey, Goldman Sachs, and the Commonwealth Bank and Taco Bell cases)
• Rory Sutherland, Alchemy: The Surprising Power of Ideas That Don’t Make Sense (2019)