Almost nobody wakes up and decides to hire a controller. What happens instead is a slow accumulation of small frustrations — the close is late again, a number looked wrong and took two days to explain, the bank asked for something that took a week to produce — until someone says out loud that this isn't working.
The gap between when a business needs a controller and when it gets one is usually about a year. That year is expensive in ways that don't show up anywhere.
A bookkeeper records what happened. A controller is responsible for whether it's right, whether it's on time, and whether the structure it's recorded in is useful.
The short version: a bookkeeper produces the numbers, a controller makes them trustworthy and timely.
The clearest single indicator. It usually means the process is undefined rather than that the person is slow.
If the owner is the review layer, there isn't one. Every hour you spend checking the books is an hour not spent running the business, and you're the least efficient person to do it.
Multiple entities, multiple locations, several service lines, deferred revenue, inventory, or a new state to register in. Complexity is what controllers exist for.
A loan, a lease guarantee, an insurance review, a buyer, an investor. All of them want reconciled, consistent, timely statements — and the scramble to produce those under deadline is where mistakes get made.
"What did we make on the Henderson job?" or "what's our margin on maintenance work?" should take minutes. If it takes a day of spreadsheet work, the underlying structure is wrong.
Not a hard rule, but around this point payroll complexity, expense approvals and departmental cost tracking usually outgrow what a bookkeeper can hold.
A full-time controller runs roughly $80,000–$120,000 plus benefits — call it $100,000–$145,000 all in. That's justifiable for most businesses somewhere north of $10M in revenue, sooner if the operation is complex.
Below that, the honest answer is that most companies need controller function without controller headcount. The work is real but it isn't forty hours a week, and hiring full-time means paying for capacity you can't use.
The third option is worth naming: sometimes what you need isn't a controller at all, it's a one-off cleanup. If your chart of accounts is wrong and your close has no process, a project to fix both may buy you another eighteen months with the bookkeeper you already have.
Don't hire a controller to compensate for a broken system. Fix the system, then decide whether you still need the role.
The visible change is timing — numbers on the 15th instead of the 23rd. The more valuable change is that you stop wondering whether they're right.
That sounds soft, and it isn't. An owner who trusts their financials makes decisions faster, commits with more confidence, and stops holding a private mental buffer against the possibility that the numbers are wrong. That buffer is expensive, and most owners don't notice they're carrying it until it's gone.