Bookkeeping keeps score.
Advisory changes it.
You already know which part of your company is expensive. This is the practice that goes in and fixes it: the process nobody has looked at in years, the software nobody adopted, and the work a machine should have been doing all along.
Everyone is talking about AI. Almost nobody has it doing actual work.
The gap is not access. You have the same tools everyone else has. The gap is that nobody has sat down, picked a specific job a machine should own, built it properly, and then checked a month later that it is still running.
The phone
A voice agent that answers, routes, and handles the questions it should handle. Scored on every call, so you can see how it is actually doing instead of hoping.
Meetings
Recorded, transcribed, and turned into assigned action items automatically. Nobody takes notes. Nothing gets dropped because the person who was going to remember did not.
The weekly numbers
Reports that assemble themselves. Your team spends the time reading them and deciding, rather than rebuilding the same file every Monday.
Approvals and requests
Purchase requests, artwork, contracts, travel. Submitted on a form, routed to whoever decides, tracked until done. Nobody chases anybody.
The busywork nobody owns
Reconciling two systems that disagree, cleaning the data, rebuilding the report by hand every month. The jobs that are too small to hire for and too annoying to keep doing.
The part most people skip
Checking it still works. An automation that quietly broke in March is worse than no automation, because everyone stopped looking.
The test is not whether it is impressive in a demo. The test is whether it is still running when nobody is watching.
Six expensive problems, and what you have instead when they are fixed
Pick the one that sounds like your company. If two of them do, that is normal, and they are usually the same problem wearing different clothes.
Efficiency you can feel
Everything takes three steps more than it should.
The work mapped, the duplicated effort cut, the handoffs closed. Usually the same team, doing more, without anybody working later. This is the one that pays for itself first.
A vendor stack that earns its keep
Nobody has read the recurring invoices in three years.
Seats for people who left, two tools doing one job, an auto-renew nobody remembers signing. Cancelled, consolidated, renegotiated, with the contracts actually read before they renew.
Leadership that holds
Your best technician now runs people and nobody taught them how.
Managers who can run a one-on-one, handle an underperformer, and make a call without routing it to you. Turnover is a bill you are already paying. This is how it stops.
Systems people actually use
You bought the platform. The team still runs the spreadsheet.
A rollout that finishes: configured for how you really work, taught to the people who have to live in it, and adopted. You stop paying for software twice.
Numbers that tell you what to do
The books are accurate and still say nothing useful.
Margin by line of business, where the cash actually goes, which customers are worth what, and which cost is quietly compounding. Accurate is the floor. Legible is the point.
A company that runs without you in the room
Every decision routes through you, because nothing is written down.
The work documented, owned by a name, and automated where automation is honest. The test is a two-week vacation where nobody calls.
The work is boring to look at. That is how you know it is real.
Nobody photographs a renegotiated contract or an approval that routes itself. Here is the actual shape of it, using the most ordinary example there is: somebody needs to buy something.
You get the artifact, not a summary
The written diagnostic, the process map, the playbook, the form, the routing rules. Documents you keep and could hand to someone else, whether or not we work together again.
It is measured before and after
How many of these happen a month. What the median one costs. How many people are involved. How long it takes today. Without that, "we improved the process" is just a feeling.
It survives you not watching
Documented, assigned to a name, and checked a month later to confirm it is still running. Most improvements quietly stop within six weeks. Checking is the difference.
Every one of these started as something everyone already sort of knew.
The numbers are what turned it into a decision. Four examples from operations work I have done. The companies are not named, and none of them are bookkeeping clients.
Approvals 79 purchase requests over eight months. Eleven a month, median $582, four above $5,000, and thirteen different people submitting.No approval trail on any of them. That is not a spending problem, it is a visibility problem. One form now, routed to one approver, every decision timestamped and searchable.
Vendor contracts A vendor's standard fifteen-section agreement, the kind most customers sign as-is. Read clause by clause, it came back with twenty changes.Payment moved from $1,500 up front to $750 twice a month. The no-hire term halved from twenty-four months to twelve. The liability cap went from three months of fees to the greater of $18,000 or twelve months. The contract nobody reads is the one that costs the most.
Ad spend Branded search took 26.7% of the budget and produced 74% of the ad revenue, at 11.9 times return. Non-branded took 73.3% and produced 26%, at 1.5 times.Three quarters of the budget was buying a quarter of the result. Not a creative problem, an allocation problem, and completely invisible until someone splits the report in two.
Customer mix The top fifty accounts were 4.2% of the customer list and 40.7% of revenue. Thirteen accounts were 1.1% of the list and 22.5% of revenue.The largest accounts had a lower average order than the tier below them. Their value was frequency, not size, which means the growth play was consistency rather than upselling. The opposite of what everyone assumed.
None of this needed new software. It needed somebody to sit down with data that was already there and that nobody had time to open.
Advice is cheap. Execution is the product.
Anyone can tell you what is wrong. You are buying the part where somebody stays until it is different.
What you are actually buying, in order
I get in the weeds
Your meetings, your contracts, your numbers, and your managers without you in the room. You get a written read on what is working, what is quietly expensive, and the order I would fix it in. Yours to keep either way.
We finish the expensive thing
One item, taken all the way to done, with a written update every week. Not a workstream. A finished thing you can point at.
It holds without me
Documented, assigned to a name, automated where that is honest. Then a straight conversation about whether there is a next thing. Sometimes there is not, and saying so is part of the service.
This one is me
The bookkeeping practice runs on a team, which is the point of it. Advisory does not. You get the same person in the first call, in the diagnostic, and in the room at week eleven.
A background across corporate, payroll, marketing, B2B sales, inventory, ERP and systems implementation, and businesses of my own. Everything on this page is work I do, not work I have read about. The voice agent, the approval workflows, the reporting that builds itself, the contracts renegotiated before they auto-renewed: those are jobs I have done, recently, at a company that has been operating since the nineties.
Mostly that shows up as knowing which problems are worth your money and which ones quietly fix themselves if you leave them alone.
Three ways in, in order of commitment
Start at the top. Stopping after the first one is a legitimate outcome and a cheap way to find out.
Diagnostic
Two weeks inside the business, then a blunt written verdict and a repair sequence in priority order. You can hand it to someone else and it still works.
Scoped project
One outcome with a finish line: a software rollout that lands, a vendor stack renegotiated, an SOP library that people use.
Fractional
Standing operator capacity and real ownership of initiatives, for when the list is longer than one project and hiring a full-time executive is premature.
Written budget first. Overages approved in advance, always.
Every invoice itemized. Fixed fee or retainer, never an open meter.
No retainer to talk. Sometimes the answer is do nothing yet.
Not a CPA firm. We do not file tax returns or perform attest work.
Where I would point you somewhere else
Not a wall. Just the four situations where hiring me would be a waste of your money, listed so you can find out now rather than in month two.
You need a deck for the board.There are people who make beautiful ones. I make the thing the deck would have described.
The decision is already made and needs agreeing with in writing.You want cover, which is a real need. It is just not this one, and I would be bad at it.
Nobody in the engagement can say yes.If the sponsor cannot authorize a change, the work stalls at week three and you paid for a diagnosis you already had.
The books are the actual problem.Then start with the other practice. It is cheaper, it is faster, and it might be the whole fix. Start with bookkeeping.
Tell me what is not working.
A straight read on whether this is an advisory problem, a bookkeeping problem, or something that resolves itself in a quarter if you leave it alone.