A corporate accounting role shows you one company in depth. Running a practice shows you twenty at once, every month, which is a different and unusually good education.
From March 2021 to January 2025 I ran Sunny Day Bookkeeping, a remote accounting and bookkeeping consultancy, alongside my full time work. It grew to more than 20 small business clients across a range of industries, with a distributed team of bookkeepers and payroll specialists, and the client base grew roughly 30% a year.
Here is what I took from it.
Seeing twenty businesses at once
The pattern recognition you get from closing the books for twenty small companies in the same week is not available any other way.
What varies by industry is less than you would expect. Seasonality, margin structure, and how quickly customers pay differ a lot. The underlying failure modes barely differ at all. The restaurant, the contractor, and the professional services firm all struggle with the same three things: knowing where they stand before the month is over, following up on money they are owed, and keeping the compliance calendar from ambushing them.
What varies enormously is the owner's relationship to their own numbers. Some could tell you their gross margin from memory. Others genuinely did not know whether the business had made money that year until someone told them. That difference had almost nothing to do with the size of the business or the sophistication of the industry.
What owners actually need
Not a tidy set of financials.
This took me longer to accept than it should have. I would deliver a clean, correct, well presented monthly package, and it would go unread. Not out of ingratitude. It just did not answer the questions the owner had.
The questions are: was this month good or bad, can I afford this decision, and what is about to become a problem.
A correct package delivered silently answers none of those. An imperfect package plus a twenty minute conversation answers all three. Once I made the review conversation the product and the statements the supporting material, the value of the engagement changed completely, and so did client retention.
The compliance calendar is the real product
Sales tax, payroll tax, 1099 and 1095 filings.
Nobody hires a bookkeeper because they are excited about filings. But missing one is the single most damaging thing that can happen in a small business's back office, and the damage is disproportionate: penalties, interest, letters that frighten the owner, and time spent fixing something that took ten minutes to do on time.
The thing that makes this work is a calendar that does not depend on anyone remembering. Every obligation, every client, every deadline, with the preparation work scheduled backward from the due date rather than the due date itself being the only entry. If your calendar tells you a filing is due today, it has told you too late.
Systems as the delivery mechanism
Standardizing on QuickBooks Online and Bill.com across clients was what made a distributed team possible.
The temptation is to accommodate whatever each client already uses. It seems client friendly. In practice it means nobody on your team can cover for anyone else, every engagement requires specific knowledge that lives with one person, and quality varies by who happens to be assigned.
A consistent stack means a reviewer can pick up any client's file and know where everything is. That is worth more than the goodwill you get from accommodating a client's existing tooling, and most clients do not actually care as long as the migration is handled for them.
Managing a distributed team
Bookkeepers and payroll specialists across locations, none of us in a room together.
Documented procedures are what substitute for proximity. When people sit near each other, knowledge transfers by osmosis and nobody notices how much of the process is undocumented until someone leaves. Remote removes the osmosis, which sounds like a disadvantage and is actually clarifying: it forces you to find out immediately what is written down and what is not.
The other thing that made it work was making review a delivery step rather than a spot check. Every client's monthly output gets reviewed before it goes out, always, as part of the process. Not sampling. That sounds expensive and it is cheaper than the alternative, because errors caught internally cost a correction and errors caught by a client cost trust.
Growth and its constraint
Roughly 30% annual growth in the client base sounds good and it exposes things.
The constraint is not capacity in hours. It is that beyond a certain point, adding clients requires changing how the work is done rather than adding more of it. The processes that work at ten clients start to fail at twenty, not because anyone is overloaded but because the informal coordination that held it together stops scaling.
That is the moment where a practice either builds real process or quietly caps out. It is also where I learned most of what I now apply to close process design, because the problems are structurally identical: a repeating cycle with deadlines, dependencies, and knowledge that tends to concentrate in individuals.
Why it ended
I closed the practice in January 2025.
Not because it stopped working. Because I was running it alongside a full time accounting management role and building Hawk Eye AI, and attention is the genuinely scarce resource. Doing three things adequately is worse than doing two of them properly, and the practice was the one where my particular skills added the least marginal value over a good alternative.
I do not think that is a failure and I would not frame it as one. Knowing when a thing has served its purpose is a legitimate outcome.
What carried forward
The most valuable thing was a diagnostic instinct that I now use constantly.
Small businesses lose work to process, not effort. Almost never to a competitor who was better. To a phone that rang while everyone was on a job. To a quote that went out three days late because the evening got away from someone. To a follow up that nobody had time for.
That observation is the entire idea behind Hawk Eye AI, and I would not have arrived at it from inside a corporate finance function. You only see it by looking at twenty small businesses from the inside, month after month, and noticing that the same thing keeps happening to all of them.