As of September 16, 2026, I had built 96 automated jobs for this business and switched off 30 of them. I published that ratio in Automate a Job, Fund the Treasury, and it’s the most useful number I own for this question, because every one of those 96 was a reinvestment. Time and money went in on the belief that something would come back. Nearly a third of the time, it didn’t.
That’s the question this piece answers. When a month closes with money left over, where does the next dollar go? I think it has three possible jobs, and the order you ask about them matters more than the answer to any one.
Every dollar of profit gets a job, whether you assign one or not
The three jobs are reinvest, reserve, and hold. Reinvest means you spend it on something you expect to pay back: a tool, a hire, a machine, a build. Reserve means you park it against a specific thing that could go wrong. Hold means you keep it, on purpose, because you have no better use for it inside your planning horizon.
The catch is that skipping the decision is still a decision. Money that nobody assigns ends up in the operating account, and the operating account is a position too. Or it ends up in a budget line that keeps living after its reason died. When I sat down to publish my own software bill on September 14, my books carried $280 of line items that had never produced a charge. One was a monthly budget for a product that never launched, still sitting there as if it were a running cost. That was a reinvestment I had stopped believing in and never formally ended.
Reserve first, and only up to a named event
I fund the reserve before anything else, and I size it against the worst of three things that could actually happen to this business, not against a round number of months. Operating, Reserve, Treasury walks through that sizing, so I won’t repeat it here.
The point for this decision is narrower. A reserve has a ceiling. Once it covers the named event, the next dollar stops being reserve money, and piling more into it is just holding with a reassuring label. If the event that worries you is a machine failing, the equipment replacement fund is the reserve for it.
Reinvest only against a hurdle written in what you buy
Most owners treat reinvestment as the default and treat the hurdle as zero: if it pays back at all, do it. I don’t think zero is the hurdle. The money you’re about to spend has another option, which is to sit there. Sitting there has a cost you can measure in the things you buy. The retained earnings purchasing power calculator takes a balance and two prices for one thing you actually purchase, and tells you what the balance lost. That loss is the floor a reinvestment has to clear.
Then discount your own forecast by your own track record. Mine says roughly a third of what I build gets retired. So when I expect a new build to pay back, the honest version of that expectation is the payback times the share of past builds I kept. Use your own share. If you’ve never counted it, count it before the next purchase, because the count is the one input a vendor can’t argue with.
If the discounted payback still clears the floor, reinvest. If it doesn’t, the money moves to the third job.
Hold is a decision, not a leftover
Whatever survives the first two questions gets held. Most of the time, for most small businesses, that amount is small, and some months it’s zero. Zero is a fine answer.
What I’d push back on is treating the hold as the place money lands when you ran out of ideas. Holding is a choice about what you expect money to do. Holding cash says you expect it to keep its purchasing power. Holding anything else says you expect that thing to do better over a long horizon, and that you can sit through a bad quarter without being forced to sell. Write the reason down either way. A hold with a written reason gets reviewed. A hold without one just accumulates.
The order I use
- Is the reserve funded against a named event? If not, the dollar goes there.
- Does a specific reinvestment clear the purchasing-power floor after I discount it by my own kill rate? If yes, the dollar goes there.
- Otherwise, hold it, and write down why.
Running the questions in this order is what keeps a business out of the common failure: deciding what to hold first, then discovering in a bad quarter that the reserve was never there and the position has to be unwound at the worst price.
Where this breaks
This assumes the month actually produced a surplus you can see. A business that books revenue on net-60 terms, or earns most of its year in one season, won’t know until the cash arrives, and running these questions early will put money to work that isn’t there yet.
It also leaves out a fourth job. Paying down expensive debt can beat all three, and if you carry a balance at a high rate, that question comes before any of this. And the kill-rate discount is only as good as your count. Mine covers 96 builds, in one business, in under a year. It’s a better input than a vendor’s case study, and it’s still a small sample.
None of this is tax, legal, or investment advice, and your CPA should see the plan before the money moves. If you want help putting your own numbers through it, the intro call is free and fifteen minutes.