At the lending business where I ran credit decisioning, we knew within hours if something was drifting. Approval rates, default signals, the shape of the loan book — all of it was watched in near real time, because software was committing real money every few minutes and a problem left running for a week was a problem multiplied by a week.
Then I ran a restaurant, and I lived the other version.
A margin dip would start quietly — a supplier price creeping up, a portion size drifting, a shift pattern that no longer matched the covers. And I would find out about it when the month's figures were finally assembled, weeks after the fact. Sometimes weeks after that, when I'd finished digging into why.
Same person. Same analytical training. Completely different outcome. The difference wasn't intelligence or effort. It was infrastructure.
The three lags
When a month goes wrong in a small business, the cost isn't one number. It's three delays stacked on top of each other.
Detection lag. The gap between the problem starting and anyone noticing. If your margin only becomes visible when the bookkeeping is done, a dip that starts in week one runs unchecked for five or six weeks minimum. Whatever it's costing you per week, multiply accordingly.
Investigation lag. Noticing "the month was bad" is not the same as knowing why. If the answer lives across your till, your ordering platform, your accounting software and a spreadsheet someone rebuilds by hand, the investigation is days of work — which means it happens slowly, or not at all. I've asked many owners how they got to the bottom of their last bad month. The most honest answer I've heard: "we never fully did."
Response lag. Even once you know the cause, the fix takes effect from now — not from when the problem started. Every week of the first two lags is a week the fix can never claw back.
A problem that took one bad decision to create can take a full quarter to stop.
Add them up, and in a business running on 5–10% margins that arithmetic is brutal.
The fix isn't more data. It's fewer numbers, sooner.
The instinct is to solve this with more reporting. More spreadsheets, more detail, a bigger month-end pack. That makes it worse — it adds hours of manual assembly and pushes the numbers even further from the moment they describe.
What actually collapses the three lags is small and boring: a handful of numbers — five or six, not forty — that arrive automatically, weekly or daily, without anyone building them. Sales against the same week last year. Gross margin. Labour against takings. Repeat versus new customers. Whatever the five are for your business.
Not because those numbers answer every question. Because they tell you when to start asking — in week one instead of week seven.
Where to start (before buying anything)
Don't start with software. Start with a list.
Write down the five numbers you most wish appeared in front of you every Monday morning without you asking anyone. Then, for each one, check where it currently lives and whether that system can export it.
That list is the entire specification for fixing this. It makes any conversation with any provider — us included — shorter, cheaper and harder to oversell.
Big companies pay entire teams to keep decision latency low. A fifteen-person business can now get most of the way there with a build measured in weeks. The gap was never about size. It was about tooling — and the tooling finally fits.
When your last bad month happened — how long did it take you to find out, and who told you?
Delphi Decide builds exactly this: the five numbers, automated, in front of you every Monday. If you'd like to talk it through — get in touch.
