Data

Saturday nights on a spreadsheet chasing €1,200 that wouldn't add up

A restaurant owner describes his Saturday night to me, once service is over: instead of heading home, he opens a spreadsheet and compares. What the till recorded, what actually landed in the bank account, what the invoices say. Three sources that should tell the same story, and never quite agree. That month, he was chasing €1,200 that wouldn't add up, and it took him the whole evening. "I'm a restaurant owner, not an accountant," he told me. He was right, on every count.

Almost nobody talks about this reconciliation work, and yet plenty of businesses live it. Taking payment is easy; making sure everything you've taken has actually arrived, at the right amount, with nothing lost along the way, is a different trade. Done by hand, it's slow, it's tedious, and it's exactly the kind of task where a tired eye lets through precisely the error you were looking for. This is exactly where a machine beats us.

Three sources, one story

What we put in place automatically reconciles the three: till takings, bank movements, invoices. When everything matches, nobody has to do anything, and that's the case most of the time. When something doesn't add up, the system flags it, with the amount and the exact place where it's stuck. The restaurant owner no longer searches for the needle in the haystack: we show him the needle directly, and he decides what to do about it.

20min / month

A Saturday night a month on a spreadsheet became twenty minutes of checking. Not because we removed the control, but because the machine does the reconciling and only shows him what deserves his attention.

The discrepancy we eventually explained

The famous €1,200 from the start, we found it, and it wasn't theft, nor a till error. It was a timing gap: a batch of card payments credited by the bank several days late, straddling two months. The money was there all along, it had just arrived "late" from an accounting point of view. Seen by hand, it looked like a hole. Seen by a system that tracks payments over time, it was clear within seconds. Most of the "discrepancies" that ruin shopkeepers' evenings are of this kind: not disasters, just timing gaps we simply don't have the means to untangle by eye.

What it protects, beyond time

Getting a Saturday night back every month is already a lot. But the real stake lies elsewhere. When nobody actually reconciles the three sources, things slip through unnoticed: a payment taken that never reaches the bank, an invoice paid twice, a subscription you keep paying for a service you no longer use. Taken one by one, these leaks are small. Added up over a year, they end up weighing heavily. A regular reconciliation is the check that catches them while they're still small.

The restaurant owner, for his part, gained above all something he couldn't quite put a name to: the peace of mind of knowing his figures add up. Before, there was always a nagging doubt at the back of his mind, that little voice whispering "what if I got something wrong somewhere." Now, when everything matches, he knows it, and when something's off, he's shown it. That constant doubt is a fatigue nobody ever counts, and yet it's one of the heaviest for an owner to carry.

The failure: at first, there were too many alerts

The first version flagged everything, including one-euro discrepancies that didn't matter, roundings, normal delays between taking a payment and its being credited to the bank. The restaurant owner found himself swamped with alerts that mostly called for no action at all. An alert that fires for nothing, repeated often enough, ends up in the bin along with the real ones. So we adjusted the setting: we let normal noise through, and only surface discrepancies that are genuinely out of the ordinary. A good control system isn't the one that talks the most, it's the one that talks at the right moment.

Worth remembering. Before automating a reconciliation, you need to know what's "normal" for your business: what's the usual gap between taking a payment and it reaching the bank, what roundings do you let through. Without that framework, the system alerts you over nothing. With it, it becomes a real safeguard. The rule holds for any control: first define what's normal, only then chase what's abnormal.

What you can do alone this month

Take a month that's already passed, and do the exercise once, by hand. On one side the total taken by the till, on the other what actually landed in the bank over the period. If the two add up within ten minutes, good, your process is sound. If you can't manage it, or if you find a discrepancy you can't explain, you've just put your finger on a real issue, and on the time you're losing to it without counting it.

Putting these figures back in order and getting your evenings back is the territory of my work on data and steering. To measure how much time this kind of task really costs you, read how many hours you lose without counting them; and if keeping watch on your figures speaks to you, it pairs with keeping watch on your prices against competitors.