What happened to our own P&L when we put AI inside our wholesaler

Most software companies show you a case study about someone else. We can show you our own. Conducta was built inside Sip Champagnes, the wholesale business we run, and we switched it on there before we offered it to anyone. That means the first customer whose P&L we can talk about honestly is ours.

The setup

The figures below come from our audited management accounts. They compare the nine months after we switched Conducta on to the same nine months a year earlier. We publish percentages rather than pounds because the absolute numbers of a private business are commercially confidential; they are available under NDA. Everything here is a real change in a real ledger, not a model.

What the P&L did

The striking part was that it was not one line. The whole statement moved, and it moved on a slightly smaller top line, which means margin did the work rather than volume.

What actually changed on the ground

Numbers like these only mean something if you can explain the operational change behind them. Here is what changed day to day:

• Order processing ran above 95 percent straight-through on priced lines: orders read and written without a person retyping them.
• Order-entry errors fell, which cut credit notes, re-picks, and the quiet loss of customers let down twice.
• Invoice chasing became consistent and polite rather than sporadic, which brought cash in sooner.
• Purchasing got cleaner and supplier terms tighter, which is most of the cost-of-sales improvement.
• The same team handled more throughput, because AI absorbed the manual work rather than the people.

Why it held

A single good quarter proves nothing. The trajectory held through Q1 2026: gross margin stayed above 35 percent nine months after switch-on, the operating run-rate reached break-even and is on track for post-tax profit this year, and order-processing automation kept running above 95 percent straight-through on priced lines. It was sustained, not a spike.

“We stopped believing the P&L could move like this. Then we watched it move, in our own business, and it stayed moved.”

The honest caveats

This is one business, ours, and we are not going to pretend it guarantees yours. The top line was slightly smaller, so this is a margin and efficiency story, not a growth miracle. The percentages are real and audited, but the absolute figures stay confidential. And nine months is long enough to be meaningful without being a decade of proof. We publish this because it is honest evidence the approach can move a real P&L, not because it is a promise.

What this is not

It is not a projection for your business, and it is not a claim that software alone did it; the discipline of running against a real P&L mattered as much as the automation. What it is: the reason we can talk about outcomes from the inside rather than pointing at someone else's logo. If you want to see how the same approach would map onto your own numbers, that is exactly what a walkthrough is for.