In most growing businesses, the Monday sales report is somebody’s job. They pull an export, paste it into a spreadsheet, fix the bits that look odd and email it round. It takes a couple of hours, it depends on one person, and when it’s late or wrong, nobody finds out until a decision has already been made on it.
We’ve automated exactly this inside a multi-channel UK business selling online and to trade. The report now builds itself every Monday and lands as a draft email for a person to check and send. Here’s how it works, and the one thing that matters more than the AI.
Start with the rules, not the tool
Before anything was automated, we wrote down in plain English what “sales” meant. Which orders count, which date they count on, what’s left out, VAT, how channels are split and who owns each customer. It sounds dull. It’s the most valuable hour in the whole project.
In our case, writing the rules down showed that an existing report everyone trusted had been missing a clean-up step. Fifteen customers who had moved to ordering online were still being counted as trade customers: a five-figure amount of sales credited to the wrong team. No AI found that. Writing the rules down did.
“I’m going to explain how we work out our weekly sales figures. Ask me one question at a time until you can write a one-page rules table: what counts as a sale, which date we use, what’s excluded, VAT, how channels are split, who owns each customer, and what must always add up. Flag anything I’ve left vague.”
If you haven’t yet got everyone working from one set of numbers, start with our guide to getting one set of numbers you trust.
The five checks that stop bad numbers going out
Automation without checks just gets you wrong numbers faster. Our weekly report won’t produce a draft unless every one of these passes:
1. The data is fresh. If it doesn’t reach last Sunday, it waits and tries again later in the day, then tells the owner rather than sending old numbers.
2. One source only. It reads one named file, never an old copy.
3. The sums add up. Channels add to the total and salespeople add to trade sales, to zero difference.
4. It’s a draft. A person always reads it and presses send.
5. It logs every run, so you can see what was reported and when.
Before we trusted it, we recounted a week independently. The totals matched to within a penny across hundreds of orders.
What it costs you today
Typical example: one person spending 4 hours a week pulling, checking and sending sales reports, at £20 an hour, is about £4,160 a year, before counting the time spent arguing about which figure is right. Swap in your own numbers.
Use a business account for your AI assistant and check its data settings before sharing customer or financial details.
Get the full playbook
This is one part of our new playbook, Numbers You Trust Every Monday. It covers the weekly sales report, a daily sales file for your team, weekly KPIs with commentary and monthly profit by channel, with 16 copy-ready prompts, every check we built in and a 30-day rollout plan. All of it is running in a real business today.
We’re not IT people. We’ve sat in the MD’s chair and built the systems behind it, and we’ll talk to you in plain business terms.
Rather have it built for you? Our Sales Report Setup does it on your own data in about two weeks. And if the numbers are a symptom of a bigger growing pain, take the free Growing Pains Check or look at our Growing Pains Diagnostic.
Leave a Reply