Ask a business owner what they'd like in a weekly report and you'll usually get a list of about thirty things. Ask them what they did differently after last month's management accounts and the answer is often nothing. That gap is the whole problem.
A report is not a record. Its only job is to change a decision. If a number can't plausibly change what you or your team do in the next seven days, it belongs in the annual accounts, not on Monday morning.
So the discipline is subtraction. Five numbers, chosen deliberately, beat a forty-tile dashboard that everyone stops opening by March. Here's the framework we use when we build a weekly brief for a client, and how to apply it to your own business whether you sell hours, products or projects.
Why five
Five is not mystical. It's about what a person can hold in their head while walking to the kettle. A weekly report gets read in ninety seconds by someone who has already got a full day planned. If it takes longer than that to see whether the week needs intervention, it will be skimmed, then ignored, then quietly cancelled.
There's a second reason. Every number you add dilutes accountability. When a report shows five things, someone owns each one. When it shows forty, nobody owns anything and the report becomes weather — commented on, never acted upon.
Write down the sentence that begins: "If this number moves the wrong way, we will…" If you can't finish the sentence with a specific action that could happen this week, the number doesn't go in the report. It's context, not a control.
The five slots
Rather than prescribe five metrics — they differ wildly between a brewery and a recruitment firm — we fill five slots. Every business has all five, whatever it sells.
1. Money in — with a comparison, not just a total
Revenue for the week is nearly useless on its own. Revenue for the week against the rolling four-week average, split by the one dimension that matters most to you, is a decision-maker.
That dimension is the choice that makes this slot work. For a wholesaler it's usually channel — trade versus direct, because the margins differ enormously. For a service firm it's usually service line. For a product business it's usually product or format. Pick one. Not three.
The comparison matters more than the number. £42,300 means nothing. "£42,300, up 9% on the four-week average, and almost all of the increase is one product" means something, and it means it instantly.
2. Money owed — and how old it is
The total owed to you is a much weaker number than the shape of what's owed. Two businesses can both be owed £30,000: one is owed it by twelve customers who all pay in 32 days, the other is owed £22,000 of it by one customer who has been quiet for 90 days. Same total, completely different week ahead.
So this slot wants ageing, not just a balance. In practice: what's overdue, who's the biggest overdue, and how that compares to the last few weeks. A trend that's worsening is a cashflow problem forming, and it's much cheaper to intervene in week three than week eleven.
3. Work committed but not yet delivered
This is the slot most owners are missing, and it's usually the one that predicts the next quarter. It has a different name in every sector: order book, backlog, work in progress, forward bookings, pipeline weighted by stage. Same idea — what have we promised that we haven't yet turned into an invoice?
Two things make it useful. First, its size relative to your usual weekly output, which tells you whether you're about to be too busy or too quiet. Second, its age: work that has been committed and hasn't moved for weeks is usually stuck on something small that nobody has escalated.
4. Capacity or stock — whichever constrains you
Every business has one thing that runs out first. For a manufacturer it's stock or production slots. For a professional firm it's available hours in the right skill. For a venue it's covers or space.
Express it as cover, not as a level. "We have 4,000 units" doesn't tell you anything. "At the current run rate, that's 2.1 weeks" tells you whether to act on Friday. The same trick works for people: not "we have six technicians", but "we have eleven days of scheduled work in a five-day week", which is a hiring conversation rather than a headcount fact.
5. One thing that's drifting
The fifth slot is deliberately not a metric. It's an exception — the single most notable thing that's slipped, chosen from a wider set of checks. A good customer whose ordering has stopped. A quote sent three weeks ago and never followed up. A job that has sat in the same status since the start of the month.
This is where reporting turns into management. Slots one to four tell you how the business is doing; slot five tells you what to do about it. If your report has no slot five, it's a scoreboard, and scoreboards don't change games.
Getting the numbers to agree with each other
Here's the practical obstacle. Most businesses can produce all five numbers — but from five different places, in five different formats, with three different definitions of "sales".
The accounting system knows about invoices, not orders. The till or web shop knows about transactions, not customers. The job system knows about work, not money. Nothing knows about all of it, so somebody spends Friday afternoon in a spreadsheet stitching it together, and every version has slightly different totals to the last one.
That's why the sequence matters: agree the five numbers first, then join up the data that produces them. Doing it the other way round — connecting everything and hoping insight falls out — is how businesses end up with an expensive dashboard nobody trusts.
Before automating anything, write one sentence per number defining exactly what counts. Does "sales" include VAT? Does it include the trade discount? Is a deposit revenue when taken or when delivered? Two people in the same business often answer differently, and until it's written down, every report will be argued with rather than acted on.
Make it push, not pull
A dashboard requires somebody to remember to look. A report arrives whether they remember or not. Given the choice, always choose the one that arrives.
Email at a fixed time on a fixed day beats a login. It gets read on a phone before the first meeting. It can be forwarded. It's archived automatically, which means you get a history of what you knew and when — surprisingly useful when you're trying to work out whether a problem was visible before it became expensive.
The dashboard still earns its place for digging into a number after the report has raised it. But the report is what makes the reporting habit stick.
Then make it explain itself
A number that has moved raises a question: why? This is the part that used to require a person with time and system access, which is precisely why it usually didn't happen.
It's also the part that has genuinely changed in the last couple of years. Given joined-up data, a language model can write the two sentences of context that turn a number into a decision — that the increase is concentrated in one product, that three accounts stopped ordering, that the overdue total is one customer rather than a general slide. Not a forecast, not advice. Just the "because" clause, written before you've had your coffee.
That's the whole ambition of a good weekly report: the numbers that matter, the reason they moved, and the one thing worth chasing — short enough to read while the kettle boils.