By Rakibul Hasan, Moonlab
An automated business report pulls numbers from your tools on a schedule and sends a short summary, usually on Monday morning, so nobody builds it by hand. The best ones fit on one screen and answer three questions: how did last week go, what needs attention and what is coming.
This guide shows what to include, how to build the report with the tools you already use, and how to keep it accurate and short. It suits owners and managers who spend an hour a week copying figures into a spreadsheet. If you want a live view instead of a message, see moonlab.space/guides/small-business-dashboard.
Why a short weekly report beats a dashboard nobody opens
A dashboard needs you to remember to look at it. An email or chat message arrives whether you remember or not. For most small teams, a weekly message that takes two minutes to read does more good than a clever dashboard.
The habit matters more than the tool. Pick a day and a time, send it to the people who act on it and review it in a short weekly meeting.
What to include
Choose 5 to 8 numbers. More than that, and nobody reads them. Pick figures that you can influence and that change from week to week.
- Sales or revenue for last week and the week before.
- New enquiries or leads, and how many were answered within a day.
- Quotes sent and quotes won.
- Invoices overdue and the total owed.
- Cash in the bank, if the owner wants it.
- Website visits and top pages, from Google Analytics or Search Console.
- Reviews received and average rating.
- Open support tickets or complaints.
Where the numbers come from
List each number and where it lives: accounting tool, CRM, booking system, shop, ad accounts, Google Business Profile or a spreadsheet. If a figure only exists in someone's head, decide where it will be recorded first.
Most of these tools have APIs or connectors, and tools like Google Sheets, Looker Studio, Zapier, Make and n8n can read them. Check that your plan allows data access, because some limit it.
How to build it, step by step
Keep the first version simple. A common pattern is a scheduled workflow that runs early Monday, collects the figures into a sheet, adds the previous week for comparison and sends a formatted message.
- Trigger: a schedule, for example Monday at 7:00 local time.
- Collect: pull each number from its source.
- Calculate: change versus last week, as a number and a simple arrow or word.
- Format: a short email or chat message with a link to the detail.
- Send: to named people, with a fallback alert if a source fails.
Adding an AI summary, carefully
An AI step can write two or three plain sentences from the numbers, such as noting that quotes sent are up but wins are flat. This is useful for people who dislike tables.
Give the AI only the numbers, tell it not to guess causes, and keep the table in the message so readers can check the facts. Do not send customer personal data to an AI tool without checking its data terms.
Keep the numbers trustworthy
A report that is wrong once loses trust for months. Agree definitions in writing. What counts as a lead? Is revenue before or after tax and refunds? Use the same definitions every week.
Add checks. If a number is zero or far outside its usual range, flag it as possibly broken instead of presenting it as fact. If a connection fails, the report should say so.
Make people act on it
Add a line for the one thing to fix this week. Review the report in a ten minute meeting. If nobody ever acts on a figure, remove it.
Review the report every quarter. Drop numbers that no longer matter and add ones that do.
A worked example and common mistakes
Here is a made-up example. A physiotherapy practice with three therapists wants a Monday summary. The owner chooses six numbers: appointments attended, appointments missed, new patient enquiries, revenue invoiced, unpaid invoices and new Google reviews.
A workflow runs at seven on Monday morning. It reads the booking system for last week's appointments, the accounting tool for invoices, the enquiry form for new patients and the Google profile for reviews. It writes the numbers into a sheet next to the previous week, works out the change, and builds a short email. The top line says, for example, that missed appointments were up on the week before, in a sentence written from the numbers only. Below sits a small table, and a link to the sheet.
If a source cannot be read, the email says so in the first line instead of showing a zero. The owner reads it in two minutes, picks one action for the week, and the team discusses it at the Monday huddle. After a month she removes the review count, which nobody acts on, and adds average wait time for a first appointment.
Avoid these mistakes.
- Too many numbers, so nothing stands out.
- Unclear definitions that change week to week.
- A zero shown as a real result when a connection failed.
- Sending it to people who never act on it.
- Letting an AI step guess reasons it cannot know.
- No history, so trends are invisible.
- Never reviewing which figures still matter.
How to decide and what to do this week
Use these questions to choose the right numbers for your own report, and to decide when to stop adding more.
- Would a different number change what you do this week?
- Is it available automatically, or does someone have to type it?
- Can your team influence it within a month?
- Is the definition written down and agreed?
- Does it compare to last week or last month, so the trend is clear?
- Who reads the report, and what will they do with it?
How Moonlab can help
We build weekly reports from your accounting, CRM and website tools in your own accounts. See moonlab.space/services/data-and-analytics and moonlab.space/services/operations.
A free 30-minute automation audit at moonlab.space/contact will help you choose the numbers worth tracking.
