Weekly sales report with AI: the Monday summary
A manager who builds the weekly report on Sunday night arrives at the meeting tired and with numbers that do not match other teams'. How to set up a weekly summary that the CRM calculates, AI writes and the manager reviews in minutes.
The Monday report with AI is a one-page document that lands in the inbox before the weekly meeting, produced by a three-part routine: the CRM calculates the numbers with fixed filters, the AI writes the text from those numbers, and the manager reviews and sends it. The summary says what changed during the week, where the risk is and what leadership needs to decide, and the part that writes itself is the assembly. If you lead sales, your decision is to define the report’s fixed sections, the definition of each number, who receives each version and who reviews it before it goes out. Once that is done, the manager gets Sunday night back and the Monday meeting starts with discussion, without anyone reading numbers aloud.
When I sit in on a sales leadership team’s Monday meeting, the first fifteen minutes usually go to reconciling numbers. The sales manager brings one pipeline total, marketing brings a different count of opportunities generated, and finance asks why neither matches revenue. Each person built their own spreadsheet the night before, with different filters. The last-minute assembly habit shows up even in software usage data: in its June 2025 Work Trend Index, Microsoft reports, from Microsoft 365 telemetry, that PowerPoint edits spike 122% in the final ten minutes before a meeting. An automated report removes the assembly work and forces the company to choose one definition for each number.
What should a weekly sales report include?
A useful weekly report answers a few questions, always the same ones, in the same order. When the format changes every week, nobody compares one week with the next and the report becomes optional reading. I use five sections that fit on one page.
| Section | Question it answers | Source | Rule |
|---|---|---|---|
| Results | How much did we sell this week and month to date, against quota? | CRM, won deals | Same definition of “won” as finance |
| Pipeline | How much came in, how much advanced, how much left? | CRM, stage changes during the week | Amount and count, by stage |
| Risks | Which large deals stalled, moved back a stage or changed date? | CRM, deal history | List with deal name, amount and reason |
| Forecast | What changes in the month and quarter number? | Forecast category in the CRM | Only deals that meet the category criteria |
| Decisions | What does leadership need to decide this week? | Manager | Written by a person, with owner and deadline |
The last section gives meaning to the other four. A report with no decisions is informational, and leadership reads it, agrees and moves on. With two or three decisions written at the end, such as “approve a discount above policy on deal X” or “reassign the accounts of the salesperson who left”, the Monday meeting already has an agenda.
The rule in the right-hand column is the one that takes the most work in the first week. “Pipeline” needs a written definition, with the stages included, the cutoff date and how deals with no amount are treated. Without that, the AI will fluently write a number each team understands differently. The tool for this is the data dictionary, and the criteria for forecast categories are in sales forecasting.
How do you automate the weekly report with AI?
The automation has a clear division of labor: the system calculates and the AI writes. Asking an AI to add up values from a table pasted into a chat is the shortest path to a wrong number stated with confidence. The CRM or the spreadsheet does the math, and the AI receives finished results to turn into text.
- Write the definitions. For each number in the five sections, one line with what goes in, what stays out and the cutoff date.
- Build the reports in the CRM with those definitions and save them under fixed names. Most CRMs on the market, such as HubSpot, let you schedule reports to be sent by email.
- Schedule the export for Sunday night or early Monday, always with the same period: Monday to Sunday of the previous week.
- Generate the text with AI. The manager pastes the results into the assistant the company has approved, with the prompt below, or uses a routine that does it automatically. People who work with exported files can use an agent like the one described in Claude Code for managers.
- Review in five minutes. Check total sales and total pipeline against the CRM, read the risk list and write the decisions section.
- Send it at the same time every week, before the meeting, so everyone arrives having read it.
In the first weeks, the manager spends more time adjusting definitions than reviewing text. That time pays off, because each definition settled applies to every following report and to the sales dashboard.
The instruction to avoid evaluative adjectives looks like a detail and changes the tone of the document. A report that says “excellent week” invites celebration, and one that says “R$ 480k won against a weekly quota of R$ 400k”, with illustrative numbers, invites the question of which deals explain the difference.
How do you stop AI from getting the report’s numbers wrong?
The first safeguard is the division of labor already described: the AI writes on top of numbers calculated outside it. The second is checking totals. Before sending, the manager compares three numbers with the CRM: sales for the week, total pipeline and the largest deal on the risk list. If all three match, the rest is likely right, because it came from the same export.
The third safeguard is in the prompt itself: asking the AI to point out inconsistent numbers. A week in which pipeline grew and no deal came in signals a filter error or a bulk change to amounts, and the AI usually catches that kind of contradiction when instructed to look for it. The mechanics that lead AI to state confidently what is not in the data are in AI hallucination.
The hardest error to catch comes from the records: deals with no updated close date, amounts nobody has reviewed since the proposal, stages changed without criteria. The automated report exposes these problems every Monday, and the list of fixes becomes a task for the team, the subject of CRM data quality.
What CRM data can go to the AI for the report?
The weekly report works with aggregated numbers and deal names, which in B2B sales are usually company names. That sharply reduces the amount of personal data involved, and the care goes into the details: the risk list does not need the contact’s name, email or phone number, and the recorded reason should not carry comments about people. The export that feeds the AI includes only the columns the text uses, and the tool is the business version approved by the company. The rules of Brazil’s data protection law as applied to this use are in AI and LGPD.
Who should receive the report, and in which version?
The same set of numbers serves different readers, and each one needs a different cut. AI makes this easier, because generating three versions from the same data costs one more prompt.
| Reader | What they need to see | Length |
|---|---|---|
| CEO and executive team | Results against quota, quarter forecast, pending decisions | Half a page |
| Sales director | All five sections, by team | One page |
| Marketing | Opportunities by source, conversion to meeting, disqualification reasons | Half a page |
| Post-sale | Deals won this week, with what was promised and the start date | List |
| Salespeople | Their own portfolio: stalled deals, overdue next steps | Individual list |
The post-sale version prevents the most trouble. The team that will implement or serve the customer learns about won deals the same week, with a summary of what was sold, and the handoff no longer depends on someone remembering to tell them. The marketing version feeds the lead-quality conversation with the same number sales uses, which cuts down the argument over who generated what.
How do you use the report in the Monday meeting?
The report changes the pipeline meeting when everyone arrives having read it. The manager opens with the decisions section, goes through the risks with the owner of each deal and closes with the revised forecast. Numbers everyone has already seen stay out of the discussion, and the time goes to the deals that need help.
Two rules sustain the habit. The report always goes out at the same time, even in a weak week, because the weak week is the one that most needs reading. And the decisions made in the meeting go into the following week’s report, with what has happened since. That way the report also records what leadership decided, and tracking sales quotas no longer depends on the memory of whoever was in the room.
A three-week plan to go live
- Week 1, definitions. Write the definition of each number in the five sections and validate it with finance and marketing. Build the reports in the CRM.
- Week 2, parallel pilot. Generate the report with AI and compare it with what the manager would build by hand. Adjust the prompt and definitions based on the differences.
- Week 3, routine. Schedule the export, set the send time, publish the versions by reader and open the Monday meeting with the decisions section.
If your leadership wants the weekly report, the Monday meeting and the sales, marketing and post-sale numbers to speak the same language, that is the work of the in-company applied AI training. I run the program on your company’s CRM and real reports, with the definitions of the numbers, prompts by reader and the review routine, and managers leave with the first automated report running.
Frequently asked questions
Can AI build the weekly report without anyone reviewing it?
It builds the text, and the review is still needed. Five minutes checking three totals against the CRM and writing the decisions section stay with the manager.
Which tool should I use to automate the report?
The CRM’s own scheduled reports handle the calculation, and any business AI assistant handles the text. For people who work with exported files, agents that read folders and generate documents shorten the process.
Why shouldn’t AI calculate the numbers?
Language models get sums and filters wrong without warning. The CRM or the spreadsheet calculates, and the AI receives finished numbers to write about.
What is the ideal length for a weekly sales report?
One page for the sales director and half a page for the executive team. Beyond that, people stop reading before the meeting.
Does the report replace the dashboard?
No. The dashboard shows the situation at any moment, and the report tells what changed during the week and what needs a decision. Both use the same definitions.
What should I do when the report’s numbers don’t match finance’s?
Compare the definitions, line by line. The difference is usually in the criterion for “won”, the cutoff date or the treatment of taxes, and the fix goes into the data dictionary.
Want to bring this to your team?
The topics in this article also become in-company training: designed from a diagnosis of your operation, with real cases from your team and a deliverable that stays with the company.
More writing on the same subject.

Meeting summaries with AI: from conversation to CRM
Logging the meeting is the task salespeople put off the most and the one managers need the most. How to turn the transcript into CRM fields with AI, with the prompt, the two-minute review and the care the recording requires.
12 min readRead more
B2B price increases: how to announce one and keep the client
An increase that arrives as a generic email from finance pushes the client toward a competitor or toward a request for an exception. How to segment the customer base before announcing, when to give notice, what to say in the conversation, which trade-offs to offer and what to measure in the following 90 days.
14 min readRead more
Sales proposals with AI: fast draft, human review
AI writes the first draft of a proposal in minutes when it receives the discovery summary, the company template and the pricing rules. The salesperson and the manager review scope, prices, timelines and promises before anything goes out. The workflow, the prompt and the review checklist.
12 min readRead more