AI prompts for client reporting and updates
Template · 2 min read · Updated 2026-10-06 · By the Cofoundr team
A good reporting prompt names the account, period, comparison, target and audience, and asks for short plain-language output that shows the figures used. Always check the figures and remove any cause or promise you cannot support.
Vague prompts get vague reports. These prompts give the AI what it needs and ask it to show its working, so you can review quickly. Replace the brackets with your own details, and paste in real data rather than asking it to recall any.
What every prompt should include
- The account name and the period.
- The comparison period and the client's target.
- Who will read it and how they like to be addressed.
- The format you want and a length limit.
- An instruction to list the figures used and to flag anything uncertain.
Prompts for the monthly report
- Write a short summary of [month] for [client]. Use the data below. Compare with [previous month] and the target cost per result of [target]. Use plain language for a non-technical owner. Finish with three next steps. List the figures you used.
- Explain in three sentences why cost per result moved between [period A] and [period B], using only the data below. Say what is a fact and what is a guess.
- Turn these results into five bullets a busy owner will read in one minute.
Prompts for recommendations
- Suggest three actions for next month based on the data below, and say what evidence supports each one.
- Which two ad sets look weakest and what should I check first? Use only the data below.
Prompts for difficult updates
- Draft a short, honest message to [client] explaining that cost per result rose above target in the second half of the month. Say what we think caused it, what we have done and when we will update them. Keep it calm and specific.
- Rewrite this update to be shorter and less technical, without hiding the problem.
Prompts to prepare for a call
- List the five questions [client] is most likely to ask about this report, with a short answer for each based on the data.
Checking the output
- Check every figure against the dashboard.
- Remove any cause stated as fact when you cannot prove it.
- Delete promises about future results.
- Read it as the client would.
Where Cofoundr helps
Hadya inside Cofoundr already has the selected account, date range and health data, so these prompts work without copying numbers. You can also connect Claude or ChatGPT through MCP. More analysis prompts are in AI prompts for Meta Ads analysis.
Frequently asked questions
Should I paste raw numbers into the prompt?
Yes, real numbers from your source of truth. Do not ask the AI to remember or estimate figures.
How long should a client update be?
Short enough to read in a minute or two. Ask the AI for a length limit.
Can I reuse these prompts for other platforms?
The structure works anywhere. Replace the platform details and use data from that platform.
