40 AI prompts for your CRM and sales day

A blank chat box is surprisingly hard to use. You know the AI assistant could help, but at 8am with a full calendar, “what should I ask it?” is one more decision you don’t have time for.

This post gives you 40 prompts you can copy, grouped by the job you’re trying to get done: pipeline review, follow-up, forecasting, coaching, admin and reporting. They’re written for an AI assistant that can read your CRM data, so the answers come from your actual leads, deals, tasks and invoices. At the end you’ll find tips for writing your own.

Before you start

Three things make these prompts work better:

• Swap in your own details. Replace names, dollar amounts and dates with real ones. “The Patel deal” gets a better answer than “that deal.”
• Expect answers scoped to you. A well-designed assistant only sees what your role allows. If you’re a rep, “my pipeline” means your deals; a manager asking the same question may see the whole team’s.
• Treat actions as proposals. Prompts that create tasks, notes, messages or automations should come back as something you review and confirm. We explain why in why your AI assistant should ask before it acts.

Pipeline review (8 prompts)

Use these on Monday morning or before a pipeline meeting. They pair well with a regular pipeline management routine.

1. “Show my pipeline by stage.”
2. “Which deals are stalled?”
3. “Which leads are most likely to close?”
4. “Which deals have an expected close date that’s already passed?”
5. “Summarize the Patel kitchen remodel deal: highlights, risks and next steps.”
6. “Which deals over $10,000 don’t have an upcoming task?”
7. “What were the most common lost reasons this quarter?”
8. “List my open deals with no activity in the last two weeks.”

Tip: Stalled deals and missing next steps are the two problems a weekly review should catch. If prompt 6 returns a long list, that’s your to-do list for the morning.

Follow-up (7 prompts)

Follow-up is where most deals are won or lost quietly. These prompts find the gaps.

9. “Who needs follow-up?”
10. “Which new leads haven’t been contacted?”
11. “What tasks are overdue?”
12. “Which conversations are waiting on a reply from me?”
13. “Which quotes were sent but haven’t been accepted yet?”
14. “Draft a follow-up text to Maria Lopez about the quote we sent last week. Keep it under 300 characters.”
15. “Draft a reply to the latest email from Jordan at Brightline Dental, answering his question about timing.”

Tip: For drafts, say the channel and length. A text and an email to the same person should read very differently.

Forecasting and revenue (6 prompts)

These help owners and managers answer “how’s the month looking?” without building a spreadsheet. For the method behind them, see sales forecasting for small teams.

16. “What is my revenue forecast for the next three months?”
17. “What’s our weighted pipeline value right now?”
18. “Which deals are expected to close this month, and what’s their total value?”
19. “What did we earn this month?”
20. “Compare revenue this quarter vs last quarter.”
21. “What’s our win rate this quarter compared with last quarter?”

Tip: Ask for the breakdown, not just the total. “Show the deals behind that number” turns a forecast you have to trust into one you can check.

Coaching and team performance (6 prompts)

For managers running one-on-ones or a weekly team meeting.

22. “How is the team performing this month?”
23. “Compare leads, conversions and revenue by rep for the last 30 days.”
24. “Which rep has the most overdue tasks?”
25. “Which rep has the most new leads that haven’t been contacted?”
26. “Summarize Alex’s open deals with the biggest risks.”
27. “How many activities did each rep log last week?”

Tip: Use these to start conversations, not to grade people. A rep with many overdue tasks may simply have been assigned too many leads.

Admin (7 prompts)

These are the small jobs that eat the end of the day. Each one should come back as an action you confirm.

28. “Remind me to call Marcus tomorrow at 10.”
29. “Add a note to the Greenway Landscaping lead: wants spring cleanup quote, prefers texts after 5pm.”
30. “Create a lead for Priya Shah, [email protected], referred by the Johnsons, interested in a bathroom remodel.”
31. “Book an appointment with Dana Ruiz on Thursday at 2pm for a site visit.”
32. “Draft a quote for my newest lead.”
33. “Create an automation that texts new leads within a minute.”
34. “Create a task to call back every lead that replies.”

Tip: Read the confirmation card before you click. Check the date, the person and the wording, especially on anything a customer will see.

Reporting and billing (6 prompts)

Questions you’d otherwise answer by exporting data.

35. “Which lead sources convert best?”
36. “Show my lead funnel for the last 90 days.”
37. “How many new leads did we get this week?”
38. “Which invoices are overdue?”
39. “How much is outstanding right now?”
40. “How much did we collect last month, and how much is still overdue?”

Tip: Ask for a chart when you’re comparing things over time or across categories (“as a bar chart by month”). Ask for a list when you need to act on each item.

How to write your own prompts

The prompts above follow a few simple rules. Anthropic’s prompting guide puts the main one well: think of the AI as “a brilliant but new employee who lacks context on your norms and workflows.” The more precisely you explain what you want, the better the result. In practice:

Instead of	Try	Why it’s better
“How are sales?”	“What’s our revenue this month compared with last month?”	Names a metric and a time frame
“Any problems?”	“Which deals over $5,000 haven’t moved stage in 14 days?”	Defines what a problem is
“Write a follow-up.”	“Draft a short, friendly text to Dana about Thursday’s site visit.”	Gives person, channel, tone and purpose
“Show leads.”	“List new leads from Google Ads this week, newest first.”	Adds source, period and order
“Help with my pipeline.”	“Which three deals should I focus on today, and why?”	Asks for a limited, ranked answer with reasons

A few more habits:

• Give the why. “Draft a reply. She’s comparing us with two other contractors, so lead with our warranty” gets a sharper draft than “draft a reply.”
• Follow up instead of starting over. “Now only show the ones assigned to me” or “make it shorter” keeps the context.
• Ask for reasons. “Why is this lead ranked first?” tells you whether to trust the ranking.
• Save the prompts that work. Keep a short list for your Monday review and your end-of-day wrap-up.

Check the answers that matter

AI assistants can be wrong in a confident tone. NIST’s 2024 Generative AI Profile calls this “confabulation”: the production of confidently stated but erroneous or false content. An assistant that answers from your CRM data, rather than from general knowledge, is on firmer ground, but it can still misread a question. “This quarter” might be calendar or fiscal; “won deals” might or might not include ones closed today.

So before you act on a surprising number, click through to the records behind it. If an answer looks wrong, rephrase the question with a clearer time frame or definition.

Where Exa fits

Exa’s AI assistant answers natural-language questions about your CRM data with charts, summarizes leads and deals with highlights, risks and next steps, drafts replies and quotes, and answers billing questions like “Which invoices are overdue?”. It can create tasks, notes, leads, messages, appointments and automations, but only after you confirm an action card, and it respects your role and permissions. You can see the prompt menu on the AI assistant page. For a broader look at what AI can and can’t do in a CRM, read AI CRM: what AI can (and can’t) do for your sales team, and for more workflows built around these prompts, see 12 ways small sales teams use an AI assistant.

Final thoughts

You don’t need 40 prompts every day. Pick three or four that match your routine: a pipeline check in the morning, a follow-up sweep after lunch, a billing question on Fridays. Make them specific, check the numbers that drive decisions, and confirm every action before it runs. The goal isn’t to talk to a chatbot more; it’s to spend less time hunting through your CRM and more time talking to customers.