15 AI prompts for sales follow-ups, organized by how long the silence has lasted: three days, two weeks, gone cold. Three mine the real inbox for owed replies.
Key takeaways
- This is not a cold-outreach page. Every prompt here works threads that already exist: the demo that went quiet, the proposal nobody answered, the intro that stalled. The pipeline hiding in sent mail, not the one in a prospecting tool.
- The right follow-up depends on the age of the silence, so the prompts are organized by it: the three-day nudge, the two-week reframe, and the gone-cold reopening, which are three different messages, not one template resent harder.
- The most valuable prompt is the one that finds the follow-ups you owe: the deals that stall because nobody remembered are found in the sent folder, and mining it by hand is why it never happens.
- Prompts 1 to 12 are copy-paste in any assistant. The last three run on a connected inbox: finding every thread where the last word was yours and a reply is owed, drafting the nudges, and keeping the pipeline honest.
Table of contents
The short version: these fifteen prompts work the pipeline that already lives in your inbox: the demo that went quiet, the proposal that got a "looks great, give us a week" three weeks ago, the warm intro that stalled. No cold outreach, no sequences; that ground belongs to the outbound tools, and this page deliberately stays off it. The organizing idea is that silence has ages, and a three-day-old silence, a two-week-old one, and a dead thread need three different messages. The first twelve prompts are copy-paste in ChatGPT, Claude, or Gemini. The last three run on a connected inbox and find the follow-ups you owe before drafting them.
Where this page sits in the library: the ChatGPT prompts collection has a broad sales section including pitches and comparisons; this page goes deep on exactly one motion, the follow-up, because it is the one that moves pipeline with mail you have already sent.
The three-day silence (prompts 1-4)
Fresh silence is usually logistics, not rejection. These nudge without weight.
1. The light nudge. The follow-up that is easy to answer beats the one that is thorough.
2. The value add-on. Following up with something beats following up for something.
3. The post-demo specific. Generic demo follow-ups die; the one that names their moment does not.
4. The multi-thread untangler. When the deal spans three people and nobody has replied.
The two-week silence (prompts 5-8)
Old enough that repeating yourself reads as nagging. The message has to change shape.
5. The reframe. Same offer, new angle, because the old angle already failed once.
6. The honest direct. Sometimes the strongest move is naming the silence.
7. The stakeholder shift. When the silence means your contact lost the internal argument.
8. The deadline that is real. Urgency works once, and only when it is true.
Gone cold: the reopening (prompts 9-12)
Months of silence. The thread is dead; the relationship might not be.
9. The no-pretense revival. Reopening without pretending the gap did not happen.
10. The trigger-event opener. The best reason to reappear is something that happened to them.
11. The graceful close-out. The breakup email that sometimes revives the deal, and always cleans the pipeline.
12. The lost-deal learning note. When they chose someone else and told you.
Drag AI
The inbox your team and your AI work in together
Shared inbox, live chat, and AI in Gmail, with an MCP server your AI tools can drive.
The prompts that mine the real inbox (prompts 13-15)
Every prompt above assumes you know which threads need following up. The connected prompts remove that assumption: the assistant reads the actual sent mail and finds the silences for you. They run against Drag's MCP server; the connection guide covers setup in about ten minutes.
13. The owed-replies sweep. The prompt that finds the pipeline leak.
What happens: search_threads walks the real conversations, and the ranked list arrives with the asks quoted. The deals that stall because nobody remembered stop being found by accident.
The result shape, from a connected inbox:
14. The silence-aged drafts. Prompt 13's list, answered in the right register per thread.
What happens: the assistant reads each thread's dates and content, applies the same age logic this page teaches, and the drafts wait in their threads. The judgment stays reviewable; the remembering stops being a job.
15. The Friday pipeline sweep. The weekly habit that keeps the sent folder honest.
What happens: search_threads and the conversation history produce both lists with dates from the real messages. The sweep takes thirty seconds, which is the difference between a habit and an intention.
Where the prompts stop and the system starts
Follow-ups fail as a memory problem before they fail as a writing problem, and prompts only fix the writing half unless the assistant can see the inbox. For disclosure: Drag is our shared inbox for Gmail, where teams run sales pipelines as boards on the mail itself, from $12 a seat with Drag AI from $18, and its MCP server (Pro plan) is what makes prompts 13 to 15 run. Everything else on this page needs only a clipboard and any assistant.
FAQ
How do I write a follow-up email after no response?
Match the message to the age of the silence. At a few days, ask one easy question with no recap. At two weeks, change the angle rather than repeating the pitch, or name the silence honestly. Months later, acknowledge the gap in a clause and lead with what changed. The constant across all three: shorter than the message that went unanswered.
How many times should I follow up on a sales email?
Three or four touches across increasing gaps covers most of the value, provided each message changes shape rather than resending the same ask. The more useful discipline than a count is an ending: a graceful close-out that leaves the door open reads as confidence, cleans your pipeline, and revives more dead deals than an eighth nudge does.
Can AI find the emails I forgot to follow up on?
Copy-paste AI cannot, because it only sees what you paste, and the forgotten thread is by definition the one you did not paste. Connected to the inbox through an MCP server, an assistant can search the real sent mail for threads where the last word was yours and an ask went unanswered, then rank and draft from what it finds.
The rest of the prompt library
This page is one of nine in the Drag prompt library. The ChatGPT prompts cover the wider sales writing, the triage prompts include the follow-up tracker this page's sweep grew from, and the reporting prompts carry the commitment-risk scan for the promises side of the same problem.
Co-founder
Building Drag for nearly ten years: shared inboxes, boards, and now the AI and agent layer, all on Gmail, plus HeyHelp for the personal inbox. Writes the honest versions of the comparisons.
