ChatGPT for sales teams: a one-page AI playbook

Why "everyone just uses ChatGPT" doesn't work

Most reps already use AI, usually their own account with their own prompts: Microsoft and LinkedIn's 2024 survey found 78% of AI users bring their own tools to work, and only 39% had any AI training from their company. On a sales team, that means eight versions of the same first email, and a manager who can't tell why one rep's outreach works and another's doesn't.

The upside of sharing is concrete. In the NBER study of a support team using an AI assistant, the least experienced staff gained most, because the assistant carried what the best performers did. A shared playbook does the same for a sales team: your top rep's approach, in every rep's AI.

The one-page playbook

A worked example: an eight-rep team

Illustration: a made-up business. An eight-rep team sells scheduling software to dental clinics. The sales manager asks the top rep, Nadia, to talk through how she writes a first email and qualifies a deal, for 20 minutes each. Her answers become the instructions for two of the four jobs.

For the deal review, a rep pastes notes from a call with a three-clinic group. The shared deal review comes back with a score, the evidence for each part, and the gaps: "Economic buyer: not identified (the practice manager said the owner signs). Next action: ask the practice manager to bring the owner to the demo." The manager reviews eight deals in the Monday meeting in 20 minutes, all in the same format, instead of eight different stories.

The rules held up in the first month: one rep wanted to paste a clinic's patient-volume export into a personal account; the playbook said company account only, and the team was on a company plan by then.

Roll it out in two weeks

1. Move the team to company accounts (ChatGPT Business or Claude Team), which don't train on your data by default.
2. Record your best rep explaining each of the four jobs, 20 minutes each.
3. Turn each recording into instructions with real examples, banned words and the fields your CRM needs.
4. Put the four where every rep's AI reads them, a shared project or shared skills, and retire the old prompts in people's notes.
5. Review deals in the shared format at the next pipeline meeting, and fix the instructions, not the reps, when the output is off.

Next step

Instructions a rep has to open or paste get skipped on a busy day. The next rung is turning each of the four into a skill, which the AI picks by itself when a rep asks for "a first email to Brightsmile Dental". How to make your first Claude skill from a doc you already have shows how, and getting your team to use AI the same way covers the rollout beyond sales.

The AI skill below is the deal review from the playbook: paste a rep's notes, and your AI scores the deal with the evidence for each part, lists the gaps with one next action each, and suggests a forecast category.

Questions

How can a sales team use ChatGPT?

For four jobs that every rep repeats: researching an account before a call, writing a first email, turning call notes into CRM notes and next steps, and reviewing whether a deal is real. The gain comes when the whole team uses the same instructions for each, written from what the best rep does.

Can sales reps put customer data into ChatGPT?

Only into a company account whose plan doesn't train on your data by default, such as ChatGPT Business or Claude Team, and only what your contracts allow. Personal accounts get public information only.

How do I stop reps' AI emails sounding the same as everyone else's?

Give the AI your best rep's real emails as examples, a list of words you never use, and the one detail each email must reference from the account research. Generic emails come from generic instructions.