Library › Customer Support & Success AI › Sales and customer success

Map who decides on a deal, with a short exact quote for each

  • For Sales and customer success
  • You practice Keeping the AI to your sources
  • Time about 5 minutes
  • Level Advanced

About this exercise

Build a plain-text table with the header row exactly "Name | Role | Evidence" and one row per person in the order given. Role is one of Decider, Influencer, Blocker, Process or Unknown. Evidence is an exact quote of 8 words or fewer copied from the thread in quotation marks, or "none" if the person never wrote anything. Do not infer roles from job titles. Plain text.

Common mistakes when you ask AI for this

  • Inferring from titles. The AI makes Sunil the Decider because he runs operations. Try: Say: "Do not infer roles from job titles; use only the thread."
  • Paraphrasing evidence. The AI summarizes instead of quoting. Try: Say: "Evidence is an exact quote of 8 words or fewer."
  • Guessing for Rob. The AI assigns legal a role. Try: Say: "If a person wrote nothing, the role is Unknown and evidence is none."
  • Getting markdown back. AI adds bold and headings by default, which show up as stray symbols where this is pasted. Try: Say "plain text, no bold, no headings, no asterisks".
  • Trusting the AI to count words. The AI runs long and cannot count its own words. Try: Cap each part ("each line under 12 words") and add "under N words in total".
  • Leaving a requirement unstated. The AI follows its own defaults unless the prompt spells this out. Try: State it plainly in your prompt: the header row is exactly 'Name | Role | Evidence' with five rows in the order Mara, Leo, Sunil, Dee, Rob.

What a strong prompt includes

  • Gives the header
  • Lists the five roles
  • Requires exact quotes
  • Bans title inference
  • Asks for plain text
  • Sets a word cap

Practice this now ↑

Related exercises