GTMcraft Plays · The Forecast Voice Play

Your forecast reads what reps type, not what buyers said; the agent flags every risk, you still own the commit.

Play visual

The 90-second read

Your forecast call runs on rep optimism and a CRM that only knows what someone typed. The agent can read every open deal and surface the risk the reps talk around. But an agent that sets stages and probabilities just automates the guessing faster. The move is governance, not another prediction. Give the agent a voice, so it flags. Keep the vote, so a human commits the number.

• The agent reads every deal and flags risk; the human still owns the call.
• A one-page risk brief replaces the ninety-minute story-telling roll call.
• Four flags matter: no next step, gone quiet, past close date, single-threaded.
• A model that overrules the rep is a mistake dressed as accuracy.
• You grade the calls, not the deals, so accuracy compounds by team.

Read time: about 6 minutes. First step runnable Monday morning.

The #1 move this week

Write the one rule that the agent flags and a human commits, then say it out loud in Friday's forecast call.

Most forecast tools fail because nobody decided who wins when the model and the rep disagree. Decide it first. The agent gets a voice, never a vote. A disagreement is a mandatory review, not an override.

Start here: draft one sentence of decision rights and paste it at the top of your forecast doc. You need your commit definition and the name of the person who makes the final call.

Ignore this week: buying a forecasting agent. Set the rule before you point any model at your pipeline.