Feed deal outcomes back into turgo so targeting and messaging improve on evidence, not opinion. Tag why deals were won or lost, compare against predicted ICP tier, and let the pattern reshape your rubric and sequences.
Capture outcomes with reasons
Log won and lost deals with a structured reason — budget, timing, competitor, fit, no decision. Free-text alone doesn't aggregate; a consistent reason taxonomy is what makes win-loss analyzable over time.
Compare to predicted tier
Cross-reference outcomes with the ICP tier the rubric predicted. If A-tier accounts convert far better than B/C, the rubric is working; if wins are scattered across tiers, the rubric needs recalibration (see Tune your ICP after 30 days).
Find the message patterns
Look at which openings, signals, and sequences preceded wins versus losses. Patterns here feed directly into sequence design and signal selection — you're letting closed deals tell you what outreach to double down on.
Close the loop into the agent
Turn findings into concrete changes: adjust rubric weights, retire underperforming signals, update the agent's context and guardrails. Win-loss analysis is only valuable if it changes what the agent does next quarter.
