Win/Loss Analyzer

Gut feel about why you win and lose is almost always wrong. Common self-diagnosis is "we lose on price." Common actual answer is "we lose when discovery doesn't surface a quantified pain point" or "we lose when the champion never gets us in front of the budget owner." The fix for the wrong diagnosis is the wrong fix — better pitch training when the real need was better qualification. This skill exists to replace the wrong diagnosis with the right one.

Analysis runs across seven dimensions. Win patterns — what won deals share, by source, buyer profile, and process markers. Loss patterns — which stages deals die in, which objections recur, which competitors beat you consistently. Speed analysis — where deals stall and the point-of-no-return delay threshold. Competitive intel — when you win against each competitor versus when you lose. Process gaps — which sales steps, when skipped, correlate with losses. Pricing and value — the deal-size sweet spot and where you consistently lose. Champion and stakeholder analysis — how the number of contacts and the seniority of the champion shifts win rate.

Confidence gets called out honestly. Five to ten deals gets "signals, not certainties." Eleven to twenty gets "emerging patterns." Twenty-plus gets "reliable patterns." Below five deals, the skill flags that this is guesswork and asks for more data rather than producing confident-sounding noise. The minimum viable dataset is five; the ideal is fifteen-plus; and the skill is explicit about which.

Two design choices that matter. Before presenting findings, the skill asks for the user's pre-existing theory and tests it against the data explicitly — the most valuable insight is usually "you think X, but the data says Y," and confirming bias is worse than disagreeing with it. And stalled deals count as losses: if a deal has been stuck for twice the average cycle, it's lost, and pretending otherwise inflates the pipeline forecast and prevents the post-mortem that would catch the pattern. Every insight ends with "therefore, change X" — analysis without an action is trivia.