Win-Loss Analysis

Sellers' loss reasons are mostly guesses, and "price" is the most common guess. You find the real reasons by combining the CRM record, what the buyer wrote and said, and short interviews with the buyers themselves, then you count. The standard: every finding is backed by a number of deals and by buyers' own words, losses to "no decision" are counted as losses, and each recommendation names who should do what by when.

When to use

• "Why are we losing deals?"
• "Run our quarterly win-loss review."
• "We keep losing to Competitor X; find out why."
• "Win rate dropped from 30 to 18 percent; what changed?"
• After launching a new offer or price, to see how buyers responded.

What you need

• Closed won, closed lost and closed with no decision deals for the period (at least a quarter; ideally 20 or more of each outcome), with amount, segment, source, competitor, stage reached, cycle length, loss reason and owner. From the connected CRM or an export.
• Call notes, emails and any proposals for those deals (connected mailbox, documents, session_search).
• The owner's permission to approach buyers for interviews, and a list of who must not be contacted (sensitive relationships, legal disputes).
• The qualification scorecards, if the team uses them.

Method

1. Define the sample. All closed deals in the period above a minimum size, plus any notable smaller ones. Include no-decision outcomes. Note what proportion have usable notes.
2. Quantitative pass with execute_code:
   • Win rate overall and by segment, deal size band, lead source, owner, competitor, product and region. Show counts beside percentages; flag any group under 10 deals as too small to conclude from.
   • The stage at which lost deals were lost. Losses after the proposal stage cost the most effort.
   • Cycle length of wins versus losses. Long cycles often signal no compelling event.
   • Qualification completeness at loss versus win: the share of deals where the economic buyer was never met, the champion was untested, or the paper process was unknown.
3. Code the reasons from the record. For each deal, read the notes and emails and assign:
   • one primary reason and up to two contributing reasons from the fixed codebook in references/win-loss-kit.md (no decision, competitor and on what, price or value, product gap and which, timing, lost champion, poor fit, sales process, trust or risk, implementation concerns);
   • the quote or line that supports each code. If evidence is thin, code "unclear" rather than guessing.
4. Recruit interviews. Aim for 6 to 12 buyers per period, a mix of wins and losses, within about 90 days of the decision while memory is fresh.
   • The request comes from someone other than the deal owner where possible.
   • It is short, honest about the purpose, offers 20 minutes, and promises no sales pitch.
   • Draft each request for the owner's approval; send nothing without it.
   • Respect any refusal, and do not follow up more than once.
5. Interview well (guide in the reference). Ask about their process and alternatives before asking about the seller:
   • what triggered the search, and who was involved;
   • what options they considered, and how they compared them;
   • what tipped the decision, and what nearly changed it;
   • what they thought of the pricing and the sales experience. Listen for the moment of decision. Do not defend, correct or pitch. With consent, record or take notes, and store notes with the deal.
6. Compare seller reasons with buyer reasons. Make a table of deals where both exist. Where they differ, the buyer's account wins. Report how often the CRM reason was wrong; this calibrates future data.
7. Find patterns.
   • Rank themes by the revenue involved, not just the count.
   • A pattern needs at least 3 deals with supporting quotes.
   • Separate what the team controls (discovery quality, follow-up speed, proposal clarity, reference availability) from what it does not (a product gap, a competitor's pricing).
8. Write recommendations that are specific and testable: what to change, who owns it, by when, and how you will know it worked (the metric and the next review).
   • Example: "Require an economic buyer meeting before proposals over the deal-size threshold. Owner: sales manager, from next month. Measure: win rate on those deals next quarter."
   • Route product gaps to the product team with evidence and customer quotes.
9. Close the loop.
   • Share findings with the team via brief_team once the owner has approved them.
   • Update the ICP, objection library and qualification criteria where the evidence supports it.
   • Record the baseline win rate in memory for the next review.

How much evidence is enough

Claim	Minimum evidence
A segment's win rate	10 or more closed deals in the segment; otherwise "indicative"
A loss pattern	3 or more deals with the same coded reason and supporting quotes
"The CRM reasons are wrong"	Seller and buyer reasons compared on at least 5 deals
A recommendation	A pattern plus a change the team controls and a way to measure it

Example finding: "In 7 of 11 losses above the size threshold, the economic buyer never met us (CRM and notes); in wins it was 2 of 9. Three interviewed buyers said the decision was made in a meeting we were not in. Recommendation: no proposals above the threshold before an economic buyer meeting."

Output

• A report (3 to 5 pages) using the kit's structure: headline findings, win rate tables, reason analysis with quotes, seller versus buyer comparison, patterns ranked by revenue, recommendations with owners and dates.
• A show_card with the win rate trend, the top 3 reasons for losses and the top 3 for wins.
• The coded dataset (CSV) so the next review can compare.
• Interview request drafts and notes (only with approval and consent).

Checks before you finish

• No-decision outcomes are included and counted.
• Every theme cites the number of deals and at least one buyer quote.
• Groups under 10 deals are flagged as indicative only.
• Buyer interview requests were approved by the owner; no-contact names were respected.
• Recommendations each have an owner, a date and a success measure.

Pitfalls

• Taking "price" at face value. It usually means the value was not proven, or the buyer did not want to give the real reason. Probe and code what you find.
• Only studying losses. Wins show what to repeat. Study both.
• The deal owner interviewing their own buyer. Buyers are polite to the person they rejected. Use someone else where possible.
• Turning interviews into a second sales attempt. It destroys trust and the data.
• Anecdotes as findings. One loud loss is a story; three similar ones are a pattern.
• A report with no owners. Findings without actions change nothing.

See also: pipeline-review-and-forecast, discovery-qualification, ideal-customer-profile, objection-handling.