A win-loss analysis agent reads the calls, emails, and CRM history on every closed deal and explains why it was won or lost, with evidence. The five templates below cover lost deal post-mortems, won deal reviews, competitor tracking, product gap digests, and a monthly pattern report. Each one is ready to paste, with its trigger, inputs, and what good output looks like.
If your win-loss program is a dropdown field and a quarterly slide, you are not alone. Most teams only interview a few buyers a quarter, and the rest of the story lives in a loss reason a rep picked on the way out. This is the second set of agent templates in our series. The first, five sales AI agent templates, covers follow-up, MEDDIC gaps, deal risk, hygiene, and forecast prep.
Why do most win-loss programs miss the real reasons deals close?
Most win-loss programs miss the real reasons because they rely on two thin sources: the loss reason a rep logs from memory, and a small number of buyer interviews. Both are useful, and neither covers every deal.
The pattern in the data is consistent. In Airspeed’s published study of 133,068 closed deals, “lost to a competitor” made up only 2.8% of 253,008 labeled loss reasons, while a product or feature gap was the most common top reason at 42 of 102 companies. The same study found deals with five or more buyer-side contacts won 46.4% of the time, against 24.3% for deals with one. Those are things a dropdown rarely captures and a call record shows plainly.
Buyers are also spending less time with your team. Gartner found in 2025 that 61% of B2B buyers prefer a rep-free buying experience. When conversations are scarce, the ones you do have are the best evidence you will get, and an agent can read all of them.
How do you write a closed-lost post-mortem with an AI agent?
What it does. Writes a short post-mortem on each lost deal from the full record, then checks whether the evidence supports the loss reason the rep logged.
When it runs. When a deal moves to Closed Lost, limited to deals that had at least one external call.
Deal {{deal_id}} just moved to Closed Lost. Read every call, email,
and CRM change on this deal from creation to close.
1. Timeline: list the five moments that most shaped the outcome,
each with a date and a one-line description.
2. Likely reasons: name up to three reasons the deal was lost. For
each, quote the buyer or cite a date or CRM field as evidence.
3. Logged reason check: compare the loss reason in the CRM with your
findings. Say Agrees, Partly agrees, or Disagrees, and why.
4. Earliest warning: the first point where the record showed the
deal was at risk, and what could have been done then.
Use only evidence from the record. If a reason is a guess, label it
[UNCONFIRMED]. Keep the whole post-mortem under 250 words.Inputs. Call transcripts and attendee lists, email activity, stage and close date history, and the logged loss reason.
Good output. A post-mortem a manager can read in a minute, where every reason carries a quote or a date and the logged reason is either confirmed or challenged.
How do you find out why you win deals?
What it does. Runs the same review on won deals, so you learn what to repeat rather than only what to avoid.
When it runs. When a deal moves to Closed Won.
Deal {{deal_id}} just moved to Closed Won. Review the full record.
1. Turning points: the two or three moments that moved the deal
forward, with a date and a quote for each.
2. People: who on the buyer side joined calls, when they first
appeared, and which of them drove the decision.
3. What we did well: up to three rep actions that made a difference,
with evidence.
4. What nearly went wrong: one risk that showed up and how it was
handled.
End with one sentence a rep could reuse on a similar deal. Keep it
under 200 words and cite evidence for every point.Inputs. The same record as the post-mortem, plus attendee lists across the whole deal so the agent can see how many buyer-side people were involved and when.
Good output. A review that names specific moves, such as bringing in the economic buyer on the third call, rather than generic praise.
How do you track competitor mentions across sales calls?
What it does. Logs every competitor a buyer names, in what context, and what they compared, so your competitive picture builds up deal by deal instead of being reconstructed at quarter end.
When it runs. After each external call tagged Discovery, Demo, or Negotiation.
Review call {{call_id}} for any mention of another vendor, product,
or in-house build the buyer is considering or already uses.
For each mention, return:
- Vendor name, exactly as said.
- Who raised it (buyer or rep) and the timestamp.
- Context: Evaluating, Currently using, Previously used, or Mentioned
in passing.
- What the buyer compared or praised, in one sentence with a quote.
- Whether the rep responded, and how, in one sentence.
If no other vendor was mentioned, return "No competitor mentions."
Do not infer a competitor that was not named.Inputs. The call transcript and the deal it belongs to.
Good output. A clean log per call that makes the quarterly question “who do we actually lose to” a lookup instead of a debate. For a ready-made version that turns these mentions into talking points, see the Competitive Positioner.
How do you send product gap feedback from calls to your product team?
What it does. Collects every feature request and product gap buyers raise, ties each one to deal stage and outcome, and sends a weekly digest your product team will read.
When it runs. Weekly on Monday morning, with previous-run context turned on.
Review all external calls from the last 7 days. Find every point
where a buyer asked for a capability we do not have, said something
did not work for them, or named a gap as a reason to hesitate.
Group the findings by theme. For each theme, return:
- A one-line description of the gap.
- How many deals raised it this week, and the stage of each deal.
- One representative quote, with the call date.
- Whether any deal that raised it closed this week, won or lost.
Compare with your previous run and mark each theme New, Growing,
Stable, or Fading. Return at most 8 themes, most deals first.Inputs. A week of call transcripts, deal stages, and close outcomes, plus the previous digest for comparison.
Good output. A one-screen digest a product manager can bring to planning, where each theme is backed by deal counts and a real quote rather than the loudest anecdote.
How do you build a monthly win-loss report with an agent?
What it does. Rolls up every post-mortem, review, competitor log, and gap digest from the month into one report for revenue leadership.
When it runs. The first weekday of each month, covering deals closed the previous month, with previous-run context turned on.
Prepare a win-loss report for deals closed last month.
1. Summary: deals won and lost, and the three findings that matter
most, in five sentences or fewer.
2. Loss reasons: the top reasons from the call evidence, compared
with the reasons logged in the CRM. Note where they disagree.
3. Win patterns: what the won deals had in common, such as number
of buyer-side contacts or buyer-owned next steps.
4. Competitors: who appeared in lost deals and who in won deals.
5. Change: what is different from your previous report.
6. Actions: three recommendations, each tied to a finding above.
Cite deals by name as evidence. Use a table wherever you compare.Inputs. All deals closed in the period with their post-mortems or reviews, the competitor log, the product gap digest, and last month’s report.
Good output. A report a CRO can read before the monthly business review, where each recommendation traces back to deals they can open. For a broader view of what this agent does inside Airspeed, see the Win/Loss Analyst.
How do you roll out win-loss agents without drowning in reports?
Start with the closed-lost post-mortem, because it runs on a clear event, the output is short, and managers can judge it against what they already know about the deal. Send early runs to an in-app history rather than Slack, read the first ten, and tighten the instructions wherever the agent guessed.
Then add the competitor tracker and product gap digest, which build the raw material the monthly report needs. Turn on the monthly report last, once a full month of post-mortems exists. Keep each output capped and evidence-backed, and agree in advance who acts on each one: managers on post-mortems, product on gap digests, and leadership on the monthly report. An agent nobody owns is the fastest way to a report nobody reads.
Where Airspeed fits
Every template above runs as written in Airspeed Agents. Airspeed already captures your calls, transcripts, attendees, and CRM history, so a post-mortem starts from the full deal record instead of a blank page. You choose the trigger, whether that is a deal stage change, a finished call, or a schedule, paste the instructions, and send the output to Slack, email, or in-app history. With previous-run context on, the weekly and monthly agents report what changed since last time. Airspeed does the reading on every closed deal, so your team gets the real reasons without anyone opening a single transcript.
If you want to run these templates on your own closed deals, book a demo.
Frequently asked questions
How can AI automate win-loss analysis?
An AI agent can read every call, email, and CRM change on a deal the moment it closes, then write a short post-mortem that names the likely reasons it was won or lost, with a quote or date behind each one. Run on a schedule, the same agent rolls those post-mortems into a monthly pattern report, so win-loss analysis covers every closed deal instead of the handful a team has time to interview.
Why are CRM loss reasons often inaccurate?
Reps pick a loss reason from a dropdown at the moment they close the deal out, usually from memory and often in a hurry. Broad options such as price, timing, or other absorb most losses, and the choice is rarely checked against what the buyer actually said. An agent that compares the logged reason with the call record catches the gap.
What should a win-loss analysis agent look at?
At minimum, the call transcripts and attendee lists for the whole deal, the CRM stage and close date history, and the loss or win reason the rep logged. Email activity and the number of buyer-side contacts add a lot, because silence and single-threading are two of the most common patterns behind a slipped or lost deal.
How often should you run win-loss analysis?
Run the per-deal post-mortem every time a deal closes, while the record is complete and the rep still remembers it. Run the pattern report monthly or quarterly, because a single loss tells you little and the value is in the trend across many deals. Competitor and product gap tracking work best after each call, so nothing is lost by the time the deal ends.
Does AI win-loss analysis replace buyer interviews?
No. Agents cover every deal from the evidence you already hold, which interviews cannot do at scale. Interviews still catch what the buyer never said on a call, such as internal politics or a vendor they never named. The strongest programs use the agent output to choose which buyers to interview and what to ask them.
