Win-Loss Analyst
Tests whether you really lost on price
- 8hard rules
- Things it must never do, enforced in every run
- 5handover rules
- Conditions where it stops and gives the work to a person
- 12output fields
- A structured result, not a paragraph you have to re-read
- ≤7steps per run
- Bounded work: at most 5 tool calls, enforced by the runtime
Read-only by design. This agent has no path to sending, spending, publishing or committing anything. Where that limit is the product, removing it would remove the reason to trust the output.
Overview
Every CRM has a closed-lost reason field and the data in it is close to worthless. It is chosen by the person who lost the deal, at the moment they lose it, from a list written by someone else — and "price" is always on that list and never wrong enough to argue with. So price dominates every closed-lost report in every company, and the conclusion drawn from it, discount more, is expensive and usually mistaken.
The evidence that contradicts it is in the same records. If price were the reason, losses would concentrate where the price gap is largest. Usually they concentrate somewhere else: in deals where one specific competitor was present, in deals that never reached a second meeting, in deals where the buyer needed something the product does not do, in deals where the champion was too junior to sign. All of that is visible in stage history, call notes, participants and stage durations.
And nobody analyses the wins. A loss analysis without them cannot say what is different about the deals you close, which is the more useful half.
This agent tests the stated reason against the evidence, reports what actually separates wins from losses, and never assesses an individual rep. A per-person win-rate table gets managed rather than acted on, and it corrupts the reason field further.
Benefits
You stop discounting for the wrong reason
Which is the most expensive mistake this data causes.
Product and positioning get real input
From evidence, not from a field chosen under pressure.
Qualification improves
Because you know which deals were never winnable.
What it does
Tests the stated reason against the evidence
If price were the reason, losses would cluster where the gap is largest. Usually they cluster somewhere else entirely.
Analyses the wins too
A loss report alone cannot tell you what is different about the deals you close.
Finds where deals actually die
The stage they stop at, and how long they sat there. Far more honest than a dropdown.
Reads competitor presence
Losses concentrated against one competitor is a positioning problem, not a pricing one.
Checks buyer seniority
A champion who could never sign is a qualification finding that gets recorded as price.
Never ranks reps
A per-person win-rate table gets managed rather than acted on, and it corrupts the reason field further.
What this changes
What this changes about your week
- It takes over
- Testing why deals were actually lost, against what was typed in the CRM.
- Done today by
- A closed-lost report that everyone reads and nobody trusts.
- How often
- Quarterly, from the same unreliable field.
- Typical manual effort
- Stage history, competitors, seniority and requirements all read across won and lost deals together.
Why Sales in particular. 'Price' is chosen by the person who lost, at the moment they lost, from a list someone else wrote — it is always available and never wrong enough to argue with, so it swamps everything and the fix it implies is expensive.
Runs unattended
Started by you or by an event, and it finishes on its own. Nothing waits for someone to be at a desk.
The same standard every time
The two-hundredth item is held to the bar the first one was. Consistency is the part people cannot sustain.
It cannot act on its own
Win-Loss Analyst has no path to sending, spending or committing. That limit is why its output is safe to act on.
How it runs
This agent runs server-side through the PROMIVO runtime. Each run is logged step by step and every tool call is permission-checked before it executes.
- 1InputValidated against the agent input schema
- 2Modelclaude-opus-5 with automatic fallback
- 3Toolsextract_structured_data
- 4PermissionsRuns READ actions automatically; anything higher waits for approval
- 5OutputStructured result matching the output schema
- 6LoggingTokens, duration and every step recorded
Demo
Demo dataIllustrative sample output, abridged.
Input
{
"deals": [],
"period": "H1 2026",
"currency": "USD",
"minimumSample": 8
}Output
{
"escalate": true,
"priceTest": {
"finding": "Median discount requested was 9% in lost deals and 11% in won ones — you discounted more on the deals you won. And 38 of the 61 price losses never reached a second meeting, where price had not yet been discussed at all. Price is what gets typed; it is not what happened.",
"statedAsPrice": 61,
"evidenceSupports": false
},
"disclaimer": "An analysis of your own CRM records. No individual salesperson is named, ranked or scored; no customer or lost buyer has been contacted; no record has been changed; no discount or price change is recommended; and patterns below the stated minimum sample are reported as too thin rather than ranked.",
"statedReasons": [
{
"deals": 61,
"share": 54,
"reason": "Price"
},
{
"deals": 22,
"share": 19,
"reason": "Timing"
},
{
"deals": 18,
"share": 16,
"reason": "Went with competitor"
}
],
"whereDealsDie": [
{
"stage": "Discovery",
"lostDeals": 44,
"medianDaysThere": 31
},
{
"stage": "Proposal",
"lostDeals": 19,
"medianDaysThere": 48
}
],
"capabilityGaps": [
"A specific reporting requirement was raised in 14 lost deals and 2 won ones. Reported as a repeated requirement rather than as a confirmed gap — some of those may be a communication problem rather than a missing capability."
],
"escalationReason": "Stated reasons and evidence disagree materially — 54% of losses are recorded as price and the discount data contradicts it — and 22 of 31 competitive losses involve a single competitor.",
"seniorityPattern": "Deals worked entirely with a manager-level champion lost 79% of the time. Almost all were recorded as price.",
"tooThinToConclude": [
"Losses in the public sector segment: 5 deals. Not enough to say anything."
],
"whatSeparatesWins": [
"Won deals reached a second meeting in a median of 9 days. Lost deals that got there took 24, and most never did.",
"Won deals involved someone at director level or above in 81% of cases. Lost deals, 34%."
],
"competitorPatterns": [
"Of 31 competitive losses, 22 involved the same competitor, and 19 of those died at proposal rather than on price. That is a positioning problem at proposal stage, and no amount of discount addresses it."
],
"recordsCannotAnswer": [
"Whether the buyers who cited price would have paid more for something else. Only a conversation with them answers that, and it is worth having with the 22 lost to one competitor."
]
}Requirements
Integrations
No integrations required.
Before you start
- Closed won and closed lost deals with value, stage history and dates
- Stated close reasons
- Competitors present, where recorded
- Contact roles and seniority, and notes or requirements captured
- Requires the All-Access plan or higher
Use cases
Quarterly win-loss review
What actually separates the two.
Before a pricing change
Whether price is really the constraint.
Competitive strategy
Which competitor you lose to, and at which stage.
Product roadmap input
Requirements raised in deals you lost.
Pricing
All-Access
Billed monthly through your PROMIVO subscription. Cancel at any time.
Runs consume your plan allowance for agent executions and tokens. See plan limits
FAQ
Does it rank salespeople?
No, deliberately. A per-person win-rate table gets managed rather than acted on, and the first thing to change is what people type into the reason field. Findings are about deals and patterns.
Why not just use our closed-lost reasons?
Because they are chosen by the person who lost, at the moment they lost, from a list someone else wrote. Price is always available and never wrong enough to argue with, so it swamps everything.
Does it interview customers?
No. It contacts nobody. It works from your own records — and where those cannot answer a question, it says which question a real win-loss interview would need to answer.
Reviews
Reviews open once customers have run this agent.
Need this agent to work differently?
Tell us what to change and our team will scope a customised version for your business.
Customize this agent