← Back to the exerciseThe full worked example: "We're losing on price": test it before you give away margin
campo.
Field guide / Sales / UC-006 · SC-02
The result

"29 of 41 lost on price"? Price is the first option in a required field. 7 of 12 customers we asked pointed to speed.

A VP Sales at a 7,200-person components maker works through her regional VPs' request for more discount authority before the pricing council, with any AI chat her company allows.

The resultWhat Carla takes to the pricing council on December 3. Fictional company.
NARROWGATE COMPONENTS · SALES · NORTH AMERICA
Discount authority: decision note
For: Daniel Ash, CRO, and the pricing council
From: Carla Ruiz · December 3

MY RECOMMENDATION

Don't approve 18% authority on the current evidence; keep the request open. The "price" count comes from a required field where Price is the first option, and "15% under us" is one buyer's comment on one order. In 12 interviews with lost customers (June to September), 7 pointed to speed and 3 to price. · Inputs 1, 2, 4, 6

WHAT THE EVIDENCE SHOWS

CRM: lost on price29 of 41
Price is the first option in a required field
Interviews: speed7 of 12
June to September; small sample
Quote time, lost vs wonAbout 2x
Two systems, two units; not yet checked
Cost of more discount$0.65M a point
Per quarter, on Q3 won volume

WHAT I'M ASKING THE COUNCIL FOR

APPROVE
A 1-business-day approval target for discounts from 10% to 18%
Only if finance confirms it can meet it. Today's median is 4 business days. · Input 5
ASK
Sales operations: quote time for won and lost deals from one system, in the same units, and the 41 losses re-checked
· Inputs 2, 3
REGIONS
East and West bring documented competitor quotes, like for like
If price still leads, a capped authority comes back to the next council. · Inputs 1, 6
Caught on review
1
"29 of 41 on price" comes from a required field where Price is the first option
2
"15% under us" is one buyer's comment on one order
3
The quote-time figures mix business and calendar days from two systems
UC-006 · SC-02 · Rev A
FOR
VP Sales, North America
SECTOR
Automotive components (IATF 16949)
COMPANY
7,200 people · 4 plants · US, Mexico
TASK
Decide on a request for more discount authority
WRITTEN FOR
ChatGPT · Claude · Gemini · Copilot
AI TEST
Claude, Nov 14, 2026 · 7 of 7 found
Step 0 · The mess

What landed on Carla's desk

A confident email from her two regional VPs, a CRM export, quote times from two systems, notes from customer interviews, a note from finance and one forwarded comment from a buyer. The request reads well. The evidence is thinner.

EMAIL · REGIONAL VPS[1]
Discount authority
“We lost 41 deals … 29 of them on price. Competitors are 15% under us.”
1
Three claims.
CRM EXPORT[2]
Lost deals Q3
Loss reason is a required field … “Price” is the first option in the list.
2
How was it recorded?
QUOTE TURNAROUND[3]
Sales ops
Won: 3.1 business days. Lost: 9.4 calendar days.
3
Same units?
WIN/LOSS INTERVIEWS[4]
Lost customers
7 of 12: another supplier sent a usable quote first.
4
What customers said.
NOTE · FINANCE[5]
Discount and margin
Average discount 11.2% … approval takes 4 business days … $0.65M a point.
5
The cost.
FORWARDED COMMENT[6]
Competitor
“Your competitor came in 15% under you on the 200-unit order.”
6
One order.
You're not alone
82%
of the time, the vendor a B2B buying group already favoured before contacting any seller won the deal.
6sense research, survey of 4,611 recent B2B buyers, November 2026. Source
Steps 1–3 · How it works

Three steps, one chat

Here the AI works through a decision your own team is pushing. The prompt asks it to check the reason before the fix.

1
Open a new chat

Use the AI tool your company approves: ChatGPT, Claude, Gemini or Copilot.

2
Paste the prompt and the inputs

Copy the prompt and the six fictional inputs from the appendix into one message.

3
Convert one figure yourself

Put both quote times in the same units and compare them. Then decide what you'd ask finance and the regions for.

The prompt, part by part
Copy-ready version in the appendix

I'm VP Sales for North America at an industrial components maker. My two regional VPs want authority to discount up to 18% without approval, because they say we're losing on price. Our CRO decides at the pricing council on Thursday, December 3. Today is Tuesday, December 1. Below are their email, our CRM loss export, the quote turnaround figures, notes from win/loss interviews, a note from finance and the competitor evidence they sent.

1
Who you are, who decides, when, and today's date.

Please give me a one-page decision note:

  1. The decision, in one sentence, and the options, including any nobody has proposed.
  2. Why we lose: what the evidence supports, what it only suggests, and what it doesn't show.
  3. Each option: what it would cost, what it would fix, and what would have to be true.
  4. My recommendation for Thursday, or "not yet" with what would settle it.
2
The decision and every option, why you lose (supported, suggested, not shown), the cost of each option, and a recommendation or "not yet".

Before you accept any reason for losing, check how that reason was recorded and how much evidence sits behind it. Keep units and periods consistent. Don't add facts that aren't in the inputs; if you assume something, say so.

3
The heavy lifting: check how each reason for losing was recorded and how much evidence sits behind it, and keep units and periods consistent. It found all seven problems.
Where to paste
Every tool has one message box. Paste the prompt, add a blank line, paste the inputs, send. CRM exports can be attached as files if your tool allows it.
Safe to try
Practise with the fictional inputs first. Real pricing, margins and customer names are confidential: use only your company's approved tool.
The AI's first answer

What Claude actually wrote

Exactly as it came back, not edited. Read it the way Carla did: check each point against the source it cites.

UNEDITED
Claude · fresh session, no other context · Nov 14, 2026 · one run

Decision note: regional discount authority (pricing council, Thursday, December 3)

1. The decision Should East and West regional VPs be allowed to discount up to 18% without approval, or does the evidence point to a different fix?

Options:

  • A. Approve the request: up to 18% at regional level, no approval needed.
  • B. A narrower version: a lower cap, or 18% only where there is a documented competitor quote, for a fixed trial period.
  • C. Not proposed by anyone: fix speed. Set a fast approval deadline for discounts above 10% (today's median is 4 business days) and cut quote turnaround. found
  • D. Not proposed by anyone: decide nothing on discount authority yet. Get the missing evidence first (section 4), then decide.

2. Why we lose

Supported by the evidence

  • Quote speed is the most common reason customers give. In the win/loss interviews it was 7 of 12 ("another supplier sent a usable quote first, or met their deadline"). Price was 3 of 12 and product specification 2 of 12 [4]. found
  • Approval adds delay. Won deals averaged 11.2% off list, and anything above 10% needs finance approval, with a median wait of 4 business days [5]. So many of our quotes probably have to wait for approval. That is an assumption: finance gives the average, not how many deals went above 10%. found

Suggested only

  • Lost deals seem to be quoted more slowly: 9.4 days for lost deals against 3.1 for won deals [3]. But the two figures are not comparable as given. found
    • Lost deals are measured in calendar days, from CRM activity dates. Won deals are measured in business days, from the quoting system.
    • Converting 9.4 calendar days at 5/7 gives about 6.7 business days, assuming no holidays. That is still about twice as slow, but the two figures come from different systems.
    • We are also not told that the 118 won deals are over $50K like the lost ones, so the groups may differ.
  • Some losses really are on price. Three interviews say so, plus one buyer's comment [6].

Not shown

  • That 29 of 41 losses were on price. Loss reason is a required field before a lost deal can be closed, and "Price" is the first option in the list [2]. 29 of 41 (71%) is the CRM field, not checked evidence. Customers' own answers put price at 3 of 12 (25%). found
  • That "competitors are 15% under us." This rests on one buyer's remark about one 200-unit order [6]. We don't know whether that is 15% under our list price or our net price (we already average 11.2% off list), or whether the specification and terms were the same. found
  • That the 12 interviews are among the 41 lost deals. The interviews ran June to September, while the export covers Q3 and deals over $50K only.
  • That 18% would have won these deals, or that we "will miss Q1". No input supports either claim.

3. The options

Option Cost (on Q3 won volume of $58.0M; finance says each extra point of average discount costs about $0.65M) What it would fix What would have to be true
A. 18% authority Depends on how far the average rises. If it rose from 11.2% to 15% (an illustration, not a forecast), about $2.5M a quarter. If every won deal went to 18%, about $4.4M a quarter. Margin is given away on deals we would have won anyway. The 3 of 12 interviewed losses that were about price. Probably some speed too, because these deals would no longer wait 4 days for approval. Price really is the main reason we lose, and competitors really are about 15% under us on a like-for-like basis. Neither is shown. found
B. Narrower or conditional authority Lower, and capped by design. It cannot be costed until a cap is chosen. Documented price losses, while protecting margin. Some approval delay disappears. Regions can produce real competitor quotes. Someone checks how the discount is used.
C. Approval and quote speed Little direct discount cost. The effort is in finance and sales operations (amount not given). The most common reason customers gave (7 of 12). Also the 4-day approval wait. The speed gap holds up when won and lost deals are measured the same way. Approval is a large part of the delay.
D. Not yet One cycle of delay. Avoids locking in a costly policy based on weak evidence. The missing evidence can be gathered quickly.

4. My recommendation for Thursday

Not yet on 18%. Don't approve blanket 18% authority on the current evidence. Keep the question open. In the meantime, ask the council to approve option C now, because it costs little and matches what customers told us: a finance approval deadline (for example 1 business day) for discounts from 10% to 18%, and a quote-turnaround target. B: finance agrees first

Revisit 18% at the next council meeting, once these are known: C: one brief to sales ops

  1. Turnaround for won and lost deals, measured from the same system in the same units, for the same deal size (over $50K).
  2. The 41 lost deals re-checked from deal notes or customer contact, rather than from the default "Price" choice in the CRM.
  3. Whether the 12 interviews overlap with the 41 lost deals.
  4. How many lost deals were waiting for discount approval when the customer decided.
  5. Competitor quotes compared like for like: net price, same specification, same quantity.
  6. Finance's estimate of where the average discount would land under 18% authority. Both cost figures above are my illustrations.

If the re-checked data still shows price as the main cause, option B with a clear cap is the safer next step than A. A: give the regions a role

foundcaught a planted problemA–CCarla changes it; see the next section
The human part

What Carla changed, and why

Claude found every planted problem, said "not yet" instead of "no", and found the cheaper fix. Carla's edits are about her own team, who owns the fix, and what the council needs.

What Claude found: 7 of 7 planted problems
One run
✓“Price” is a default option
✓Quote times in two units
✓Interviews: small, mixed period
✓“15% under” is one order
✓Approval adds 4 days
✓The cost of the request
✓The option nobody proposed
Three edits, all judgment calls
A
CLAUDE WROTE
“If the re-checked data still shows price as the main cause, option B with a clear cap is the safer next step than A.”
CARLA'S VERSION
Carla calls both regional VPs before Thursday and asks them to bring documented competitor quotes, like for like. Their case gets tested on evidence, not turned down.
WHY · RELATIONSHIP
They will hear "not yet" as "no". Giving them the job of producing the price evidence keeps them on side and gets the data the decision needs.
B
CLAUDE WROTE
“Ask the council to approve option C now … a finance approval deadline (for example 1 business day).”
CARLA'S VERSION
Carla checks with finance before Thursday whether a 1-business-day approval is workable. The council approves it only if finance can meet it.
WHY · OWNER
Finance approves the discounts and hasn't agreed to the deadline. Promising it without them sets up the next missed target.
C
CLAUDE WROTE
Six data requests listed for the council.
CARLA'S VERSION
The council sees the recommendation, four numbers and three asks. The data requests go to sales operations as one brief for the next council.
WHY · LENGTH
The council is deciding whether to grant authority. Sales operations needs the full list.
What only you can do
Bring your own team along when the answer is "not yet", check the fix with the people who'd have to deliver it, and give the council only what it decides. The AI weighed the evidence. You lead the team.
Your turn

Make it yours

Use it whenever your team asks for something to fix a problem they've already diagnosed: more discount, more headcount, a new tool, a policy exception.

Your prompt: fill in the brackets

I'm the [your role] at a [industry] company. [Who] wants [the request], because they say [the reason]. [Who decides] decides on [date]. Today is [date]. Below are [the request, the data behind the reason, any customer feedback, finance's view].

Please give me a one-page decision note: 1. The decision, in one sentence, and the options, including any nobody has proposed. 2. Why it's happening: what the evidence supports, what it only suggests, and what it doesn't show. 3. Each option: what it would cost, what it would fix, and what would have to be true. 4. My recommendation for [date], or "not yet" with what would settle it.

Before you accept any reason, check how it was recorded and how much evidence sits behind it. Keep units and periods consistent. Don't add facts that aren't in the inputs; if you assume something, say so.

Five checks before you decide
1
You know how the reason was recorded, not just how often.
2
Customer evidence is counted, with its sample size and period.
3
Figures compared side by side are in the same units, from comparable sources.
4
The cost of the request is stated, per point or per head.
5
Whoever owns the cheaper fix has agreed to it.
Make it repeatable (optional)
Save the prompt once and reuse it whenever a team asks you to fund a fix: a custom GPT in ChatGPT, a project in Claude, a Gem in Gemini, or an agent in Copilot. Some companies switch these off; copy and paste always works.
With real data
Pricing, margins and customer names are confidential. Use only the AI tool your company approves, and follow its rules.
What this example shows
One run of this prompt in Claude found all seven problems planted in fictional inputs and proposed a cheaper fix nobody had asked for.
What it doesn't show
That every tool, setting or run will do the same, how much time it saves you, or whether price or speed decides deals in a real business. Narrowgate Components and its people are fictional.
Appendix

Copy-ready: the prompt and the six inputs

Copy the prompt, then all six inputs, into one message in a new chat. Everything here is fictional and safe to paste.

The prompt

I'm VP Sales for North America at an industrial components maker. My two regional VPs want authority to discount up to 18% without approval, because they say we're losing on price. Our CRO decides at the pricing council on Thursday, December 3. Today is Tuesday, December 1. Below are their email, our CRM loss export, the quote turnaround figures, notes from win/loss interviews, a note from finance and the competitor evidence they sent.

Please give me a one-page decision note:

  1. The decision, in one sentence, and the options, including any nobody has proposed.
  2. Why we lose: what the evidence supports, what it only suggests, and what it doesn't show.
  3. Each option: what it would cost, what it would fix, and what would have to be true.
  4. My recommendation for Thursday, or "not yet" with what would settle it.

Before you accept any reason for losing, check how that reason was recorded and how much evidence sits behind it. Keep units and periods consistent. Don't add facts that aren't in the inputs; if you assume something, say so.

[1] Email from the regional VPs (East and West)

Subject: Discount authority

Carla, we lost 41 deals over $50K last quarter, 29 of them on price. Competitors are 15% under us. Give the regions authority to discount up to 18% without approval, or we'll miss Q1.

[2] CRM export: lost deals over $50K, Q3

41 deals, $23.6M in total. Loss reason: Price 29, Timing 5, Product fit 4, Other 3.

Note from sales operations: loss reason is a required field to close a lost deal, and "Price" is the first option in the list.

[3] Quote turnaround, Q3 (sales operations)

Median time from customer request to quote sent: - Won deals (118): 3.1 business days, from the quoting system. - Lost deals (41): 9.4 calendar days, from CRM activity dates.

[4] Win/loss interviews with customers we lost (June to September)

12 interviews completed. Main reason given: - 7: another supplier sent a usable quote first, or met their deadline - 3: price - 2: product specification

[5] Note from finance

Won deals in Q3: $58.0M booked, at an average discount of 11.2% off list. Discounts above 10% need my approval; the median approval time is 4 business days. Each extra point of average discount on Q3's won volume would have cost about $0.65M.

[6] Competitor evidence, forwarded by the East regional VP

From a customer's buyer: "Your competitor came in 15% under you on the 200-unit order."