pain-mapper

---
name: pain-mapper
description: Turns discovery notes into a costed map of what standing still costs the buyer, priced only in figures the buyer said out loud. Use after a discovery call, before a demo or proposal.
---

# Find the cost of doing nothing, in their numbers

A one page cost of inaction map: three or four problems priced in the buyer's own figures, plus the numbers you still have to go and ask for.

## Claude does

1. Ask for the call notes or transcript, the currency and the deal name. Work only from what was said, never from memory of similar deals.
2. Pull out every problem stated, then keep the three or four carrying most weight. Gong's 519,000 call study found the best discovery uncovers three to four business problems. Nine reads as a survey, not a case.
3. Quote the buyer verbatim under each. No quote means it is a hypothesis, labelled as one.
4. Attach the buyer's metric: unit, quantity, frequency, period. Tag every figure BUYER SAID, SELLER ESTIMATE or MISSING. Never promote an estimate to a buyer figure.
5. Write the missing questions in the buyer's words, using the two openers Gong found buy the longest answers: "Can you help me understand" and "Can you walk me through". Keep the next call inside 11 to 14 questions, fewer with the C suite, where extra questions correlate with harm.
6. Name the indecision under the pain. The JOLT Effect puts 40 to 60 percent of losses on indecision, split 44 percent status quo preference, 56 percent fear of getting it wrong. Say which this account shows.
7. Decide one recommendation and one way to take risk off the table. No menu.

## Then Codex does

Codex takes the arithmetic and layout: quantity by frequency by period per row, the annual total, a low, expected and high range, then the table, a six field situation summary and a CSV. Repetitive, easy to fumble by hand, identical every time.

```
codex exec --sandbox danger-full-access --skip-git-repo-check -C "<working folder>" - < CODEX.md
```

Fill in first: {DEAL_NAME}, {CURRENCY}, {PROBLEM_TABLE} (a line per problem: label, quote, metric, quantity, frequency, period, source tag) and {MISSING_LIST}.

## Claude checks

1. Recompute two rows by hand against the notes. Any disagreement, reject.
2. Every figure carries a source tag. An untagged number was invented.
3. No row gives an outcome in place of a cause. Clozd checked 1,000 closed lost deals against buyer interviews: 44 percent of recorded reasons were outcomes. "They went with a rival" is not a problem.
4. No urgency, deadline or scarcity wording. Urgency backfired 84 percent of the time.
5. Three or four rows, and MISSING rows carry no invented number.

Any failed check, rerun Codex. Never patch the output by hand and call it verified.

## Rules

- Three or four problems, never nine.
- No number without a source tag. MISSING beats a guess.
- Recommend and remove risk instead of applying pressure.
- A problem is a cause, not an outcome.
- Public information only.
- Never invent a fact, a number or a quote.
- Anything sent in someone's name says whose name it is.

## If Codex is not installed

Claude does the arithmetic and layout itself, and says so at the top.

## Built from

- Gong Labs discovery study, 519,000 recorded calls: https://www.gong.io/resources/labs/deal-closing-discovery-call/
- The JOLT Effect, Dixon and McKenna: https://www.jolteffect.com/blog/what-is-the-jolt-effect
- Clozd, 1,000 closed lost deals vs buyer interviews: https://www.clozd.com/blog/5-lies-your-crm-is-telling-you-about-your-buyers
- Great Demo!, Peter Cohan, Seven Validated Habits: the six field Situation Slide. Full URL in SOURCES.md.

# Find the cost of doing nothing, in their numbers - Codex task

## You are given

- {DEAL_NAME}: the account this is for.
- {CURRENCY}: the currency every figure is written in.
- {PROBLEM_TABLE}: three or four lines, one per business problem. Each line holds: problem label, buyer quote, metric unit, quantity, frequency, period, source tag. The source tag is one of BUYER SAID, SELLER ESTIMATE or MISSING.
- {MISSING_LIST}: the questions still to be asked, already written out.

## Produce

Three files in the working folder.

1. `cost-of-inaction.md`
   - Heading with {DEAL_NAME} and the date.
   - One table. Columns: Problem, Buyer's words, Metric, Working, Annual cost, Source.
   - Three or four rows, no more, matching the input exactly.
   - The Working column shows the sum in full, for example 12 x 4 x 52.
   - Buyer figure total, in {CURRENCY}, on its own line.
   - A low, expected and high range: low is buyer figures only, expected adds seller estimates, high applies the highest quantity given for any row where a span was quoted. Say in one line how each was built.
   - A "Still to confirm" list, taken verbatim from {MISSING_LIST}.

2. `situation-summary.md`
   - Six headed fields in this order: Overarching goals, Current situation, Pain and problems, Specific capabilities sought, Value desired, Required timeline.
   - Two sentences maximum per field, drawn only from the input. A field with no input says "not yet given".
   - Must read aloud in about two minutes.

3. `cost-of-inaction.csv`
   - Header row, then the same rows, machine readable, one problem per line.

## Rules

- Show the arithmetic on every row. Never print a bare total.
- A row tagged MISSING gets no number at all. Write "not yet given" and carry the question into the Still to confirm list. Do not estimate, benchmark or interpolate.
- SELLER ESTIMATE rows are totalled separately and never folded into the buyer figure headline.
- Round to whole {CURRENCY} units and state the rounding once, at the foot of the table.
- Do not add urgency, deadlines, discounts, scarcity or a call to action.
- Do not invent quotes, company names, industry averages or comparisons to other customers.
- Do not add rows, columns, sections or files beyond those listed above.
- British English throughout.

## Return

Print, in this order: the row count, the buyer figure total, the seller estimate total, the low, expected and high range, the count of MISSING fields, and the three filenames with their byte sizes. Then print rows one and two of the table in full so the arithmetic can be rechecked against the source notes.