Skip to content

Makes

Buying moments

Turns your deals, interviews and sales notes into your ideal customer, the buying group and the moments that make buyers start looking, ranked for ads and search.

Other tools

Add this skill to Claude

Claude takes skills as a ZIP file. Your download has started; if it didn't, use the button below.

  1. 1In Claude, open Customize, then Skills.
  2. 2Choose +, then Upload a skill, and pick buying-moments.zip.
  3. 3Ask for it by name, or let Claude use it when a task matches.

Skills need code execution turned on in Claude's settings. On a Team or Enterprise plan, an admin can share it with everyone.

Download the folder

Try it now, without installing

See the skill work on a made-up input before you add it. Nothing is installed, and you can swap in your own input.

Open in the Claude app

Opens a new chat in the Claude desktop app with the prompt filled in. Nothing is sent until you press enter.

Includes the whole SKILL.md and the sample input. Paste it into Claude, ChatGPT or any chat.

Influenced by
Category entry points, Switch interviews
Needs
Won and lost deals with why-now notes and the contacts involved
Works with
Claude, ChatGPT, Codex, Cursor and any agent that reads SKILL.md
Version
1.0.0 · Last checked · License MIT
Maintained by
Sabih AhmedCreative
SkillStrategy and messaging

Buying moments

Example output

Rank the moments that start a search
MomentSourcesPriority
An audit date is set9 of 20 storiesHigh
A new CFO wants a faster month-end4 of 20 storiesMedium
Moving to a new ERP1 sourceWatch

Turns your deals, interviews and sales notes into your ideal customer, the buying group and the moments that make buyers start looking, ranked for ads and search.

The skill

The whole skill, in plain text.

Nothing is hidden. Read it, change it to fit your team, or take the whole folder.

Scripts
Optional Python scripts for the counting. It works without them and says which way it ran.
Network
Opens web pages or your tools if your agent can. Works from pasted data if not.

3 test cases

buying-moments/SKILL.md

In the folder

  • buying-moments/SKILL.md
  • buying-moments/references/switch-interview.md
  • buying-moments/scripts/moment_score.py
---name: buying-momentsdescription: >-  Turns won and lost deals, buyer interviews, sales notes and support tickets into the ideal customer, the buying  group and the moments that start a search for the category, each traced to its sources, worded in the buyer's own  phrasing and scored for priority. Writes the profile's Buyers and Buying moments sections, which seed ad themes,  keyword selection and AI answer prompt panels. Returns READY, DRAFT, NEEDS DECISION or BLOCKED after its own  check. Use it after positioning and before messaging, ads or search work, or when new interviews come in. Not for  deciding the category or what makes the company different (use positioning) or for writing copy (use  messaging-house).license: MITcompatibility: >-  Works in Claude, ChatGPT, Codex, Cursor and any agent that reads SKILL.md. Uses a connected CRM, call recorder or  help desk if the agent has one, otherwise exports, transcripts or pasted notes. Python 3 is optional, for scoring.metadata:  author: SHAKEUP  version: "1.0.0"  stage: make  plugin: strategy-messaging  homepage: https://www.shakeuphq.com/skills/buying-moments--- # Buying moments Finds who buys, who else has to agree, and the events that make them start looking, so ads, pages and prompt panels start where buyers start. ## Purpose - One job: write `## Buyers` (ideal customer, bad fit, buying group) and `## Buying moments` in `company-profile.md` from buyer evidence.- Stage: make. It ends with its own check, at most 2 rounds.- A buying moment is an event that starts a search for the category, told from the buyer's side ("an audit date is set"), never a feature or a pain adjective.- It never invents a buyer, a quote, a search phrase or a search volume. A phrase no buyer said is marked `inferred`. ## Use when / not | Situation | Use || --- | --- || The profile has no buyers or buying moments, or they were written from opinion | this skill || New win and loss interviews, or a new segment is winning | this skill, to update || Before `ad-themes`, `keyword-selection` or `ai-answer-panel` on a new company | this skill first || You need the category and what makes the company different | `positioning` (run it first; this skill reads its best fit) || You need the words for a page or an ad | `messaging-house`, then the copy skills | ## Inputs | Input | Required | Ways to provide, in order | If missing || --- | --- | --- | --- || Won and lost deals with why-now notes, contact roles and the role that signed | yes, or interviews | a connected CRM (closed deals, contact roles, first-call notes); a CRM export; a pasted table | ask once; without deals or interviews, `BLOCKED` || Switch interviews with recent buyers (and buyers who chose something else) | no, but primary | uploaded transcripts; a connected call recorder; pasted notes | moments come from deal notes only; coverage `partial`; offer the guide in `references/switch-interview.md` || Sales call notes and recordings | no | a connected call recorder; exports; pasted notes | fewer role objections; coverage `partial` || Support tickets, onboarding notes, reviews | no | a connected help desk; an export; pasted text | skip; only lines about why they bought count anyway || Search phrasing (Search Console queries, ad search terms) | no | a connected Search Console or ad account; an export; pasted rows | phrases stay as buyers said them or `inferred` | ## Profile read From `company-profile.md`: `## Company` (category, business model, ACV, sales cycle, regions), `## Product`, `## Buyers` (the current version), `## Positioning` (best-fit customers, competitive alternatives, unique attributes), `## Competitors`, `## Accounts and data`, and every `## Learnings` entry for `buying-moments` with status `noted`, `fact` or `accepted`. No profile: ask in one message for what the product does, the deal list or interview notes, and who usually signs. Say that `company-profile` saves this next time. ## Steps 1. **Apply learnings.** Accepted learnings for this skill override the defaults below. Name each one that fires. 2. **Label the evidence.** Give every item an id (`D1` deals, `I1` interviews, `C1` calls, `T1` tickets, `R1` reviews, `Q1` search rows). Sort each by what it can show:    | Source | What it can show | Counts as |   | --- | --- | --- |   | Switch interview, decision in the last 12 months | The whole timeline, the 4 forces, the buyer's words | primary (memory of the first thought fades after a year) |   | Won deal with why-now notes | The trigger, the roles, the signer | primary |   | Lost or no-decision deal notes | What held them back, the role that blocked | primary for worries and habits |   | Sales call transcript | Words, roles, objections | primary |   | Ticket, onboarding note, review | Mostly use after purchase; only "why we bought" lines count | supporting |   | Search rows | How buyers phrase a need | phrasing only, never a moment on its own |    Drop how-to and bug tickets from moment work: they describe using the product, not starting a search. 3. **Write the ideal customer and bad fit.**   - If `## Positioning` has best-fit customers, start there and confirm it against this evidence. Don't re-decide it.   - Otherwise, put a trait in the ideal customer when at least half of won deals share it (a majority describes the typical win, not an edge case), and add the situation they were in.   - Bad fit: traits that show up in 3 or more lost, no-decision or churned deals and in fewer than a quarter of won deals.   - If the evidence says the ideal customer differs from the profile's confirmed one or from `## Positioning`, stop at `NEEDS DECISION`. 4. **Map the buying group.** List every role that appears in the deals: who found the product, who ran the evaluation, who signed, who blocked, who uses it.   - Keep a role that appears in 3 or more deals. 1 or 2 appearances is that account's org chart, not the pattern.   - For each role write: what they care about, their objection, and their part (`champion`, `decision maker`, `influencer`, `blocker`, `user`), each from the evidence with ids. An objection nobody voiced is `not available`, not a guess.   - The decision maker is whoever signed or owned the budget in the deals, not the most senior title. 5. **Rebuild each buyer's timeline.** For every interview and every won deal with enough notes, write the sequence in the buyer's words with dates where given: the first thought (what made the old way stop being good enough), passive looking, the event that turned it into active looking, active looking (what they searched, asked or who they asked), deciding (what almost stopped them), the purchase. Record the 4 forces in each:    | Force | Question it answers |   | --- | --- |   | Push | What was going wrong with the current way? |   | Pull | What did they hope the new way would give them? |   | Worry | What did they fear could go wrong with something new? |   | Habit | What kept them with the current way, even while it hurt? |    Interview guide and how to recruit: `references/switch-interview.md`. 6. **Group timelines into moments.** A moment is the event that turned passive looking into active looking, plus the situation it happens in. Describe each with 5 questions: who it happens to (the role), what they need done, when (the event or deadline), where (the meeting, team or tool where it surfaces), why now (the push).   - Merge timelines with the same event, even when worded differently.   - Split them when the role or the job differs: "an audit date is set" for a finance manager (clean receipts) and for an IT lead (access logs) are 2 moments, because the job differs.   - Name the moment as the event, in the buyer's terms. Rewrite any name that is a feature ("needs multi-touch") or an adjective ("frustrated with data"). 7. **Capture how they looked.** For each moment, quote what buyers typed into search, asked an AI assistant, or asked a peer or community, with the source id. When no buyer said it, write 1 or 2 candidate phrases marked `inferred` and send them to `keyword-selection` or `ai-answer-panel` to confirm. Never attach a search volume here. 8. **Score and prioritise.** Count the sources per moment, then run:    ```   python3 scripts/moment_score.py --json moments.json   ```    Without Python, apply the same rules by hand and say so:    | Factor | Rule | Why |   | --- | --- | --- |   | Park | Credibility `none` (the product can't serve this moment yet) | A moment we can't win wastes the campaign that targets it |   | Watch | Fewer than 2 sources | 1 source is an anecdote |   | Commonness | Share of buyer stories (won, lost, interviews): 25% or more +3; 10% to 24% +2; under 10% +1 | 1 in 4 stories is common enough to carry its own campaign; 1 in 10 is a real supporting angle |   | Win rate | With 5 or more decided deals where it appears: at or above overall +1; below overall by more than one deal's swing (100 / deals) -1 | Moments where we win deserve the budget first |   | Deal size | Median ACV of its won deals at or above the overall median +1 | Bigger deals pay for more reach |   | Competition | Low +1, medium 0, high -1, unknown 0 | A moment 2 or more competitors already own costs more to win |   | Credibility | Full (a unique attribute or claim with proof covers it) +1; partial 0 | We can prove we solve it |   | Priority | High 5 or more; medium 3 to 4; low 2 or less | |    Competition comes from `ai-answer-panel` results, ad libraries or competitor pages when available: `high` when 2 or more competitors lead with the moment, `low` when none do. Without that data, write `unknown`. 9. **Write the sections.** Replace `## Buyers` and `## Buying moments` using the template's exact line formats:    ```   Ideal customer: <company type, size, industry, trigger>   Bad fit: <who, and why, with deal ids>    ### Buying group    - <Role>: <what they care about> | <their objection> | <champion, decision maker, influencer or blocker>    ## Buying moments    - <Moment>: <what happens> | <what the buyer searches or asks> | <priority: high, medium, low>   ```    Parked and watch moments stay in the output, not in the profile. If the agent can't edit the profile file (a chat app, or a profile pasted into the chat), give both sections in one code block ready to paste and set `profile_sections_written: [] (ready to paste)`. ## Check and verdict | Verdict | Condition (first match wins) || --- | --- || `BLOCKED` | No buyer evidence of any kind: no deals, interviews, call notes, tickets or reviews || `NEEDS DECISION` | The ideal customer the evidence shows conflicts with `## Positioning` best fit or the profile's confirmed ideal customer || `DRAFT` | Fewer than 3 primary sources; or a high-priority moment rests on a single source type; or a loop check still fails after 2 rounds || `READY` | Every loop check passes | Named stops: - `BLOCKED`: stop. Don't write the profile. Give the interview plan from `references/switch-interview.md` and the CRM report to pull. If the reader asks for a brainstorm anyway, label every line `hypothesis` and keep it out of the profile.- `NEEDS DECISION`: stop. Show both ideal customers with the deal ids behind each. Write nothing until the reader picks, or reruns `positioning`.- Overwriting a `## Buyers` the reader marked confirmed: show the old and new side by side and write after the reader says yes. `DRAFT` writes the sections with every unproven line marked `hypothesis`. `READY` writes without a stop. ## Loop Check the draft, fix what fails, check again. At most 2 rounds, then hand over with what is still open, labelled. | Check | Fails when | Fix || --- | --- | --- || Sources | A moment in the profile has fewer than 2 sources | Move it to watch || Buyer words | A search phrase or objection has no source id and isn't marked `inferred` | Add the id, or mark it || No invented numbers | Any count, rate or volume isn't from the evidence | Remove it or write `not available` || Events, not features | A moment name is a feature, a product term or an adjective | Rename it as the event || Before the purchase | A moment comes only from after-purchase tickets | Drop it from moments || Roles | A buying group role lacks an objection or a part | Fill from evidence, or `not available` || Format | A line doesn't match the template format | Rewrite it to the format | ## Learn Append to `## Learnings`: - the reader changes a moment's priority or merges 2 moments: a `proposal` with their reason;- results come back for a moment, such as the keywords `keyword-selection` confirmed, the prompts `ai-answer-panel` ran, or ad results by theme from `ad-themes`: a `fact` with the source;- new interviews change a moment's count: a `fact` with the new count and ids. Example: `- 2026-10-24 | buying-moments | fact | "A new CFO wants a faster month-end" seen in 2 more interviews, now 6 of 20 stories | Why: switch interviews I7, I8 (2026-10) | Status: noted` ## Output ```Verdict: <READY | DRAFT | NEEDS DECISION | BLOCKED> · Coverage: <complete | partial (what's missing) | blocked> · Scoring: <script | by hand>Evidence: <n> won, <n> lost, <n> interviews, <n> call notes, <n> supporting (<n> dropped as after-purchase) Ideal customer: <line> (from <ids>)Bad fit: <line> (from <ids>) Buying group| Role | Cares about | Objection | Part | Deals | Moments| Moment | Who | What happens | Searched or asked (source or inferred) | Sources | Points | Priority |Watch: <moments with 1 source>Parked: <moments the product can't serve yet> Forces seen most often: push <..>, pull <..>, worry <..>, habit <..> ## Buyers and ## Buying moments   (the sections as written, or "not written: <stop>") Loop: round <1 or 2>, <checks passed>/<7>; open: <list>Learnings applied: <list, or none>``` Then the handoff block. ## Hand off ```handoffskill: buying-momentsversion: 1.0.0verdict: READYcoverage: partial (no search data; 2 phrases inferred)for: <company>outputs:  top_moments: ["An audit date is set", "A new CFO wants a faster month-end"]  phrases: ["receipt matching software", "expense policy software (inferred)"]  decision_maker: Controller  blocker: IT reviewer  profile_sections_written: [Buyers, Buying moments]next: [messaging-house, keyword-selection, ad-themes, ai-answer-panel]open: [confirm 2 inferred phrases with keyword-selection]``` Made-up values for illustration. `messaging-house` reads the buying group and moments for pillar order; `keyword-selection` reads `phrases`; `ad-themes` reads `top_moments`; `ai-answer-panel` turns each high and medium moment into prompts. ## Failure modes | Problem | Do this || --- | --- || Only the founder's or sales lead's opinion is available | Record it as `hypothesis`, verdict `DRAFT` at best; plan 5 switch interviews first || Interviews only with happy customers | Add buyers who chose something else or nothing; without them, worries and habits are blind; say so || The CRM has no contact roles | Use first-call notes and signers; mark the buying group coverage `partial` || Every moment looks the same ("needs better data") | The event is missing. Ask what happened the week they started looking || The reader wants search volumes per moment | Not this skill's job. Hand the phrases to `keyword-selection` || A moment only exists in one region or segment | Keep it, with the segment named in "who" | ## Influenced by - Jenni Romaniuk's category entry points, as set out for B2B by LinkedIn's B2B Institute: the buying situations a brand should be linked to, described by who, what, when, where and why, and prioritised by how common, how contested and how credible they are: https://business.linkedin.com/advertise/resources/b2b-institute/cep-in-b2b- Bob Moesta's switch interviews, which rebuild the timeline from the first thought to the purchase and the forces that pushed and held the buyer: https://businessofsoftware.org/talks/understanding-your-customer-jtbd/- What we changed: the buying situations come from deal and interview evidence with ids, the 3 prioritising questions became a scored rule with thresholds, the switch timeline is the unit of evidence, every search phrase is quoted or marked inferred, and the result is written into the company profile for ad, keyword and AI answer skills to read.

How it thinks

Inputs in, a verdict out, and a stop only where you decide.

What we built

We made the switch interview timeline the unit of evidence, and turned how common, how contested and how credible a moment is into a scored rule with thresholds. Every search phrase is quoted from a buyer or marked inferred, so the ad and search skills never build on a guess.

It reads your company profile if you have one, or asks for only what it needs. Then it works through its steps, checks its own work and ends in a verdict.

The steps

  1. 01

    Read your learnings first

    Rules your team accepted for this skill, and results from past moments, change the defaults before it starts.

  2. 02

    Sort the evidence

    Interviews and won deals with why-now notes come first. How-to and bug tickets are dropped, because they describe using the product, not starting a search.

  3. 03

    Write who buys and who doesn't

    It starts from your positioning's best fit and checks it against the deals. A trait goes in when at least half of your won deals share it.

  4. 04

    Map the buying group

    A role goes in when it shows up in 3 or more deals, with what it cares about and its objection in its own words. The decision maker is whoever signed.

  5. 05

    Rebuild each buyer's timeline

    From the first thought to the purchase, with what pushed them, what pulled them, what worried them and what kept them with the old way.

  6. 06

    Group timelines into moments

    A moment is the event that turned looking into searching, named in the buyer's terms. Each one keeps the phrases buyers typed or asked, with their source.

  7. 07

    Score, rank and write

    A script scores how common, how winnable, how contested and how credible each moment is. The result goes into your profile, and 1-source moments stay on a watch list.

What you need

  • Required

    Won and lost deals with why-now notes and the contacts involved

    A connected CRM with contact roles and first-call notes → A CRM export → A table you paste

  • Optional

    Switch interviews with recent buyers, and with buyers who chose something else

    Transcripts you upload → A connected call recorder → Notes you paste

  • Optional

    Sales call notes and recordings

    A connected call recorder → Exports or notes you paste

  • Optional

    Support tickets, onboarding notes and reviews

    A connected help desk → An export or text you paste

  • Optional

    How buyers phrase the need, from Search Console or ad search terms

    A connected Search Console or ad account → An export or rows you paste

What it can return

  • READY

    Every moment has 2 or more sources, every phrase is quoted or marked inferred, and every role has its objection.

  • DRAFT

    There are fewer than 3 interviews or won deals, or a high-priority moment rests on 1 kind of source. Unproven lines are marked as hypotheses.

  • NEEDS DECISION

    The deals point to a different ideal customer than your positioning or your confirmed profile.

  • BLOCKED

    There's no buyer evidence of any kind. It gives you an interview plan instead of guessing.

Where it stops for you

  • When there's no buyer evidence at all
  • When the evidence points to a different ideal customer than your positioning
  • Before it replaces a buying group you confirmed

Example output

What you get back.

A run for a made-up company. Your output depends on your inputs.

Here's a run for a made-up B2B SaaS company that sells expense software to finance teams. It had 6 switch interviews, 10 won deals and 4 lost deals.

Verdict

READY

Coverage: partial (no search data, so 1 phrase is inferred)

Ideal customer

finance teams at companies with 200 to 1,000 employees and a card program (7 of 10 won deals). An audit or a new CFO usually sets the deadline.

Bad fit

companies with fewer than 20 cards, who stalled on price in 3 of 4 lost deals.

RoleCares aboutObjectionPart
Finance managerA clean audit with no receipt chasing"If the card feed breaks, I'm the one who rebuilds it"Champion
ControllerClosing the books before the board pack"Another system my team has to learn"Decision maker
IT reviewerSingle sign-on and the ERP link"Who maintains the integration?"Blocker until shown
MomentSearched or askedSourcesPriority
An audit date is set"receipt matching software" (I3), asked peers in a finance community (I1)9 of 20 storiesHigh
A new CFO wants a faster month-end"close the books faster" (I5)4 of 20 storiesMedium
The card program passes 50 cards"expense policy software" (inferred)3 of 20 storiesLow
Moving to a new ERPnone1 sourceWatch
Forces seen most often

receipts missing at audit time push buyers to look, and a clean audit pulls them. A broken card feed worries them, while the shared spreadsheet keeps them where they are.

The handoff block the next skill reads
skill: buying-moments
version: 1.0.0
verdict: READY
coverage: partial (no search data, so 1 phrase is inferred)
for: expense software company
outputs:
  top_moments: ["An audit date is set", "A new CFO wants a faster month-end"]
  phrases: ["receipt matching software", "close the books faster", "expense policy software (inferred)"]
  decision_maker: Controller
  blocker: IT reviewer
  profile_sections_written: [Buyers, Buying moments]
next: [messaging-house, keyword-selection, ad-themes, ai-answer-panel]
open: [confirm "expense policy software" with keyword-selection]

What good looks like

  • Every moment named as an event, in the buyer's words
  • Source ids beside every moment, role and objection
  • Search phrases quoted from buyers, or marked inferred
  • Moments with 1 source kept on a watch list, out of the profile
  • A buying group led by the person who signed, not the most senior title

Install

Use it in the tool you already have.

Pick your tool. Each tab has the shortest path we know of.

  1. 01
  2. 02
    In Claude, open Customize, then Skills.
  3. 03
    Choose +, then Upload a skill, and pick the ZIP. Code execution must be on in your settings.
  4. 04
    Ask for it by name, or let Claude use it when a task matches.

Questions

Before you add it.

What's a buying moment?

An event that makes a buyer start looking for your category, told from their side, like "an audit date was set". It's never a feature or a feeling.

Do we need interviews?

They're the best source, and won deals with why-now notes work too. With neither, it stops and gives you an interview plan instead of guessing.

Where do the moments go next?

Into your profile. Ad themes take the top moments, keyword selection confirms the phrases with your data, and the AI answer panel turns each moment into prompts.

Will it give search volumes?

No. It quotes what buyers said they typed or asked. Keyword selection checks demand with your own data.