sales-meeting-recap

{
  "skill_name": "sales-meeting-recap",
  "evals": [
    {
      "id": 1,
      "prompt": "Just got out of a demo with Kestrel Freight. Notes: Dana Okonkwo (VP Operations) and Ravi Shah (IT) on their side, me and my SE Tom on ours. Dana said they're \"bleeding about 9 hours a week having two people retype order data from the portal into NetSuite\". Ravi wants to see how our API handles their custom fields - he said he'd need to check with his security team about the questionnaire but didn't put a timeline on it. Dana said she'd \"try to get procurement looped in at some point\". We agreed to do a technical deep dive but nobody picked a day. Deal is around $85K ARR. She also mentioned their CFO Bernard Achterberg has to sign off on anything over $50K, he wasn't on the call. Write me the follow-up email.",
      "expected_output": "A plain-text recap email leading with dated action items, Dana's pain quoted word-for-word, plus an explicit gaps block covering the undated deep dive, Ravi's uncommitted questionnaire, Dana's vague procurement promise, and Bernard's absence.",
      "files": [],
      "expectations": [
        "The action items appear at the very top of the email body, ahead of any greeting pleasantry, thanks-for-your-time line, or summary of what was discussed.",
        "Dana's pain is reproduced word-for-word as \"bleeding about 9 hours a week having two people retype order data from the portal into NetSuite\", not paraphrased into vendor or marketing language.",
        "No date is invented for the technical deep dive; the output either asks the user for one or proposes it as a forced choice between two named days.",
        "The missing technical deep dive date is listed in an open-gaps block rather than filled in with a plausible-looking date.",
        "Ravi's security questionnaire item is labelled \"proposed - confirm before sending\" (or an equivalent explicit marker) because he committed to no timeline.",
        "Dana's \"try to get procurement looped in at some point\" is never carried into the email in that form; it is either converted into an owner, date and deliverable, or listed as a gap to confirm.",
        "Every action item that does appear in the email carries all three of an owner, a calendar date, and a deliverable.",
        "Each action item names one individual person as owner, never a team, department, or \"all\".",
        "Bernard Achterberg is surfaced as a stakeholder gap or a send-list consideration rather than silently omitted, given he approves spend above $50K and was absent.",
        "The email body is plain text with no markdown headers, asterisk bold, or other markup.",
        "The subject line is under 50 characters and anchored to the next step rather than to the meeting that just happened.",
        "The email closes with a short ratification question inviting Dana to correct or add to the record."
      ]
    },
    {
      "id": 2,
      "prompt": "Quick one. I sell a scheduling tool at $290/month. Had a 20-minute call with Marisol Vega, she owns a 6-person dental practice and is the only person who decides anything. She said she's \"double-booking patients at least twice a week and my front desk is drowning\". She wants to be running on the 1st. I booked a 15-minute setup walkthrough for Tuesday the 24th at 9am and the invite is sent. I do about 40 of these conversations a week and I have maybe three minutes each for follow-up. Give me the follow-up email and a mutual action plan so I can keep her on track.",
      "expected_output": "A three-to-five sentence recap built around Marisol's own words and the already-booked walkthrough, with the requested mutual action plan refused outright rather than offered in a lighter form, plus a note that an automated sequence beats hand-written recaps at this volume.",
      "files": [],
      "expectations": [
        "Refuses to build a mutual action plan for this deal and gives the reason: a single decision-maker, transactional motion far below the roughly $50K threshold where a plan pays off.",
        "Does not offer seeded milestones or a full plan as an optional extra anywhere later in the output - those options are removed, not parked at the bottom.",
        "States that the already-booked Tuesday the 24th walkthrough is itself the appropriate level of plan for this deal.",
        "Delivers a short recap of roughly three to five sentences rather than a full committee-weight recap.",
        "Reproduces Marisol's words verbatim: \"double-booking patients at least twice a week and my front desk is drowning\".",
        "Names the single next step with its real date and time (Tuesday the 24th, 9am) and notes the invite is already sent.",
        "The subject line is under 50 characters.",
        "Given 40 conversations a week at roughly three minutes each, flags an automated lifecycle sequence as the higher-efficiency option at that volume, while noting it buys coverage and never the buyer's own words.",
        "No list of product features or capabilities appears anywhere in the email body.",
        "The email body is plain text with no markdown formatting."
      ]
    },
    {
      "id": 3,
      "prompt": "Northwind Grid, $240K, four months in. Buying committee: Alicia Brandt (Director of Grid Ops, my champion), Sam Iyer on security, Kofi Mensah in procurement, and the CTO Lena Vos signs. On today's call Alicia said their regulator audit lands June 30 and they want us live before that. My quarter closes March 31 and my VP is all over me about it. Build me the full mutual action plan and put it in the email tonight so they can see exactly what we need from them to get this signed by end of March.",
      "expected_output": "A short recap email plus a small set of seeded milestones anchored backwards from the June 30 audit, with the full plan deliberately not sent cold, an ask for a working session and sign-off, seller-owned rows, paper-process steps, and milestones running past signature to go-live.",
      "files": [],
      "expectations": [
        "Anchors the plan's dates to Northwind's June 30 regulator audit and their go-live target, working backwards from it.",
        "Explicitly declines to work the plan backwards from the March 31 quarter close, and explains why anchoring to the seller's fiscal calendar undermines the plan.",
        "Does not send a fully built plan cold: proposes only two or three milestones drawn from what was actually agreed on today's call.",
        "Asks Alicia for a working session to build out the rest of the plan together, and for explicit sign-off once it is complete.",
        "Every milestone row carries both a named buyer owner and a named seller owner, so the document reads as mutual rather than as buyer homework.",
        "Includes paper-process milestones - security review or questionnaire, DPA or MSA redlines, procurement onboarding - rather than stopping at evaluation steps.",
        "Extends milestones past signature to implementation or onboarding, go-live, and first measured value.",
        "Each milestone is concrete and verifiable, for example \"security questionnaire returned\", never a vague progress-style row.",
        "Each owner column holds exactly one named person, never a department such as \"procurement\" or \"security\".",
        "Does not assert or imply a March signature date, which the buyer never committed to.",
        "The recap email itself stays to roughly one phone screen with next steps at the top; the plan is delivered as its own separate artifact rather than pasted into the email body in full."
      ]
    },
    {
      "id": 4,
      "prompt": "Had a call with Bellweather Health this afternoon. Priya Raghunathan, Director of Revenue Cycle. Good conversation - she said \"claims denials are eating about $180K a quarter and my team spends Fridays chasing them\". I walked her through the denial-prediction piece and she asked a lot of questions about the scoring model. At the end she said \"this is really interesting, let me take it back to the team and I'll be in touch\". That's where we left it. Draft the follow-up so I keep momentum.",
      "expected_output": "A recap that states plainly there is no committed next step, refuses to invent one, lists the gap, and prescribes the 5 Minute Drill with its three questions for the next touch.",
      "files": [],
      "expectations": [
        "States plainly that the notes contain no committed next step and does not manufacture one.",
        "No fabricated meeting, date, or buyer commitment appears anywhere in the draft.",
        "The missing next step is listed in an open-gaps block.",
        "Prescribes the 5 Minute Drill for the next touch and names all three of its questions: whether they want to buy, when they want to buy, and how they want to buy.",
        "Any next step the user could propose is marked \"proposed - confirm before sending\" or an equivalent explicit marker.",
        "Rejects \"I'll be in touch\" and \"circle back\" style phrasing as a next step, and if a time is proposed it is a forced choice between two named days rather than an open invitation.",
        "Reproduces Priya's words verbatim: \"claims denials are eating about $180K a quarter and my team spends Fridays chasing them\".",
        "Contains no recap of the denial-prediction walkthrough, the scoring model, or any other capability.",
        "Flags that every candidate action is seller-owned and that no buyer-owned step exists in the notes.",
        "States which close-plan rung applies to this deal and why, rather than silently producing or omitting a plan."
      ]
    },
    {
      "id": 5,
      "prompt": "I'm pitching my VP next week on rolling mutual action plans out across the whole AE team. I need numbers for the deck - I've seen a 26% win rate lift quoted and one vendor's site says 57 to 200%. Which one is right, what do I put in the slide, and what should we track afterwards to prove it worked? Right now about 60% of our open opportunities have a scheduled next step.",
      "expected_output": "A refusal to treat either circulated uplift figure as evidence, an explanation of why they fail, a redirect to coordination as the plan's defensible value, and a measurement plan that fixes the sub-benchmark next-step-set rate before any rollout.",
      "files": [],
      "expectations": [
        "Refuses to endorse either the 26% or the 57-200% figure as established fact.",
        "Explains that these are vendor self-reports with no disclosed sample, methodology, or control.",
        "Names selection bias as a specific flaw: buyers willing to co-build a plan were already more engaged than average.",
        "Recommends presenting them, if at all, as directional vendor claims rather than as evidence.",
        "Identifies the team's 60% next-step-set rate as sitting below the roughly 80% practitioner benchmark.",
        "Recommends fixing next-step discipline first and gating the mutual action plan rollout behind it, rather than running both together.",
        "Names next-step-set rate as the metric to instrument first, on the grounds that the CRM already holds the count.",
        "Recommends a before-and-after comparison on the team's own pipeline as the success measure for close plans, instead of any vendor uplift claim.",
        "Names multithreading rate - buyer contacts engaged per deal - as a second metric already sitting in the CRM.",
        "Frames the defensible value of a mutual action plan as coordination - dated and owned milestones, a surfaced paper process, multithreaded visibility - rather than as a win-rate number."
      ]
    },
    {
      "id": 6,
      "prompt": "Real estate. I did a listing presentation this morning with the Calloways, Tomas and Renee, for their place on Fairview Road. Renee said \"we got burned last time, our agent listed it and then we didn't hear from him for three weeks\". They're picking between me and two other agents by Friday. I do four or five of these a week. What do I send them?",
      "expected_output": "A three-to-five sentence note sent within about two hours, built on Renee's exact words, carrying one dated next step offered as a forced choice, with no close plan and no services list.",
      "files": [],
      "expectations": [
        "Delivers a short recap of roughly three to five sentences rather than a full-weight recap.",
        "Reproduces Renee's words verbatim: \"we got burned last time, our agent listed it and then we didn't hear from him for three weeks\".",
        "Recommends sending within roughly two hours of the meeting, not that evening and not the next morning.",
        "Contains exactly one next step, carrying an owner, a real calendar date, and a deliverable.",
        "Proposes the time as a forced choice between two named options rather than \"whenever works for you\" or a request to reply with availability.",
        "Produces no mutual action plan and no seeded milestones for this engagement.",
        "The subject line is under 50 characters.",
        "No list of the agent's services, marketing package, or credentials appears in the body.",
        "The body is plain text with no markdown formatting and no templated enthusiasm such as \"wonderful to connect\" or \"looking forward to working together\".",
        "Applies the same anti-fabrication discipline as a B2B recap: nothing the Calloways did not say or agree to is asserted, and any gap is flagged rather than filled."
      ]
    },
    {
      "id": 7,
      "prompt": "Pasting my notes from the Vireo Analytics discovery call. Can you write the recap email, and also tell me honestly how the call went? I felt like my discovery was weak and I want to know whether I should have pushed harder when she dodged the budget question. Notes: Elena Park (Head of Data) said \"our analysts spend the first two hours of every day just reconciling numbers from three systems\". I asked about budget and she deflected, said it depends on the scope. Agreed on the call: I send the security overview by Friday March 6; Elena gets me the list of the three source systems by Wednesday March 4; follow-up call Tuesday March 10 at 2pm, invite already sent.",
      "expected_output": "The recap email delivered as asked, with the call-quality critique declined and handed off to the call-review skill, and the three agreed commitments carried through exactly as stated.",
      "files": [],
      "expectations": [
        "Declines to score, grade, or critique how the call was run, including the discovery quality and the handling of the budget question.",
        "Points the user to the separate call-review skill, referenced as mbfinotti/sales-skills@sales-call-review.",
        "Still produces the recap email rather than refusing the request as a whole.",
        "The action items appear at the top of the email body, above any greeting pleasantry or summary of the discussion.",
        "Elena's own item - the list of three source systems by Wednesday March 4 - is present and visually distinguished from the seller's items.",
        "Reproduces Elena's words verbatim: \"our analysts spend the first two hours of every day just reconciling numbers from three systems\".",
        "All three agreed commitments appear with the owner, date and deliverable exactly as stated in the notes, and no fourth commitment is added.",
        "The March 10 call is shown with its date, its 2pm time, and the fact the invite is already sent.",
        "The deflected budget answer is not presented as a resolved figure, an assumed range, or a qualified budget."
      ]
    },
    {
      "id": 8,
      "prompt": "Halcyon Systems, $120K. On the call: Jonas Brekke (Director of Engineering, he's the one driving this), plus his two staff engineers Mei Tanaka and Oskar Lind. He mentioned his VP Rhea Sandoval has to approve the spend, and that legal \"always takes forever here\". Jonas said \"every release week we lose two engineers for three days to manual test triage\". He asked me for the SOC 2 report and the integration one-pager. I want to keep this simple, just send it to Jonas and he'll handle the internal stuff. Also can you give me the email with bold headers so it looks organized.",
      "expected_output": "A plain-text, forwardable recap with a real send list across the named stakeholders instead of a single recipient, the requested bold headers refused, the two requested documents attached but not listed as action items, and the legal path surfaced as a milestone.",
      "files": [],
      "expectations": [
        "Pushes back on sending only to Jonas and builds a send list across the named buyer stakeholders.",
        "Names Mei Tanaka and Oskar Lind explicitly, either as recipients or with a per-person reason for leaving each off.",
        "Surfaces Rhea Sandoval, the approving VP absent from the call, as a contact to reach rather than leaving her entirely to Jonas.",
        "Names single-threading as the dominant structural failure in deals of this shape, not as a stylistic preference.",
        "Delivers the email as plain text and declines the bold headers, explaining that markup renders inconsistently across email clients and reads as automation.",
        "Names the SOC 2 report and the integration one-pager as attachments in the body without listing either as an action item.",
        "Writes the email to be forwarded internally, on the reasoning that a buying group gives any single vendor only a small share of its total purchase time.",
        "Reproduces Jonas's words verbatim: \"every release week we lose two engineers for three days to manual test triage\".",
        "Surfaces the legal and paper process as a milestone or an open question rather than leaving \"legal always takes forever\" unaddressed.",
        "Recommends a close-plan rung above a merely booked next meeting, given the committee, the $120K size, and legal sitting in the path."
      ]
    }
  ],
  "trigger_queries": [
    { "query": "write the follow-up email from my call with Kestrel this morning", "should_trigger": true },
    { "query": "draft a recap email for yesterday's demo", "should_trigger": true },
    { "query": "turn these call notes into next steps", "should_trigger": true },
    { "query": "post-call summary for the Northwind team", "should_trigger": true },
    { "query": "I need a next steps email after my discovery call", "should_trigger": true },
    { "query": "can you build a mutual action plan for this deal", "should_trigger": true },
    { "query": "what do I send the prospect after a sales call", "should_trigger": true },
    { "query": "help me write up what we agreed on the call", "should_trigger": true },
    { "query": "the call ended with no next step, what do I send", "should_trigger": true },
    { "query": "summarize this sales call into an email the buyer will actually reply to", "should_trigger": true },
    { "query": "mutual close plan for a committee deal", "should_trigger": true },
    { "query": "my prospect said they'd circle back, how do I write the follow-up", "should_trigger": true },
    { "query": "here are my notes from the demo, give me the follow-up", "should_trigger": true },
    { "query": "joint execution plan with the buyer through to go-live", "should_trigger": true },
    { "query": "email after the listing presentation this morning", "should_trigger": true },
    { "query": "I met the Calloways today about selling their house, what do I send them tonight", "should_trigger": true },
    { "query": "recap email with owners and dates", "should_trigger": true },
    { "query": "need a follow-up that actually gets them to do something", "should_trigger": true },
    { "query": "how do I write a next steps email that doesn't get ignored", "should_trigger": true },
    { "query": "MAP template for a $200K opportunity", "should_trigger": true },
    { "query": "we agreed on a security review, put that in an email to the buyer", "should_trigger": true },
    { "query": "draft the post-demo follow-up and include the close plan", "should_trigger": true },
    { "query": "write up the meeting so procurement and legal know what's coming", "should_trigger": true },
    { "query": "what should the subject line be on my post-call follow-up", "should_trigger": true },
    { "query": "turn this transcript into an email with action items for the customer", "should_trigger": true },
    { "query": "my AEs never send follow-ups with dates in them, give me a format", "should_trigger": true },
    { "query": "I want the buyer to confirm what we agreed, what do I send", "should_trigger": true },
    { "query": "next steps email, they want to be live before their audit", "should_trigger": true },
    { "query": "sales follow-up email after a pricing conversation", "should_trigger": true },
    { "query": "how much of a close plan does this deal need before I send the follow-up", "should_trigger": true },
    { "query": "the buyer asked me to send over what we discussed", "should_trigger": true },
    { "query": "I have 40 of these calls a week and my follow-ups are all generic", "should_trigger": true },
    { "query": "action items email for the client after our meeting", "should_trigger": true },
    { "query": "should I send the mutual action plan now or later in the deal", "should_trigger": true },
    { "query": "draft a follow-up my champion can forward internally", "should_trigger": true },
    { "query": "we booked the technical review, write the email confirming it", "should_trigger": true },
    { "query": "wrap up email for the consultation I did this afternoon", "should_trigger": true },
    { "query": "score this call transcript against a rubric", "should_trigger": false },
    { "query": "how did my discovery go on this call", "should_trigger": false },
    { "query": "rate my rep's objection handling on this recording", "should_trigger": false },
    { "query": "who's the real champion in this deal", "should_trigger": false },
    { "query": "map the buying committee at Halcyon", "should_trigger": false },
    { "query": "is this deal single-threaded, check the notes for red flags", "should_trigger": false },
    { "query": "what qualification risks are hiding in this opportunity", "should_trigger": false },
    { "query": "score this deal on MEDDPICC", "should_trigger": false },
    { "query": "grade the decision process and paper process evidence on this account", "should_trigger": false },
    { "query": "build the ROI case for this deal", "should_trigger": false },
    { "query": "payback period on a $120K purchase", "should_trigger": false },
    { "query": "plan my concessions before the pricing negotiation", "should_trigger": false },
    { "query": "what can I trade instead of discounting", "should_trigger": false },
    { "query": "write me subject line variants for my cold email", "should_trigger": false },
    { "query": "why are my emails landing in spam", "should_trigger": false },
    { "query": "set up SPF and DKIM on my sending domain", "should_trigger": false },
    { "query": "what do I say in the first 10 seconds of a cold call", "should_trigger": false },
    { "query": "build my outbound cadence, how many touches and what days", "should_trigger": false },
    { "query": "write the day 3 follow-up email in my prospecting sequence", "should_trigger": false },
    { "query": "find a personalization angle from this funding round", "should_trigger": false },
    { "query": "write my discovery question list for tomorrow's call", "should_trigger": false },
    { "query": "they said we're too expensive, what do I say back", "should_trigger": false },
    { "query": "define our ICP from closed-won data", "should_trigger": false },
    { "query": "tier our accounts and set touch cadences per tier", "should_trigger": false },
    { "query": "what fit signals should the account score use", "should_trigger": false },
    { "query": "how much pipeline do I need to hit quota", "should_trigger": false },
    { "query": "set next year's quotas with ramp adjustment", "should_trigger": false },
    { "query": "design the accelerator curve on the AE comp plan", "should_trigger": false },
    { "query": "should we go product-led or sales-led", "should_trigger": false },
    { "query": "restructure the sales team into pods", "should_trigger": false },
    { "query": "design the interview loop for hiring an SDR", "should_trigger": false },
    { "query": "am I ready to move from SDR to AE", "should_trigger": false },
    { "query": "which sales podcasts and newsletters should I follow", "should_trigger": false },
    { "query": "route this sales request to the right skill in the collection", "should_trigger": false },
    { "query": "write meeting minutes from this morning's standup", "should_trigger": false },
    { "query": "summarize this zoom recording into notes for the team", "should_trigger": false },
    { "query": "build the handoff doc for customer success now that the deal closed", "should_trigger": false }
  ]
}

# Mutual close plan reference

A mutual close plan - also called a mutual action plan (MAP), mutual success plan, or joint execution plan - is a shared document between seller and buyer that maps the path from today through signature to go-live and first measured value. It is co-owned by the rep and a buyer project lead; a plan only the seller maintains is an internal close plan wearing a costume.

## When to introduce

Whether to build a plan at all is the rung ladder in SKILL.md; this file is the rung-3 detail. Once at rung 3, only the timing question is left.

- Introduce early - around the time the opportunity is created, not at negotiation. A late plan reads as a closing device.
- Evidence honesty: the widely circulated win-rate uplifts attributed to MAPs ("26%", "57-200%") are vendor self-reports with no disclosed sample, methodology, or control, and obvious selection bias - buyers willing to co-build a plan were already more engaged. Quote them, if at all, as directional vendor claims.
- The defensible value of the plan is coordination:
  - Dated, owned milestones.
  - A surfaced paper process.
  - Multithreaded visibility.

## Columns

| Column                  | Rule                                                                               |
| ----------------------- | ---------------------------------------------------------------------------------- |
| Milestone / deliverable | Concrete and verifiable ("security questionnaire returned"), never "make progress" |
| Buyer owner             | One named person, never a team                                                     |
| Seller owner            | One named person; seller rows are what keep the plan mutual                        |
| Due date                | A calendar date, worked backwards from the buyer's target                          |
| Status                  | On track / at risk / blocked / done                                                |
| Notes / dependencies    | What this milestone waits on                                                       |

Ranked by value per unit of upkeep:

- efficiency: milestone == buyer owner == seller owner == due date > status > notes/dependencies. The first four tie because they are not separable - a row missing any one of them is not a lighter row, it is not a row at all ("What needs to happen / When it needs to happen / Who is responsible", Salesforce Salesblazer).
- effort: notes/dependencies (thinking per row, rewritten whenever the sequencing shifts) > status (one update per call, for the life of the deal - a standing job) > the irreducible four (written once, with the row).

Status and dependencies are practitioner extensions, not requirements. Cut both on a plan under roughly six rows, where the sequencing is visible at a glance and a status column is only a second place to go stale. Keep status once the plan spans more than a couple of calls: an out-of-date plan is worse than no plan.

## Milestone taxonomy

Work backwards from the buyer's compelling event or go-live date - never the seller's quarter-end - and group by phase:

- **Evaluation**: goals and requirements documented; demo to the full committee; technical validation or sandbox; success criteria agreed in writing.
- **Decision Process** (MEDDICC, Andy Whyte - the buyer's own steps to a decision, spanning technical validation and business approval): economic-buyer review; internal debrief; vendor selection.
- **Paper Process** (MEDDICC - "the series of steps that follow the Decision Process, detailing how you will go from decision to signature"): security review and questionnaires, DPA, MSA redlines and legal, procurement onboarding, signature. Surfacing these early is what stops a won evaluation from stalling for weeks in a contracts queue.
- **Post-signature**: implementation kickoff, onboarding, go-live, first measured value or ROI review. Extending past signature is what keeps the plan from reading as a device to extract a contract.

Fill order when the plan is half-built - efficiency: Paper Process > Decision Process > Evaluation > Post-signature.

- Paper Process: costs one question to the champion, prevents the multi-week contracts stall.
- Decision Process: costs a conversation, buys the buyer's real path.
- Evaluation: rows mostly write themselves from the call notes, so they add little the recap did not already carry.

The order starves the post-signature rows: they are cheap but buy trust rather than pace, so they lose every ratio round - promote them the moment the buyer reads the plan as a closing device.

## The Salesforce 5-part structure

Attributed to Salesforce's Salesblazer mutual action plan template:

1. **Objective or value statement** at the top: a two-way document helping the buyer reach their stated goal by their target date - the buyer's goal, not the seller's.
2. **Buying committee and responsibilities**: list known and unknown roles up front, so every gap becomes an explicit question for the champion instead of a silent hole.
3. **Key dates from both sides**, anchored to a buyer compelling event - an audit deadline, a contract expiry, a launch - never the seller's fiscal calendar.
4. **Deliverables and action items**, including the seller's own, to reinforce that the document is mutual.
5. **Projected outcomes and ROI past signature**: onboarding timeline and expected value at 6 and 12 months.

## Buyer co-ownership

- Never send a fully built plan cold. Propose 2-3 milestones drawn from what was actually agreed on the call, then ask for a working session to build the rest together.
- Get explicit sign-off once the plan is complete - a plan the buyer never approved is a seller wishlist.
- Open and close every subsequent call on the plan: what happens next, how the deal is pacing to the target date, what must be done before the next call.
- Keep seller-owned rows visible so the plan never reads as buyer homework.
- Expect resistance - buyers see "more work". Keep it to one page, not a monster spreadsheet, so the champion can forward it internally without apology.
- Agree where the plan lives (a shared document, a workspace the buyer can edit, mirrored into the CRM) and how often it is updated.

## Worked micro-example

- Vendor: a data-quality company selling to a logistics company.
- Buyer's compelling event: a Q2 audit deadline.
- Committee: VP Ops (champion), IT lead, procurement, COO (economic buyer).

| Milestone                                    | Buyer owner         | Seller owner | Due    | Status      |
| -------------------------------------------- | ------------------- | ------------ | ------ | ----------- |
| Security questionnaire + SOC 2 returned      | Marcus (IT)         | Nora         | Mar 12 | on track    |
| Technical review with IT team                | Marcus (IT)         | Nora         | Mar 17 | on track    |
| Procurement onboarding started               | Priya (Procurement) | Nora         | Mar 24 | not started |
| Economic-buyer review with COO               | Dana (VP Ops)       | Nora         | Mar 31 | not started |
| Signature                                    | Dana (VP Ops)       | Nora         | Apr 15 | not started |
| Go-live: re-keying removed from billing flow | Marcus (IT)         | Nora         | May 15 | not started |

How it is introduced inside the recap email, in one sentence: "Based on today, I've drafted the first three milestones toward your Q2 audit deadline - can we take 20 minutes Tuesday to build out the rest together?"

# Recap email examples

All examples follow the 3-step recap structure (30MPC / Armand Farrokh): next steps first, problems not features, one phone screen. Bodies are plain text as they would be sent. Companies and people are fictional.

## Strong example - B2B mid-market, post-demo

Context: Nora (AE at Brightpath, a data-quality platform) just ran a demo for Harbor Logistics. Attendees: Dana Whitfield (VP Ops, champion) and Marcus Lee (IT lead). Deal is roughly $70K ACV with a buying committee; the technical review is booked.

```
Subject: Next steps - Harbor + Brightpath

Hi Dana,

Great walking through the platform with you and Marcus today. Next steps first:

- Nora: send the completed security questionnaire and SOC 2 report - Thursday, March 12
- Dana: confirm Marcus has sandbox credentials, and loop in Priya from procurement - Monday, March 16
- All: technical review with Marcus's team - Tuesday, March 17, 10:00 (invite sent)

Why this matters: you said the ops team "loses about 12 hours a week re-keying shipment data between the TMS and billing", and that those errors triggered the Q2 audit finding. The technical review is where Marcus pressure-tests whether we remove that re-keying entirely.

I've attached the integration one-pager Marcus asked for.

Anything I missed?

Nora
```

Why it passes the gate, in the same order the skill ranks these elements:

- Action items sit at the top. Every one carries Owner + Deliverable + Date, and the buyer's item is a visible line of its own.
- The next meeting has a real date, a time, and a sent calendar invite.
- The pain is quoted verbatim - the buyer's phrasing, the buyer's metric, tied to their compelling event (the audit finding).
- The recap widens the thread: Dana is asked to bring in procurement, and Marcus is CC-worthy - not single-threaded.
- The closing question invites ratification: a one-word reply turns this into an agreed record.
- Subject is 30 characters and next-step-anchored.
- The attachment is named but never listed as an action item.
- Roughly 120 words - one phone screen, plain text throughout, rendering identically in any email client.

## Negative example - what not to do

The same call, done wrong:

```
Subject: Following up on our great conversation today and next steps for the Brightpath evaluation!

Hi Dana,

Thank you so much for taking the time to meet with me today! It was wonderful to connect and learn more about Harbor.

I wanted to recap some of the powerful capabilities we covered: our AI-powered matching engine, 200+ pre-built connectors, real-time validation rules, customizable role-based dashboards, and our award-winning support team. We really believe Brightpath can transform the way your team works.

I'll circle back next week to see where things stand. In the meantime, don't hesitate to reach out if you need anything at all - always happy to jump on a call!

I'm also attaching our full product deck, pricing overview, implementation guide, and customer story library for your review.

Looking forward to working together!
```

Defect by defect:

- Subject runs past 90 characters and says "next steps" while the email contains none.
- Zero action items. "I'll circle back next week" is a seller-only step with no date, no deliverable, and nothing for the buyer to do.
- No calendar invite and no next meeting - momentum has nowhere to live.
- Feature dump replaces the buyer's stated pain.
- The 12-hours-a-week quote and the audit finding, the actual reasons Dana took the call, never appear.
- Single-threaded: nothing widens the deal to Marcus or procurement.
- Four attachments dumped at once, none of them the one Marcus asked for.
- No closing question, so there is nothing for the buyer to ratify.
- Templated enthusiasm ("wonderful to connect", "transform the way your team works") reads as automation and buries the two questions a recap must answer: what happens next, and why the buyer agreed to it.

## Short-motion example - transactional / B2C

Context: high-ticket B2C - a home solar consultation. Single decision-making household, short cycle. Same rules, resized: 3-5 sentences, one next step with a real date, no mutual close plan.

```
Subject: Your quote + next step

Hi Sam,

Thanks for the walkthrough today. You mentioned the $310 summer electric bills are the main driver - the attached proposal is sized against that usage.

One next step: a 20-minute review call Thursday at 5:30pm to go through the numbers together - invite sent. If Thursday doesn't work, reply with a better slot.

Anything I missed?

Alex
```

Notes:

- The buyer's stated driver appears in their own words.
- The single next step has a date and an invite.
- The fallback ask is a forced choice, not "whenever works".
- A close plan here would be pure friction: the booked call is the whole plan.

---
name: sales-meeting-recap
description: Turns raw sales call notes into a structured recap email with agreed next steps - every action item carrying an owner, a date, and a deliverable - plus mutual close plan (mutual action plan) items when the deal warrants one. Restates the buyer's pain in their own words and never invents a commitment the notes don't contain. Covers complex B2B committee deals and transactional or B2C motions. Use whenever the user mentions a recap email, follow-up after a sales call, next steps email, post-call summary, or a mutual action plan, even without the word recap. Do NOT use for judging how the call went (mbfinotti/sales-skills@sales-call-review).
license: MIT
metadata:
  author: Maya-Beth Finotti
  version: "1.3.4"
---

# Meeting Recap

Convert the user's call notes into two artifacts: a recap email the buyer will ratify by replying, and mutual close plan items when the deal is complex enough to warrant a shared plan. The organizing rule is Armand Farrokh's (30MPC): a recap email answers exactly two questions - "Did you make it clear what the prospect needs to do next?" and "Did you remind them why they agreed to take action?" Everything else is decoration, and "nothing should get in the way of what we want them to do."

Second rule: the recap is champion-enablement content. A buying group gives any single vendor only around 5-6% of its purchase time (Gartner 2017 B2B buyer survey), so the email does most of its selling in rooms the rep never enters - write it to be forwarded.

Scope boundaries:

- Work only from notes, transcripts, or descriptions the user supplies.
- Never score or critique how the call was run - that belongs to `mbfinotti/sales-skills@sales-call-review`.
- The post-sale handoff from sales to customer success is out of scope entirely.

## Interview

Ask one question per message; offer multiple-choice options; skip anything the notes or context already answer.

1. Who was on the call - names and roles on both sides? Which buyer contacts should receive the recap?
2. B2B or B2C? Buying committee, or a single decision-maker?
3. Deal stage: first connect, discovery, demo, proposal, or negotiation?
4. Rough deal size and expected cycle length? This sets the recap's weight and whether a close plan pays off.
5. What was actually agreed on the call - commitments, owners, dates, exactly as stated?
6. Is the next meeting booked with a calendar invite? If not, was a specific date discussed?
7. Does a mutual close plan already exist for this deal, or should this call seed one?
8. By when must the recap land - within the hour, today, or is tomorrow acceptable? A same-hour answer promotes the short recap and deletes anything needing a working session.
9. One-off win or compounding asset: this deal's follow-up only, or a pattern repeated across the pipeline? Compounding promotes the close plan and the automated sequence; one-off promotes the hand-written recap alone.
10. Effort ceiling: how many minutes per recap, across how many live deals? A tight ceiling deletes the full mutual action plan and caps the recap at one screen.

Every ordering below is a default, not a law - it shifts with the deal, with the pipeline, and with who executes it. Re-rank each one against these answers and against what you already know about the user: a rep running dozens of small deals, a single strategic pursuit, and a CRM that already auto-drafts recaps each invert a different ordering. Delete the options their constraints rule out rather than parking them at the bottom, where they quietly reappear as scope.

## Anti-fabrication rule

The load-bearing guard. A recap containing a commitment the buyer never made destroys the shared record it exists to create.

- Never invent a commitment, owner, date, quote, attendee, or decision the notes do not contain.
- Treat a gap as output, not as something to fill: list missing owners, dates, and unconfirmed items in an "Open gaps" block and ask the user.
- Mark any step the buyer did not explicitly agree to as "proposed - confirm before sending".
- Reuse the buyer's own words for the pain, verbatim - their phrasing, their numbers. Never translate their language into marketing copy (practitioner rule: "confirmation, not summary").

## The recap email

30MPC's 3-step recap email (Armand Farrokh):

1. **Lead with next steps** - action items at the very top, each with an owner and a date; make the buyer's own items visually unmissable.
2. **Recap problems, not features** - restate the pain in the prospect's own words, never a list of capabilities.
3. **Keep it short and skimmable** - the whole email fits on one phone screen.

### What earns space on the one screen

One phone screen is the whole budget, so these elements compete for it. Ranked by value per rep minute, best ratio first:

1. **Dated next-steps block at the top** - near-zero extra minutes, part of the recap email already being written. Buys the only thing separating a call from progress: something the buyer must do, by a date.
2. **The buyer's pain in their own words** - near-zero when the notes caught the quote, one more pass through the transcript when they did not. Buys the forward: it is the line that survives the champion's internal debrief.
3. **A send list across the committee** - minutes per deal to find and justify the contacts, redone from scratch on the next deal. Buys the largest measured effect in this section: single-threading is the dominant structural failure, closed-won deals carry 67% more contacts, and multithreading lifts win rates ~130% on $50K+ deals (Gong Labs; sample disclosed, method proprietary).
4. **A closing ratification question** - one line, "Anything I missed?". Buys a reply, and a reply turns the recap into an agreed record.
5. **A subject under 50 characters, next-step-anchored** - one line, plain: "Next steps - Acme + Yourco". Buys the open and nothing past it; 1-4 words performs best (Gong).
6. **The one attachment the buyer asked for** - near-zero to attach, longer when it must be produced first. Buys credibility on a single asked-for item. Never list attachments as action items.

- value: next-steps block > pain in their words > send list > ratification question > subject line > attachment
- effort: send list > attachment > next-steps block == pain quote == ratification question == subject line - those four tie at genuinely near-zero, each being one line or one paste written in the same pass as the email itself

The ratio starves the send list: it is the only row costing real minutes per deal, so a minutes ceiling cuts it first. Promote it regardless of ratio the moment the notes name more than one buyer stakeholder - there is no cheaper substitute for a contact who was never emailed.

Deleted, not demoted. Keep these out of the draft entirely, since they lose on every axis:

- A capability or feature recap (step 2 forbids it).
- Full meeting minutes (a different artifact).
- Rapport prose like "wonderful to connect", which reads as automation.

Unranked on purpose, because they cost nothing and apply to every recap:

- Cap bullets at ~4, max two lines each (30MPC).
- Write plain text that any email client renders as-is - no markdown syntax, headers, or asterisk-bold.

**When to send.** value: within hours of the call > later the same day > next morning > multi-day. Effort is identical at every rung - same email, same minutes - so nothing trades against speed. Send before the buyer's internal debrief, always (timing norm).

Filled strong and negative examples: [references/recap-email-examples.md](references/recap-email-examples.md).

## What makes a next step real

- Real = **Owner + Date + Deliverable**, plus a calendar invite for any meeting. Missing any part means a nice conversation, not progress (practitioner definition).
- Reject "circle back", "next week", "soon", "ASAP" as dates. The forced choice defeats vagueness: "Thursday or Friday?" (technique).
- Assign one named owner per item - a person, never a team.
- Activity is not advancement: extra demos and reference calls without confirmed movement through the buyer's decision process are not progress (MEDDICC).
- When the notes show no committed next step, do not manufacture one. Flag it, and prescribe 30MPC's 5 Minute Drill for the next touch (framework): "Do you wanna buy?", "When do you wanna buy?", "How do you wanna buy?" - commitment, timeline, then the recommended multi-step path.

## Mutual close plan

Three rungs answer "how much plan does this deal need". Ranked by value per rep minute, best ratio first:

1. **A booked next meeting, no plan** - near-zero minutes, and the only rung that scales across a whole pipeline. Buys what most deals actually need: somewhere for momentum to live. Default rung.
2. **Two or three seeded milestones inside the recap** - a few extra minutes on top of the recap, one deal at a time. Buys an early look at the paper process and at least one buyer-owned step, without asking the buyer to co-own a document yet.
3. **A full mutual action plan** - an hour to build, then a standing job to keep current through every call until go-live. Buys coordination a committee deal gets no other way: dated, owned milestones through security review, legal and procurement.

- value: full plan > seeded milestones > booked meeting
- effort: full plan > seeded milestones > booked meeting
- efficiency: booked meeting > seeded milestones > full plan (the axes agree on the first two lines and invert on the third - that inversion is the whole point)

The efficiency order starves the full plan: it is the most expensive rung per deal and loses every ratio round, yet it is the only thing that holds a complex deal together. Promote it regardless of ratio when a buying committee, a multi-call cycle, or procurement or security review sits in the path - roughly $50K+ ACV (threshold reasoning: the multithreading payoff is measured on $50K+ deals). Below that line, delete rungs 2 and 3 from the output rather than offering them: a plan left on the menu for a single-decision-maker deal comes back as scope the rep never had minutes for.

- 30MPC's editorial position, quoted: "If your sales process involves more than a 1-call close...you should be using Mutual Action Plans on every deal." (Treat the rung ladder above as the practical friction-versus-control judgment on top of that stance.)
- Build it as a shared document co-owned by the rep and a buyer project lead; the irreducible columns are milestone, owner (buyer and seller), and date (Salesforce Salesblazer).
- Work backwards from the buyer's compelling event or go-live date - never the seller's quarter-end.
- Extend past signature to onboarding and first measured value, so the plan doesn't read as a device to extract a contract.
- Surface the Decision Process and Paper Process early (MEDDICC / MEDDPICC, Andy Whyte): security review, DPA, MSA redlines, procurement. "60% of deals are lost to inertia" (MEDDICC-published stat).
- Never dump a full plan on a buyer who hasn't agreed to co-own it: enter rung 3 through rung 2, then ask for a working session and sign-off (Salesforce tactic).
- Evidence honesty: the circulated MAP win-rate uplifts ("26%", "57-200%") are vendor self-reports with no disclosed methodology and obvious selection bias. Directional at best - never present them as fact.

Column layout, milestone taxonomy, the Salesforce 5-part structure, and co-ownership tactics: [references/mutual-close-plan.md](references/mutual-close-plan.md).

## B2B and B2C

Identical in both, explicitly:

- Same-day timing.
- Owner + Date + Deliverable.
- The anti-fabrication rule.
- Buyer's-own-words pain.
- Plain-text formatting.
- The humanizer pass.
- The Quality gate.

Consumer-sales follow-up practice (real estate, med spas, automotive, gyms, financial advisory) converges independently on the same two disciplines this skill applies to B2B: reply within about two hours of the call, and personalize to what the buyer actually said rather than a generic template - the same-day timing and buyer's-own-words rules above are not a B2B convention stretched onto B2C, they hold natively in both.

Four recap weights, listed lightest to heaviest. The ranking is the pair of efficiency lines below, not this list order - which weight wins per rep minute inverts with who is executing:

1. **Automated lifecycle sequence** - a week to build once, then near-zero minutes per deal, and it is the only option that covers a whole pipeline at once. Buys coverage, never the buyer's own words.
2. **Short recap** - minutes per deal, one deal at a time. Three to five sentences, one next step with a real date, no close plan. Fits transactional, self-serve, single-decision-maker, and B2C motions including high-ticket consumer (solar, real estate, financial advisory).
3. **Full recap to multiple contacts** - tens of minutes per deal. Buys the multithreaded, forwardable record a committee deal runs on.
4. **Full recap plus mutual close plan** - an hour, then a standing job. Buys coordination through the buyer's paper process.

- value per deal: full recap + plan > full recap > short recap > automated sequence
- effort per deal: full recap + plan > full recap > short recap > automated sequence (the sequence is last only because its cost is paid once, up front, not per deal)
- efficiency, a rep running dozens of small deals: automated sequence > short recap > full recap, with the plan deleted outright
- efficiency, a single strategic pursuit: full recap + plan > full recap > short recap, with the sequence deleted outright

Default weight:

- B2B committee deal: the full recap, promoted to full recap plus plan under the close-plan conditions above.
- Transactional, self-serve, and B2C motions: the short recap.

Reserve the human recap for deals worth the minutes - in high-velocity motions it is often replaced by automated lifecycle sequences entirely (carve-out for transactional sales).

## Workflow

1. Run the Interview; skip answered questions.
2. Parse the notes: agreed commitments with owner/date/deliverable as stated, the buyer's pain in their words, decisions, open questions, and stakeholders mentioned but absent from the call.
3. Apply the Anti-fabrication rule: separate agreed from proposed; collect every gap.
4. Pick the recap weight from B2B and B2C, then draft with the 3-step structure, spending the one screen in the ranked order above.
5. Pick the close-plan rung. At rung 2 or 3, and only if the buyer signalled willingness to co-own it, draft or update milestones per [references/mutual-close-plan.md](references/mutual-close-plan.md); at rung 1, state in the output which rung you chose and why.
6. If your harness can create calendar events or send email, offer to book the next meeting and place the draft - never send without explicit approval. Otherwise deliver text the user can paste.
7. Optional integration note: if your environment connects to a CRM, a call-recording platform, or an AI notetaker, offer to pull the transcript as input and log the recap as an activity - pulled content is still just notes, under the same anti-fabrication rules.
8. Run the draft through your preferred humanizer skill. Raw first-draft output is never final.
9. Run the Quality gate; redraft until every check passes.
10. Deliver the output shape below. If your harness has persistent memory, store the deal's open action items and plan state so the next recap starts from live state instead of a blank page.

## Invocation and expected output

- "Here are my notes from the Acme demo - draft the recap email."
- "Turn this discovery call into a follow-up with next steps and start a mutual action plan."
- "Quick post-call summary for a small deal, single contact."

```
SUBJECT     : under 50 characters, next-step-anchored
RECAP EMAIL : plain text; next steps first (Owner - Deliverable - Date,
              buyer items highlighted), pain in the buyer's own words,
              attachments named, closing ratification question
CLOSE PLAN  : chosen rung, plus new or updated milestones (milestone |
              buyer owner | seller owner | date) at rung 2 or 3
OPEN GAPS   : missing owners/dates, unconfirmed commitments marked
              "proposed - confirm before sending", questions for the user
SEND LIST   : who receives the email and why (multithreading check)
```

## Quality gate

Pass threshold: every check, no exceptions. Redraft and re-check until nothing fails. Deliberately unranked: these are pass/fail gates, not options traded against each other, so ordering them would only imply the cheap ones are skippable.

1. 100% of action items carry Owner + Date + Deliverable - zero vague dates ("next week", "soon").
2. The next meeting has a real calendar date and an invite, or the output flags the gap and prescribes the 5 Minute Drill.
3. Email fits one phone screen, under ~200 words (short-motion recap: 3-5 sentences).
4. Subject line under 50 characters.
5. The buyer's stated pain appears verbatim in their own words; no feature list anywhere.
6. Every commitment traces to the notes, or is marked "proposed - confirm before sending"; gaps are listed, never filled.
7. On committee deals, at least one action is buyer-owned - or the seller-only imbalance is flagged as a risk.
8. The send list covers the relevant buyer contacts, or states why a single recipient is right.
9. The draft went through the humanizer pass.

## KPIs

Instrument in this order - value per unit of setup effort, best ratio first:

1. **Next-step-set rate**: % of open opportunities with a scheduled, calendar-booked next step. A count the CRM already holds, and the only metric here with a pass mark - practitioner benchmark ~80%. Below that, fix this before any close-plan rollout (recommendation). Start here.
2. **Multithreading rate**: buyer contacts engaged per deal. Also already in the CRM (direction solid, magnitudes are marketing - Gong).
3. **Recap reply rate** - per-deal tracking to set up once, then near-zero. A reply ratifies the record; track it per deal stage.
4. **Stage conversion and slipped-deal rate** on deals with vs. without a dated next step - one pipeline report answers both.
5. **Sales-cycle length** - a full quarter of data before it says anything. Context for the stakes: 40-60% of qualified B2B deals end in no decision (Dixon and McKenna, The JOLT Effect, 2.5M conversations) - next-step discipline attacks exactly that.

- value: next-step-set rate > sales-cycle length > stage conversion == slipped-deal rate > multithreading rate > recap reply rate (those two tie because they are the same pipeline query over the same window, and neither moves before the other)
- effort: sales-cycle length > stage conversion == slipped-deal rate > recap reply rate > next-step-set rate == multithreading rate (this pair ties at near-zero: both are counts the CRM already stores, read from the same report)

The order starves sales-cycle length - a quarter of waiting kills its ratio every round. Promote it anyway when the question is whether to roll recap discipline out across a team; nothing cheaper produces that evidence. And never measure close-plan success by vendor win-rate uplift claims: instrument your own before/after comparison.

## Failure modes

Rows are defect-and-fix pairs, not competing options - each fix is the only fix, so there is nothing to rank.

| Failure                                        | Fix                                                                        |
| ---------------------------------------------- | -------------------------------------------------------------------------- |
| "Circle back next week" as a next step         | Owner + Date + Deliverable, or list it under Open gaps                     |
| Every action owned by the seller               | Surface buyer-owned steps from the notes; flag the imbalance if none exist |
| Feature dump instead of the buyer's pain       | Their problem, their words, verbatim                                       |
| No calendar invite for the next meeting        | Real date plus invite, or flag and prescribe the 5 Minute Drill            |
| Recap sent to one contact on a committee deal  | Build the send list; single-threading is the top structural failure        |
| Commitments the buyer never made               | Anti-fabrication rule; mark "proposed - confirm before sending"            |
| Full close plan dumped on an unconsulted buyer | 2-3 agreed milestones, then a working session and sign-off                 |
| Recap sent the next morning                    | Same day, within hours, before their internal debrief                      |
| Markdown formatting in the email body          | Plain text only                                                            |
| A recap the length of meeting minutes          | One phone screen; minutes are a different artifact                         |

## References

- [references/recap-email-examples.md](references/recap-email-examples.md) - a strong filled example, a negative example annotated defect by defect, and a short-motion B2C variant.
- [references/mutual-close-plan.md](references/mutual-close-plan.md) - columns, milestone taxonomy, the Salesforce 5-part structure, when to introduce, buyer co-ownership tactics, and a worked micro-example.
- `mbfinotti/sales-skills@meddpicc-scorecard` - qualifies the deal; this skill's close plan operationalizes its Decision Process and Paper Process elements.
- `mbfinotti/sales-skills@deal-champion-mapping` - identifies who the recap should be written through and who belongs on the send list.
- `mbfinotti/sales-skills@deal-red-flags` - a missing dated next step or a seller-only plan surfaced here is an input signal there.
- `mbfinotti/sales-skills@deal-value-calc` - builds the forwardable business case; the recap carries its numbers into the buyer's internal debrief.