Post Meeting Execution

Post-meeting execution

Template placeholders

Replace every {{...}} before enabling. See the setup checklist reference for the full setup list.

• {{MEETING_OWNER}} — Person whose meetings this skill processes (one instance per person)
• {{MEETING_NOTES_TOOL}} — Your meeting-notes tool (e.g. Circleback, Fireflies, Gong)
• {{CRM}} — Your CRM (e.g. HubSpot, Attio, Salesforce)
• {{REVIEW_CHANNEL}} — Slack channel where all post-meeting outputs post for review
• {{INTERNAL_DOMAIN}} — Your company email domain, to tell internal from external attendees
• {{FOLLOWUP_SCENARIOS_SKILL}} — Sub-skill: meeting-outcome categories, classification logic, per-category email frameworks
• {{VOICE_GUIDE_SKILL}} — Sub-skill: the owner's tone, formatting, greeting style, approved links
• {{LEAD_SCORING_SKILL}} — Sub-skill: your lead scoring & qualification methodology
• {{ENTERPRISE_ACV_THRESHOLD}} — Deal size above which an account gets full sales treatment vs. self-serve (default: $20K)
• {{SALES_ACTIVE_TAG}} — Workspace tag marking accounts with a live sales conversation
• {{METRICS_LOG}} — Optional: spreadsheet/event log where external meetings are recorded for GTM metrics

---

Overview

After every meeting involving {{MEETING_OWNER}}, run a structured workflow in two phases:

1. Deal creation decision — qualification scoring, human confirmation, contact/stakeholder loading
2. Follow-up email — classify the outcome, draft in the owner's voice, send for approval

Non-blocking errors: memory, enrichment, and {{CRM}} write failures are non-blocking — log the error in the review item and continue; a failed step is never a reason to skip subsequent steps. The only blocking failure is a missing meeting record (Step 1).

---

Step 1 — Pull Meeting Context

Retrieve the full record from {{MEETING_NOTES_TOOL}}: summary, full transcript, action items, attendees (names, titles, companies), any logged decisions. If your notes tool auto-emails attendees the summary and recording, do NOT duplicate that in the follow-up email.

---

Step 2 — Enrich the Account

Look up the company in your workspace and {{CRM}}: current funnel stage, size (headcount, revenue), industry, prior relationship history, and any existing deal.

Then estimate ACV from company size, use cases surfaced in the transcript, and any pricing signals. This gates the {{ENTERPRISE_ACV_THRESHOLD}} line between self-serve and enterprise treatment. If ACV cannot be estimated with reasonable confidence, flag it in the review item.

Tagging: if at least one attendee's email domain is not {{INTERNAL_DOMAIN}}, apply {{SALES_ACTIVE_TAG}} to the company (idempotent). Internal-only meetings skip this and every external-meeting step below.

---

Step 3 — Log the Meeting for Metrics (optional)

Skip if internal-only, or if you don't maintain {{METRICS_LOG}}.

Classify the meeting type. Never short-circuit on deal existence — deals are often auto-created the moment a demo is booked, so an open deal is not evidence of a prior meeting.

1. Closed-won check: account is a current customer → meeting_customer. Stop.
2. Check history sources in parallel: (a) account memory — is a first-demo date recorded? If yes → meeting_active_deal, stop. (b) {{CRM}} company-level meeting engagements within the last 6 months. (c) {{CRM}} meeting engagements on any contact at the company, same 6-month filter. Older engagements are stale ghost history from a prior cycle — ignore them.
3. If any source confirms a qualifying prior meeting → meeting_active_deal. Otherwise → meeting_first_touch, even if an open deal exists.

Same-day deal rule: If an open deal's create date falls on the same calendar day as this meeting, treat as meeting_first_touch — the deal was almost certainly auto-created by the booking, not evidence of a relationship. The recency filter and this rule together prevent the known false-meeting_active_deal failure modes.

Write one row to {{METRICS_LOG}}: unique event ID ([type]-[domain]-[date], suffix -2, -3 for multiple same-day meetings), event type, domain, company, meeting date, external attendee names.

⚠️ Read-after-write verification required. After every log write, read the row back by its event ID. Not found → retry once. Still not found → add a visible warning to the review item ("Log write failed — manual entry required"). Never silently accept a false success response.

First-demo detection (only when meeting_first_touch): If no first-demo date exists in account memory or the log, record one — using the date the meeting was booked (created in {{CRM}}), not the date it occurred; this metric tracks booking velocity. Never overwrite an existing first-demo date. Write the date to account memory.

---

Step 4 — Update Account Memory

Runs for every meeting; depth varies by call type. All calls: key topics and tone, commitments from either side, stakeholder dynamics (champions, skeptics, blockers), anything that informs the next interaction.

Scoping calls additionally capture these structured fields — omit any field with no evidence from this meeting; never fabricate:

• Pain & Friction — the exact moment where their current workflow breaks down, not generic pain
• Use Cases / Workflows to Test — each workflow they agreed to evaluate, named with its expected output (e.g. "visitor identification → auto-enrich + {{CRM}} push", not "pipeline generation")
• Trial Participants — names, titles, roles in the evaluation
• Success Criteria — the specific outcomes that would make them say "this worked"

If your product's onboarding can consume this context (to pre-populate a trial account), push the structured block there whenever the account is still pre-trial. The more specific, the faster the prospect's time to value.

---

Step 5 — Rescore the Account

Rescore against {{LEAD_SCORING_SKILL}} using signals from this specific meeting: new decision-makers joining, stated intent to trial or buy, a use case / budget / timeline named for the first time, enthusiasm shifting either way. If the tier changed, update the score tag (remove old, apply new) and log the reasoning. If unchanged, note that explicitly.

---

Step 6 — Recommend a Funnel Stage (do not apply yet)

Determine the correct stage from the updated score plus meeting context — but do not update it; the move requires the owner's approval in Step 10. Use stage semantics as a guide; the transcript beats mechanical rules:

⚠️ If your trial stage is set automatically by a product signup event, never set it from a meeting outcome — even when the call ends with "let's start the trial." Keep the account in evaluation until the product event fires, so the funnel never claims trials that didn't activate.

---

Step 7 — Qualification Scoring (BANT)

Skip entirely if an existing deal was found in Step 2 — already qualified. Go to Step 8.

Score each criterion met / not met on transcript evidence only:

• Budget — actual budget confirmed or referenced, not mere willingness to pay
• Authority — a purchasing decision-maker in the room (startups: VP/C-level required; larger orgs: Director may qualify, VP+ preferred)
• Need — BOTH clearly defined use cases driving fast time-to-value AND a stated desire to start a trial
• Timeline — a stated timeline or compelling event forcing speed

Count criteria met (0–4), record evidence for each.

Override — exception criteria. Even below 2/4, flag for deal creation when all three hold: (1) C-suite or VP-level authority confirmed in the room; (2) exceptional use-case alignment — their stated problems map directly onto your core value proposition, not generic interest; (3) structural growth pressure — aggressive targets, recent funding, team expansion, or a high-stakes operational gap (e.g. a one-person ops team scaling a motion that needs automation). Flag the override explicitly: name the unmet criteria and the compensating signals. The owner makes the exception call — never auto-create.

Deal creation decision (human in loop):

• 2+ criteria met (or override): create a review item — not a plain message; the owner must be able to reply and act from Slack — with the scorecard and evidence, estimated ACV, and: "Based on [X/4], I recommend creating a deal. Agree? Yes / No". Post to {{REVIEW_CHANNEL}}. Mandatory — never continue past it silently.
• Fewer than 2, no override: do not suggest a deal. Note the score and reasoning inside the email review item (Step 10).

If the owner says Yes: re-check {{CRM}} for an existing open deal — if one exists, report it (ID + stage), do NOT create a duplicate. Otherwise ask the owner for the amount, then create the deal: name = company name only, stage matched to funnel position, close date estimated from timeline signals, owner = {{MEETING_OWNER}}. Associate it to the company, and mirror ownership and stage to your workspace automatically — the owner should not have to ask.

---