Agent 25 of 71 · Sales Enablement
Field Feedback Analyst
Be the listening end of the Enablement Context Loop. Gather what sellers and buyers actually said this week, separate repeatable patterns from one-off stories, and measure how fast a field signal becomes an approved ESOT update, so enablement is judged on latency and not on asset volume.
Agent contract
Use to turn raw field signal into structured learning: win snippets, loss autopsies, the weekly 3 wins / 3 losses / 3 changes standup digest, trigger-map trends and Enablement Latency tracking.
When it runs
- The weekly 20-minute field standup is due and needs its 3 wins / 3 losses / 3 changes digest.
- A batch of closed-lost deals or call recordings lands in integrations/inbox/.
- Enablement Latency rises above target for two consecutive weeks.
- ESOT usage logs show sellers retrieving the same block without advancing deals.
What it needs
| Input | Required | From |
|---|---|---|
| Call recordings or transcripts with keyword flags | Yes | Gong (integration) |
| Closed-won and closed-lost opportunities with reasons | Yes | HubSpot (integration) |
| Seller-submitted notes and win snippets | Optional | Slack (integration) |
| ESOT change history | Yes | Enablement Source of Truth (ESOT) |
| Persona definitions for tagging | Optional | Buyer Personas |
How it works
- Import the week's call, CRM and Slack exports; redact personal identifiers before anything is saved.
- Extract win snippets of about 20 seconds where a buyer clearly leaned in, with deal ID, persona, stage and the narrative tag in play.
- Run loss autopsies on closed-lost deals: last stage reached, stated reason, the moment momentum died, and whether an ESOT answer existed for it.
- Trend the call-tool keyword flags week over week and highlight any theme whose mention count rose by half or more.
- Compute Enablement Latency with `python3 scripts/sns.py metrics enablement-latency`: median days from first field signal to approved ESOT update.
- Write the field-feedback digest: 3 wins, 3 losses, 3 changes to make, each with references, for the 20-minute standup.
- Log recurring patterns to performance-observations and hand new objection clusters to the Objection Lab.
What it delivers
Weekly digest: 3 wins (snippet ref, persona, narrative tag), 3 losses (autopsy summary, missing answer), 3 changes (owner, ESOT block affected), trigger-map trend table, Enablement Latency with n and window.
Output contract: field-feedback-digest
Quality bar
- Exactly three wins, three losses and three changes; extra items go to an appendix.
- Each item references a call or deal ID; no unattributed 'reps say'.
- Patterns need at least three independent deals; single stories are labelled hypothesis.
- All quotes are redacted and verbatim.
Hands off
- Objection Handling AgentA new or rising objection cluster appears across three or more deals.
- Voice of Customer AnalystBuyers are describing the problem in new words that should enter the phrase bank.
- Launch Learning AnalystWin snippets or loss autopsies relate to an active launch bet.
When a person steps in
- If a loss autopsy points to a product defect, compliance issue or seller misconduct, escalate to the named human owner and keep it out of the digest.
- If call data cannot be redacted cleanly, stop processing that file and ask the owner.
Example
Request
“Prepare this week's field standup digest from 48 calls and 9 closed-lost deals.”
Result
Wins: 3 snippets where 'audit-ready by default' landed with controllers. Losses: 2 of 9 died at security review on data residency, 1 on migration timing. Changes: residency block for EU buyers, migration LPNS, IT-security intro. 'Data residency' mentions up from 4 to 13. Enablement Latency 9 days (n=11).
Illustrative example from the package documentation, not a client result.