How I built a RevOps Intelligence skill for Claude and what it does
Table of Contents:
RevOps lives and dies by the quality of its analysis. When you walk into a forecast call, you want numbers you can defend, not a best-guess built on rep updates and CRM fields nobody's touched in two weeks.
Building that analysis takes time, though. Cross-referencing call data against CRM stages, checking email engagement on Commit deals, flagging what's actually at risk, it's the right work, but it's slow enough that it gets cut short or skipped entirely.
I built the RevOps Intelligence skill to automate it. It connects Avoma and your CRM, runs the cross-reference, and returns a report you can stand behind: named deals, named signals, prioritized next steps. Download the skill.
Most of this analysis can be done directly in Avoma. The skill simply brings those workflows into Claude for teams that prefer to run analysis and decision-making from a single workspace.
The problem with pipeline inspection today
CRM stages reflect what reps believe about deals, not what buyers have said.
A deal moves to Verbal Commit because a rep felt the call went well. Another sits in Demo Completed for 44 days because nobody flagged it. A third shows a close date of last Thursday and is still marked open. None of these surface in a standard pipeline review unless a manager happens to look at the right deal at the right time.
The frustrating part is that most teams already have Avoma recordings and tracked emails that would catch all of it. The last call transcript for the Verbal Commit deal might show an unresolved pricing objection. The Demo Completed deal's last email went unanswered three weeks ago. The evidence exists. The workflow to use it doesn't.
That's what I set out to build.
What the RevOps Intelligence skill does
The skill runs structured analysis across Avoma (call transcripts, tracked emails, AI notes, meeting data) and your CRM (HubSpot or Salesforce). For each task, it pulls data from available sources, applies a specific analysis framework, and delivers a structured output with a summary and prioritized recommendations.
Every output names the deal, the owner, the signal, and the recommended action with a timeframe. "Some deals may be at risk" is not useful. This is built to produce the specific.
What it covers
12 analyses across two clusters.
Pipeline and forecast:
1. Pipeline inspection pulls all open deals, flags dark deals (no Avoma activity in 14+ days and no CRM activity in 7+ days), surfaces close dates that have passed, identifies stage concentration risk, and cross-references CRM stage against actual call evidence.
2. Forecast validation checks every Commit and Best Case deal for supporting evidence: was there a call in the last 14 days, did the buyer confirm a next step, are there unresolved blockers in the transcript? It separates supported forecast from at-risk forecast with dollar amounts.
3. Deal risk scores each open deal on 5 signals: dark deal, competitor mentioned, unresolved objection, stakeholder drop-off, and close date slip. Deals scoring 3 or more get flagged high risk.
4. CRM hygiene runs 6 checks per deal: close date set, amount populated, contact linked, next step field populated, stage updated in the last 30 days, and Avoma action items reflected in CRM. Scores overall hygiene and surfaces the worst offenders.
5. Next-step compliance checks whether next steps agreed on Avoma calls made it into CRM, whether next step fields are blank or stale, and whether calls ended without a confirmed follow-up. Produces a rep-level compliance rate.
6. Qualification audit scores active deals against MEDDICC, MEDDPICC, or BANT by pulling evidence from call transcripts. The most commonly missing component across your pipeline comes out in the summary.
7. Stage validation checks whether deals are in the right CRM stage based on what actually happened in calls and emails. Flags where evidence doesn't match the stage and estimates how much pipeline may be overstated.
8. Win/loss analysis looks across all closed deals in a period to surface what correlates with wins and losses: champion presence, economic buyer engagement, how competitors were handled, where deals died in the stage sequence.
CS and customer
9. Churn risk scores each account on 6 signals (renewal timing, sentiment decline, champion departure, executive disengagement, email silence, and support spikes) to flag at-risk accounts before renewal.
10. Expansion discovery surfaces accounts where call transcripts show unmet use cases, feature requests, or growth signals.
11. Customer health produces a per-account scorecard across 5 dimensions: engagement, sentiment, executive sponsorship, action item follow-through, and renewal signal.
12. Executive report synthesizes pipeline health, forecast confidence, key risks, and top opportunities into a single output built for a leadership review or QBR insert.
What the output looks like
Here's an example from a pipeline inspection:
Pipeline Inspection — June 12, 2026
Sources: HubSpot · Avoma
100 open deals · $6.5M pipeline · $65K avg deal size
Summary
- Stage concentration: Proposal/Evaluation holds $3.3M across 27 deals — 51% of pipeline in a single stage
- Renewals: 53 deals, $2.3M — majority of deal count, smaller individual sizes
- Avoma coverage: 99 meetings completed this week across demos, trial onboardings, and renewals
- 26 deals have past close dates — single largest pipeline health issue
- 3 deals close within 7 days totaling $215K — need immediate attention
- 6 deals have no documented next step in CRM
- 6 of 8 deal owners are marked inactive in HubSpot
CRITICAL — Closes within 7 days ($215K total)
| Deal | Amount | Close Date | Owner |
|---|---|---|---|
| Acme Corp – Renewal | $148K | Jun 13 | Rep A |
| Initech – 70x Org | $24K | Jun 15 | Rep B |
| Globex – 2026 Renewal | $42K | Jun 18 | Rep A |
CRITICAL — 26 deals with past close dates ($1.6M in overdue pipeline)
| Deal | Amount | Close Date |
|---|---|---|
| Umbrella Corp – Field Sales | $250K | Feb 27 2026 |
| Initech – Sales (200 seats) | $186K | Feb 27 2026 |
| Globex – 500 Seats | $176K | Dec 31 2025 |
| Acme Corp – Revenue Team | $144K | Feb 27 2026 |
| Vandelay Industries – 51-100 | $124K | Feb 27 2026 |
| Sirius Cybernetics – 140 seats | $99K | Feb 26 2026 |
| Dunder Mifflin – Sales (100) | $70K | Feb 26 2026 |
| Bluth Company – 50 seats | $52K | Feb 26 2026 |
| Rekall Inc Renewal | $52K | Apr 30 2026 |
| Cyberdyne – CS (70 seats) | $49K | Feb 27 2026 |
| + 16 more past-close deals | | |
MEDIUM — 6 deals with no next step
| Deal | Amount | Owner |
|---|---|---|
| Acme Corp Renewal | $57K | Rep A |
| Initech Renewal | $35K | Rep A |
| Globex Renewal | $30K | Rep A |
| Vandelay Industries Renewal | $30K | Rep A |
| Sirius Cybernetics Renewal | $29K | Rep C |
| Dunder Mifflin Renewal | $25K | Rep C |
Recommendations
| # | Action | Owner | Why | When |
|---|---|---|---|---|
| 1 | Triage 3 imminent closes | Rep A / Rep B | $215K at risk this week; Helix closes June 13 | Today |
| 2 | Audit 26 past-close deals | RevOps + Rep A | $1.6M of pipeline is unreliable in forecast | This week |
| 3 | Reassign deals from inactive owners | RevOps | 6 owners inactive; deals will go dark without reassignment | This sprint |
| 4 | Fill next steps on 6 blank deals | Rep A (4 of 6) | No next step = no follow-up path | This week |
| 5 | Review Proposal/Evaluation stage | Sales leadership | $3.3M (51% of pipeline) concentrated in one stage | Before next forecast call |
What to watch
Helix Health close · Past-close deal cleanup rate week over week · Inactive owner reassignment progressHow to run the RevOps Intelligence skill
You need the Avoma connector active in Claude. HubSpot or Salesforce and Gmail improve coverage, but the skill notes what's missing and runs on whatever is connected. If you are new to Claude connectors, you can learn more here.
Exact trigger phrases:
Plain language works too. "Where are my deals stalling," "which deals are at risk," "how clean is our CRM." These all trigger the right task.
The only required input is a date range. The skill defaults to completed calls over 5 minutes and runs across all reps unless you say otherwise. Download the skill here.
The design decision that took the longest
The analysis logic wasn't the hard part. Defining what counts as evidence was.
A deal being in Verbal Commit in the CRM means a rep advanced the stage. A deal actually being in Verbal Commit means a buyer said yes on a recorded call, a multi-stakeholder meeting happened, and no unresolved blocker appeared in the last transcript. Getting the skill to verify that distinction, cross-referencing CRM stage against Avoma transcripts and tracked emails, is where most of the work went.
The same problem comes up in forecast validation. An unsupported Commit isn't just a deal with no recent call. It's a deal where the last transcript shows an unresolved pricing objection, tracked email engagement has gone cold, and the CRM next step hasn't been touched in two weeks. Each signal matters on its own. Together, they're the real picture.
The other thing I kept coming back to: specificity. An analysis that says "some deals may need attention" is not useful to anyone. One that says Northline SaaS is in Verbal Commit but CFO approval is still pending, last call was June 1, and $112K is probably overstated: that's something you can act on. The output format was built around that, and it took a few iterations to get the structure right.
Where to go from here
If you're on Avoma, your meeting data is already connected. Download the skill, open Claude, make sure your connectors are active, and type run pipeline inspection.
If you want to understand what Avoma's revenue intelligence looks like at the platform level before going further, this is a good place to start.
Frequently Asked Questions
The RevOps Intelligence Claude skill requires the Avoma connector to analyze meeting transcripts, call recordings, AI notes, and scorecards. HubSpot or Salesforce can add CRM fields such as deal stage, amount, owner, close date, contacts, and next steps. Gmail, Google Calendar, Slack, and Teams can improve coverage when available, but the skill is designed to run with available connected sources and note any missing data.
If HubSpot or Salesforce is not connected, the skill can still use Avoma data where available, but CRM-based checks will be limited. Analyses such as CRM hygiene, close date validation, stage validation, forecast category review, and owner-level pipeline reporting require CRM data to be complete. The output should state which sources were available and explain how missing CRM access affects coverage.
The skill identifies dark deals by checking for lack of recent Avoma and CRM activity. In pipeline inspection, a deal is flagged when there is no Avoma activity in the last 14 days and no CRM activity in the last 7 days. In deal risk scoring, a dark deal signal is triggered when there is no Avoma call and no CRM activity in 14 or more days.
The skill validates Commit and Best Case deals by comparing CRM forecast categories with recent buyer evidence. For Commit deals, it checks for an Avoma call in the last 14 days, a buyer-confirmed next step, and no unresolved blocker in the most recent call transcript. For Best Case deals, it looks for a recent Avoma call and buyer email engagement. Deals without supporting evidence are flagged as forecast risk.
The skill scores open deals using five risk signals: dark deal, competitor mentioned, unresolved objection, stakeholder drop-off, and close date slip. Each signal adds one point. Deals with three or more signals are marked High Risk, while deals with two signals are marked Medium Risk. The output includes deal name, owner, amount, triggered signals, last activity, and total at-risk pipeline value.
Yes. The CRM hygiene task checks whether CRM records contain key fields and whether Avoma action items are reflected in the CRM. The audit reviews items such as close date, amount, linked contacts, next step fields, recent stage updates, and Avoma-to-CRM action item coverage. This helps identify where missing or stale CRM data may reduce forecast accuracy.
Yes. The skill includes customer-focused tasks for churn risk, expansion discovery, customer health, and executive reporting. These tasks use Avoma CS calls, CRM account data, renewal timing, stakeholder engagement, sentiment trends, email activity, and support-related signals where available. The goal is to surface account risk, expansion opportunities, and customer health patterns from connected revenue data.
Users should check that the Avoma connector is active and that CRM access is available if the analysis depends on deal, forecast, or account fields. For broader coverage, HubSpot or Salesforce, Gmail, Google Calendar, and communication tools can be connected when relevant. The skill should still run with partial data, but the output should clearly note unavailable sources and any resulting limitations.
Dive into Our Latest
Avoma Insights.
What's stopping you from turning every conversation into actionable insights?
