Product Insights — Sprint Planning Intelligence

Turn every customer conversation, support interaction, and onboarding session into structured product input. This skill feeds your engineering team with prioritized, evidence-backed work items.

Core Principle

Engineers shouldn't build from gut feel. Every sprint item should trace back to a real customer moment. This skill extracts, categorizes, and prioritizes product signals so sprint planning starts with data, not opinions.

Data Collection (run in parallel)

Source 1: Feature Requests from Calls (Fireflies)

Pull customer-facing calls:

fireflies_get_transcripts: mine=true, fromDate={periodStart}, toDate={periodEnd}, limit=50

Get summaries for each:

fireflies_get_summary: transcriptId={id}

Search for product-specific signals:

fireflies_search: keyword:"can you" scope:sentences from:{periodStart} limit:30
fireflies_search: keyword:"wish" scope:sentences from:{periodStart} limit:30
fireflies_search: keyword:"would be nice" scope:sentences from:{periodStart} limit:30
fireflies_search: keyword:"doesn't work" scope:sentences from:{periodStart} limit:30
fireflies_search: keyword:"broken" scope:sentences from:{periodStart} limit:30
fireflies_search: keyword:"confusing" scope:sentences from:{periodStart} limit:30
fireflies_search: keyword:"slow" scope:sentences from:{periodStart} limit:30
fireflies_search: keyword:"error" scope:sentences from:{periodStart} limit:30
fireflies_search: keyword:"workaround" scope:sentences from:{periodStart} limit:30
fireflies_search: keyword:"manually" scope:sentences from:{periodStart} limit:30

Categorize each signal:

For each signal, capture:

• Customer name and company
• Exact quote (verbatim from call)
• Context (what they were trying to do)
• Frequency (first mention or repeat?)
• Impact (blocks their workflow? Annoyance? Workaround exists?)

Source 2: CRM Notes (HubSpot)

Pull contacts with recent notes:

search_crm_objects: objectType="contacts"
  filterGroups: [{ filters: [
    { propertyName: "notes_last_updated", operator: "GTE", value: "{periodStart}" }
  ]}]
  properties: ["firstname", "lastname", "company", "jobtitle",
               "notes_last_updated", "hs_lead_status"]
  limit: 50

Pull deals with notes:

search_crm_objects: objectType="deals"
  filterGroups: [{ filters: [
    { propertyName: "notes_last_updated", operator: "GTE", value: "{periodStart}" }
  ]}]
  properties: ["dealname", "amount", "dealstage", "notes_last_updated"]
  limit: 30

Extract from deal/contact notes:

• Feature requests mentioned in sales conversations
• Blockers preventing deal close ("they need X before signing")
• Integration requirements ("they use EHR Y, need compatibility")
• Competitive feature gaps ("competitor Z has this, we don't")

Source 3: Onboarding Patterns

Search for onboarding calls specifically:

fireflies_search: keyword:"onboarding" scope:title from:{periodStart} limit:20
fireflies_search: keyword:"setup" scope:title from:{periodStart} limit:20
fireflies_search: keyword:"training" scope:title from:{periodStart} limit:20

Extract:

• Where new users get stuck (first 48 hours patterns)
• Questions asked repeatedly (docs/UX gap)
• Time-to-first-value (how long until they see results)
• Setup steps that require manual intervention (automation opportunities)
• Drop-off points (started but never completed a key action)

Source 4: Competitive Feature Gaps (Web Research)

For key competitors, check recent product updates:

WebSearch: "Waystar" product update features {current year}
WebSearch: "CoverMyMeds" new features {current year}
WebSearch: "Availity" product launch {current year}
WebSearch: "CentralReach" product update {current year}

Also check G2/Capterra for user complaints about competitors:

WebSearch: site:g2.com "prior authorization software" reviews
WebSearch: site:capterra.com "insurance verification" reviews

Extract:

• Features competitors launched (are we behind?)
• Features users wish competitors had (our opportunity)
• Common complaints (our differentiation)

Analysis & Prioritization

RICE Scoring (adapted for early-stage)

Score every product signal using this framework:

RICE Score = (Reach x Impact x Confidence) / Effort

Revenue Impact Multiplier

Boost priority for signals tied to revenue:

• Deal blocker: Item blocking a deal worth $X → 2x multiplier
• Churn risk: Item causing a paying customer to consider leaving → 1.5x multiplier
• Expansion blocker: Item preventing upsell/cross-sell → 1.3x multiplier

Category Buckets

Group scored items into sprint-planning buckets:

Report Structure

# Product Insights Report
## Period: {startDate} — {endDate}
## Sprint: [current/next sprint identifier]

---

## SPRINT RECOMMENDATIONS (top 10, ranked by RICE)

| # | Item | Category | RICE | Customers | Revenue Impact | Effort |
|---|------|----------|------|-----------|---------------|--------|
| 1 | ... | BUG | XX | [names] | $X deal blocked | S |
| 2 | ... | FEATURE | XX | [names] | Churn risk at [company] | M |
| ... |

---

## MUST FIX (Bugs & Breaks)

### Bug 1: [title]
- **Reported by**: [customer, date, call link]
- **Quote**: "[exact words]"
- **Impact**: [what it breaks in their workflow]
- **Frequency**: [one-off or recurring]
- **Suggested fix**: [technical direction if obvious]

### Bug 2: ...

---

## MUST BUILD (Feature Requests)

### Feature 1: [title]
- **Requested by**: [customer list with dates]
- **Customer quotes**: "[verbatim from calls]"
- **Use case**: [what they're trying to accomplish]
- **Current workaround**: [what they do today without it]
- **Revenue impact**: [deals blocked, churn risk, expansion]
- **Competitive context**: [do competitors have this?]
- **Suggested scope**: [MVP definition — what's the smallest useful version]

### Feature 2: ...

---

## SHOULD IMPROVE (UX & Performance)

### Improvement 1: [title]
- **Signal**: [what customers said/did]
- **Current experience**: [what happens today]
- **Ideal experience**: [what should happen]
- **Impact**: [time saved, confusion eliminated, drop-off reduced]

---

## ONBOARDING GAPS

| Step | Friction Point | Frequency | Fix Type |
|------|---------------|-----------|----------|
| ... | ... | X customers | Docs / UX / Automation |

### Time to First Value: [current avg] → Target: [goal]

---

## COMPETITIVE FEATURE MAP

| Feature | Us | Waystar | CoverMyMeds | CentralReach | Customer Demand |
|---------|-----|---------|-------------|-------------|----------------|
| ... | Y/N/Partial | ... | ... | ... | X asks |

---

## WORKFLOW GAPS (manual steps customers still do)

| After Annie Does... | Customer Still Has To... | Automation Opportunity | Effort |
|---------------------|------------------------|----------------------|--------|

---

## INTEGRATION REQUESTS

| System | Customers Asking | Use Case | Effort |
|--------|-----------------|----------|--------|
| ... | [names] | [what they need] | S/M/L |

---

## BACKLOG (lower priority, track for later)

| Item | Category | RICE | Notes |
|------|----------|------|-------|
| ... | ... | XX | [why it's deferred] |

---

## ENGINEERING HEALTH SIGNALS

- Repeat bug rate: [are we re-introducing fixed bugs?]
- Feature request backlog growth: [growing or shrinking vs. last period]
- Onboarding friction trend: [improving or getting worse]
- Customer-reported vs. internally-found bugs ratio: [should trend toward internal]

Delivery

1. Save report to ~/Documents/Claude/SALES/Product_Insights_{YYYY-MM-DD}.md
2. Display in terminal: Top 3 sprint items with RICE scores, biggest bug, biggest feature gap
3. Ask: "Want me to create Jira/Linear tickets from these, drill into a specific customer's requests, or scope any feature?"

Important Rules

• Every item traces to a customer. No "we should probably build X." Only "Customer Y said Z on [date]."
• Quote verbatim. Engineers need to hear the customer's actual words, not paraphrased requirements.
• RICE everything. No prioritization by gut. Score it, rank it, defend it with data.
• Scope to MVP. For every feature request, define the smallest useful version.
• Separate bugs from features. Bugs get fixed. Features get scoped. Don't mix them.
• Track workarounds. A customer's workaround tells you exactly what the MVP needs to do.
• Competitive context matters. If a competitor has it and customers mention it, priority goes up.
• Default period: Last 14 days (sprint-aligned). User can specify any range.