Voice of Customer & Product Feedback
Prism Customer Success Handbook — Section 2.8
Why this exists
CS sits closer to the customer than any other function, which makes it the natural conduit for what customers need from the product. But feedback only has value if it closes — if the customer eventually hears what happened to what they told us. A capture system that never reports back is a suggestion box: customers use it once, learn nothing comes of it, and stop.
The whole point of this page is the fourth step below. The first three are easy; most functions do them and then stop.
The loop
Capture → Categorize & route → Disposition → Close the loop with the customer.
1. Capture
Feedback arrives everywhere: cadence calls, EBRs, support tickets, escalations, offhand comments. Capturing it means recording it as a feedback item, not leaving it buried in a call note where it can't be aggregated.
What a captured item needs: what the customer wants, the account and segment, ARR context, whether it's blocking anything, and the customer's own framing of why it matters.
2. Categorize & route
| Category | Definition | Routes to |
|---|---|---|
| Bug | Something is broken or behaving wrongly | Support → Product/engineering |
| Feature request | A capability that doesn't exist | Product, with context |
| Friction / UX | The capability exists but is hard to use | Product |
| Praise | Something is working notably well | Logged; feeds advocacy identification (2.10 Advocacy Program) |
| Churn-risk driver | Feedback that is actually a renewal threat | Escalates (2.7 Escalation Management) and revises forecast (2.6 Renewal Forecasting) — does not just sit in a queue |
That last category is the important distinction. "We'll probably leave if this isn't fixed" is not a feature request — it's a risk signal wearing a feature request's clothing, and it gets treated as one.
3. What CS owns: synthesis, not relay
CS owns pattern synthesis. It does not own the roadmap.
The value CS provides Product is not a stream of forwarded requests — that's noise Product already gets from ten channels. The value is aggregation with context:
"Four Mid-Market accounts asked for scheduled-report branding controls this quarter. One ($40K) named it as a renewal blocker. Two are actively working around it in ways that add support load."
A single anecdote is weak. A pattern with ARR and renewal context attached is a prioritization input Product can actually use. Turning scattered individual requests into that is the CSM's real job in this loop.
CS does not decide what gets built, promise features to customers, or argue roadmap priority as though it owned the decision. It makes the strongest possible evidence-based case and accepts Product's call.
4. Disposition & closing the loop
Product owes CS a disposition on routed items: shipped, planned, declined, or parked. Without disposition, CS can't close the loop, and the loop is the point.
Closing the loop means telling the customer what happened — proportional to how the feedback arrived:
- Shipped: tell them, specifically, and connect it to what they asked for. This is one of the highest-trust moments available — the customer sees their input became real. It also feeds advocacy (2.10 Advocacy Program).
- Planned: tell them it's on the roadmap, without committing to a date CS doesn't control.
- Declined or parked: tell them honestly, with the reasoning. "We looked at this and decided not to pursue it, because ___" respects the customer more than silence. Customers accept no far better than they accept being ignored.
The failure mode this prevents: a customer requests something, hears nothing for a year, and concludes Prism doesn't listen — even if Product quietly shipped it. Unclosed loops waste the feedback and damage the relationship it came from.
Cadence
- Continuous capture — feedback is recorded when it arrives, not reconstructed later
- Monthly synthesis — the CSM aggregates the month's feedback into patterns and routes the synthesis to Product, rather than forwarding items one at a time
- Loop-closing — as dispositions come back, and surfaced deliberately at EBRs and cadence calls where "here's what happened with what you asked for" is a natural, high-value agenda item
Interfaces
- Product (2.5 Cross-Functional Interfaces): receives synthesis, owes disposition, owes advance notice of changes flowing the other way
- Escalation (2.7 Escalation Management): churn-risk drivers escalate rather than queue
- Forecast (2.6 Renewal Forecasting): feedback naming a renewal threat revises forecast immediately
- Advocacy (2.10 Advocacy Program): praise and shipped-feature moments identify advocacy candidates
Not yet modeled — v2 scope
The CSP logs activities but has no structured feedback record — category, disposition, linked accounts, loop-closed status. Without it, feedback lives scattered across activity notes and can't be aggregated into the patterns this page depends on. A structured VoC record is logged in BUILD_BACKLOG.md.
Until then, this page defines the target process; capture happens in activity notes.
Related pages
- 2.5 Cross-Functional Interfaces — the CS↔Product interface, both directions
- 2.6 Renewal Forecasting — churn-risk feedback as a forecast trigger
- 2.7 Escalation Management — where churn-risk drivers go
- 2.10 Advocacy Program — praise and shipped features as advocacy signals
- 4.3 EBR Kit — where loop-closing becomes an agenda item