NPS Proxy

Measure user satisfaction from behavior when you don't have survey data — frustration signals, engagement patterns, and conversion behavior as a real-time proxy for how users feel.

When to Use

• "What's our user satisfaction looking like?"
• "We don't run NPS surveys — can Fullstory give us a satisfaction signal?"
• "Which users are likely promoters vs detractors based on behavior?"
• "Did satisfaction improve after the redesign?"
• "Show me our 'frustration score' trend over time"

Mental Model

NPS (Net Promoter Score) asks users: "How likely are you to recommend us?" Fullstory doesn't do surveys. But you can build a behavioral proxy from three signals:

1. Success signals (promoters): Completed key tasks, returned multiple times, used advanced features, high engagement
2. Frustration signals (detractors): Rage clicks, dead clicks, errors, form abandonment, one-and-done sessions
3. Engagement signals (passives): Consistent but shallow usage, moderate session frequency, no strong positive or negative signals

The ratio of success to frustration signals gives you a directional satisfaction measure. It's not a survey. It's a behavioral proxy. Treat it accordingly.

Workflow

Step 1: Define success and frustration signals

Work with the user to define what "happy" and "unhappy" look like in their product:

Success signals (any 2+ = likely satisfied):

• Completed a purchase, signup, or key conversion
• Used the product 3+ times in 30 days
• Session duration >5 minutes (engaged, not bouncing)
• Used advanced features (beyond basics)
• Returned after first session (retained)
• No frustration events in any session

Frustration signals (any 1 = likely unsatisfied):

• Rage clicked any element
• Dead clicked any element
• Hit a console error or network failure
• Abandoned a form mid-completion
• Bounced after 1 page view (<10 seconds)
• Visited help/support pages repeatedly

Step 2: Build satisfaction segments

fullstory:build_segment("users with at least 2 success signals and 0 frustration signals in last 30 days") → promoters
fullstory:build_segment("users with at least 1 frustration signal in last 30 days") → detractors
fullstory:build_segment("users with neither strong success nor frustration signals in last 30 days") → passives

Step 3: Compute the proxy score

fullstory:build_metric(query="unique users", output_type="single_number")
→ compute for promoters segment → 4,200
→ compute for detractors segment → 1,800
→ compute for passives segment → 6,400

Behavioral Proxy Score = % Promoters - % Detractors
= (4,200/12,400) - (1,800/12,400)
= 33.9% - 14.5%
= +19.4

Step 4: Present with caveats

## Behavioral Satisfaction Proxy — Last 30 Days

### Signal Distribution
- Promoters (2+ success, 0 frustration): 4,200 (33.9%) ✅
- Passives (neither strong positive nor negative): 6,400 (51.6%) 
- Detractors (1+ frustration): 1,800 (14.5%) 🔴

### Proxy Score: +19.4
(A positive score means more promoters than detractors. The theoretical range is -100 to +100.)

### What This Means
About 1 in 3 users show strong satisfaction signals — they're completing tasks, returning, and not hitting frustration. About 1 in 7 users show frustration — they're rage-clicking, hitting errors, or abandoning.

### Top Frustration Drivers
1. Rage clicks on "Apply Promo": 412 users (the biggest detractor driver)
2. Checkout form abandonment: 340 users
3. Console errors on /dashboard: 89 users

### Trend
- This month: +19.4
- Last month: +17.2
- Improvement: +2.2 points 🟢
→ Frustration is decreasing, satisfaction trending up

### ⚠️ Important Caveat
This is a behavioral proxy, NOT an NPS survey. It measures what users DO, not what they SAY. Some frustrated users might still recommend you. Some happy users might not. Use as a directional signal, not a precise metric.

Step 5: Segment the score

Break down by user properties to find where satisfaction lives:

By plan:
Enterprise: +42 (high satisfaction — they get value)
Free: +8 (mediocre — some get value, many don't)

By device:
Desktop: +24
Mobile: +8 (mobile users are twice as frustrated)
→ Mobile UX is dragging down overall satisfaction.

By acquisition:
Organic: +28
Paid ads: +4 (paid users are less satisfied — possible expectation mismatch)

Step 6: Track over time

Build a trend of the proxy score:

fullstory:build_metric(query="unique users", output_type="trend")
→ compute for promoters, detractors, and passives over time
→ plot: proxy score over the last 6 months
→ annotate key events (deploys, feature launches, campaigns)

Behavioral NPS by Product Type

Guidelines

• Always caveat: "This is a behavioral proxy, not a survey." Repeat it in every report. Don't let anyone mistake this for real NPS.
• The proxy is most useful as a trend. A single month's score (+19.4) is less meaningful than "up from +17.2 last month." Track direction, not absolute value.
• Frustration signals are more reliable than success signals. A rage click definitely means frustration. A completed purchase might mean satisfaction, or it might mean "I needed this and powered through a bad experience." Weight frustration signals higher.
• Don't compare your proxy score to industry NPS benchmarks. An NPS of +40 is "excellent." A behavioral proxy of +40 means something completely different. They're different scales.
• If you DO have survey data (NPS, CSAT, CES), correlate it with the behavioral proxy. You might find that session rage clicks are a leading indicator for NPS drops 2 weeks later — that's much more valuable than the proxy alone.