Survey Results
Synthetic survey · 100 respondents · 94% completion
⚠️ Synthetic data reflects modeled patterns, not real human judgment. Use for direction, not decisions.
1. Response Quality
| Total respondents | 100 |
| Fully completed | 94 |
| Completion rate | 94% |
| Avg completion time | 189s (fully completed) |
| Realism score | 8/10 |
Bias patterns observed
2. Screening (All 100)
Primary Computer
| Computer | Count | % |
|---|---|---|
| Mac | 85 | 85% |
| Windows | 15 | 15% |
Daily Typing Hours
| Hours | Count | % |
|---|---|---|
| 8+ | 24 | 24% |
| 4–6 | 41 | 41% |
| 6–8 | 15 | 15% |
| 2–4 | 17 | 17% |
| <2 | 3 | 3% |
Heavy typers (6+ hrs): 39% — strong overlap with power users
Primary Work
| Role | Count | % |
|---|---|---|
| Software dev | 30 | 30% |
| Writing/Content | 16 | 16% |
| Management | 12 | 12% |
| Marketing | 12 | 12% |
| Student | 12 | 12% |
| Sales | 7 | 7% |
| Other | 11 | 11% |
3. Current Behavior (94 completed)
Most-Used Apps (select up to 3)
| App | Mentions | % |
|---|---|---|
| Email client | 83 | 88% |
| Slack/Teams | 65 | 69% |
| Google Docs | 55 | 59% |
| Code editor | 30 | 32% |
| Notion/Confluence | 29 | 31% |
| iMessage/WhatsApp | 15 | 16% |
Repetitive Typing Frequency
| Frequency | % |
|---|---|
| Constantly | 36% |
| Often | 32% |
| Sometimes | 20% |
| Rarely | 9% |
| Never | 3% |
68% type repetitive phrases often or constantly. This is the core pain point.
Current Speed-Up Tools
| Tool | % |
|---|---|
| Clipboard manager | 61% |
| AI writing assistant | 55% |
| Text expander | 41% |
| Autocomplete (specific apps) | 36% |
| Voice dictation | 10% |
| Nothing | 10% |
90% already use at least one tool. They're not starting from zero — they're dissatisfied.
4. Product-Market Fit
"Very Disappointed"
Superhuman method
| Response | Count | % |
|---|---|---|
| Very disappointed | 39 | 41.5% |
| Somewhat disappointed | 35 | 37.2% |
| Not disappointed | 20 | 21.3% |
Verdict: Borderline strong PMF. 41.5% crosses the 40% threshold. The 37.2% "somewhat disappointed" is a large swing segment convertible with better messaging or feature execution.
5. Feature Prioritization
Rank 1–5, 1 = most important
| Feature | Avg Rank | Order |
|---|---|---|
| Works in every Mac app | 2.17 | 🥇 |
| Learns writing style | 2.22 | 🥈 |
| 100% local (privacy) | 2.73 | 🥉 |
| Zero lag | 3.40 | 4th |
| One-time purchase | 4.66 | 5th |
Insight: Top two features are about universality. Privacy is important but secondary. Speed is table stakes. Pricing model is least important as a differentiator — people just want it to work.
6. Pricing
Max WTP (one-time)
| Price | % | Segment |
|---|---|---|
| $79 | 20% | Power users |
| $49 | 16% | Mixed (strongest single point) |
| $99 | 15% | High-intent power users |
| $39 | 15% | Casual, students |
| $29 | 15% | Students, budget |
| $59 | 9% | Mid-range (gap) |
| $19 | 6% | Skeptical |
| $129+ | 3% | Enthusiasts |
Key signals: Median WTP: $49–$59. Sweet spot ($49–$79): 51%. $49 is strongest single point (16%). $79 surprisingly strong (20%). $59 falls in a gap (only 9%).
Pricing Model Preference
| Model | % |
|---|---|
| One-time purchase | 86% |
| Free with limited features | 14% |
| Annual subscription | 0% |
| Monthly subscription | 0% |
Signal is overwhelming: 86% want one-time purchase. Zero chose subscription. This validates the pricing model decision strongly.
7. Competitor Landscape
| Tool | % Tried | Key Weakness |
|---|---|---|
| Apple Intelligence | 62% | Generic suggestions, limited app support |
| Grammarly | 51% | Cloud-based, changes voice, subscription |
| GitHub Copilot | 32% | IDE-only, not for prose |
| Espanso | 32% | Manual setup, static snippets |
| TextExpander | 27% | Expensive subscription, dated UI |
| Superwhisper | 10% | — |
Top Frustrations (coded from open-ended)
- Narrow app support — "copilot only works in the editor"
- Generic/personality loss — "makes everything sound the same"
- Privacy concerns — "don't trust my data isn't used for training"
- Manual setup — "espanso requires too much configuration"
- Performance — "lag interrupts my thinking"
- Cost — "grammarly premium is too expensive"
8. Privacy Attitudes (Likert 1–5)
| Statement | Mean | Interpretation |
|---|---|---|
| "Comfortable with AI tools sending writing to cloud" | 2.14 | Leans uncomfortable |
| "Privacy is a key factor" | 4.18 | Strong agreement |
| "Trust local-only AI more than cloud" | 4.02 | Strong agreement |
| "Care more about quality than where processed" | 3.00 | Neutral split |
Privacy is a genuine purchase driver. But they won't sacrifice quality for privacy (3.00/5 neutral) — the product must deliver on both.
9. Segment Analysis
PMF by Segment
| Segment | N | Very Disappointed | Somewhat | Not |
|---|---|---|---|---|
| Power user | 39 | 100% | 0% | 0% |
| Casual | 35 | 0% | 100% | 0% |
| Skeptical | 19 | 0% | 0% | 100% |
| Disengaged | 1 | 0% | 0% | 100% |
Power users (39%): 8+ hrs typing, mostly devs/writers, use multiple tools, highest WTP ($79–$129). This is the beachhead.
Casual (35%): 4–6 hrs typing, mixed roles, moderate WTP ($39–$49). Growth opportunity — convert with better onboarding and clearer value prop.
Skeptical (19%): Mix of Windows users, low typing hours. Not the target market — don't optimize for them.
10. Privacy × PMF Correlation
| Privacy Score | N | Very Disappointed |
|---|---|---|
| High privacy (4–5) | 72 | 54% |
| Low privacy (1–2) | 3 | 0% |
Strong signal: Privacy-conscious users overlap heavily with highest intent. Local-first positioning is a competitive advantage against cloud-based competitors.
11. Key Takeaways
PMF is borderline strong (41.5%). Above the 40% threshold but needs real-world validation.
The beachhead is power users. Devs and writers who type 6+ hours and already juggle multiple tools.
"Works everywhere" is #1. The universal-app value prop is the killer differentiator.
One-time purchase is non-negotiable. 86% chose it, zero chose subscription.
$49–$79 is the WTP band. $59 is defensible if value is clear.
Privacy is a genuine purchase driver. 4.18/5 agreement. Not just marketing fluff.
Competitors have clear weaknesses. Narrow app support, generic suggestions, cloud dependency.
The "somewhat disappointed" 37% is the conversion opportunity. They need clearer value demonstration, not more features.
12. Caveats
- Synthetic data — real survey may differ significantly
- Segment sizes (skeptical N=19, disengaged N=1) too small for robust conclusions
- 85% Mac split may be optimistic for a Mac-only product survey
- Open-ended responses are simulated
- No attention checks included
- 100% power-user → "very disappointed" mapping is a synthetic model artifact