What Users Talk About
9 intent categories across 2,500 conversations · Character.ai companion data
Topic Map
Size = conversation volume · Color = quality score · Click Roleplay to explore sub-topics
≥75≥55≥40<40
Topic Breakdown
Click a row to see intent deep dive
| Intent | Conversations ↓ | Avg Quality | Satisfaction | Avg Turns | Top Failure | Trend (7d) |
|---|---|---|---|---|---|---|
Roleplay5 sub-topics | 700 | 72/100 | 71% | 35.2 | Context Loss | +3.2% |
Emotional Support | 450 | 65/100 | 58% | 22.4 | Tone Break | -1.4% |
Casual Chat | 375 | 76/100 | 78% | 18.1 | Repetition | +0.8% |
Creative Storytelling | 300 | 74/100 | 73% | 30.3 | Coherence Loss | +2.1% |
Advice Seeking | 200 | 59/100 | 52% | 15.4 | Hallucination | -2.8% |
Companionship | 175 | 71/100 | 70% | 20.6 | Tone Break | +1.5% |
Humor & Entertainment | 125 | 77/100 | 80% | 12.3 | Repetition | +4.2% |
Learning & Exploration | 100 | 63/100 | 58% | 18.7 | Hallucination | -0.6% |
Philosophical Discussion | 75 | 70/100 | 68% | 25.1 | Context Loss | +1.0% |
Failure Heatmap
Intent × failure type hot spots — where AI breaks down most in companion conversations
High
Medium-High
Medium
Low
| Intent | 😤Tone Break | 🔀Context Loss | 🎭Character Break | 🌀Hallucination | 💭Coherence Loss |
|---|---|---|---|---|---|
| Roleplay | LOW | HIGH | MED-HIGH | LOW | LOW |
| Emotional Support | HIGH | LOW | LOW | LOW | — |
| Casual Chat | LOW | — | — | — | LOW |
| Creative Storytelling | — | LOW | LOW | — | MED-HIGH |
| Advice Seeking | LOW | — | — | MED-HIGH | LOW |
| Companionship | MED-HIGH | LOW | LOW | — | — |
| Humor & Entertainment | — | — | — | LOW | — |
| Learning & Exploration | — | LOW | — | MED-HIGH | LOW |
| Philosophical Discussion | LOW | MED-HIGH | — | LOW | LOW |