Containment Quality
What it mesures
Inquio reads each non-escalated conversation and judges whether the user's request was genuinely handled, not just kept away from a human agent.
A conversation contributes positively when the bot:
A conversation contributes positively when the bot:
- resolves the request without a handoff to a human
- gives an answer the user accepts, rather than one they push back on
- ends with a signal that the user is satisfied — thanks, confirmation, or simply moving on without repeating the request
- keeps the user in the conversation because it was helping them, not because there was nowhere else to go
A conversation drags the score down when the bot:
- avoids escalation but also avoids actually answering
- gives an irrelevant or generic response that technically ends the conversation
- leaves the user visibly frustrated, repeating themselves, or rephrasing the same request
- contains no positive signal at all — no gratitude, no "that's all I needed," nothing to suggest the user got what they came for
This is the distinction most readers get wrong: containment is not the same as deflection. A bot that avoids escalating users is not automatically doing a good job. If it is simply refusing to hand off while failing to help, Containment Quality will score this as a failure, even though the conversation never reached a human.
How it affects your Inquio score?
| Dimension | Weight |
|---|---|
| Resolution Accuracy | 40 % |
| Containment Quality | 25 % |
| Dialog Fluency | 20 % |
| Safety and Compliance | 15 % |
A ten-point improvement in Containment Quality moves your overall score by 2.5 points.
Reading your grade
| Score | Grade | What it means |
|---|---|---|
| 90-100 | A | Non-escalated conversations are reliably and genuinely resolved. |
| 80-89 | B | Mostly resolved, with a recurring gap in one or two topics. |
| 70-79 | C | Resolved often enough to be useful, but users push back or repeat themselves regularly. |
| 60-69 | D | A substantial share of non-escalated conversations end without real resolution. |
| 0-59 | F | The bot is avoiding escalation without actually helping users. |
The trend arrow next to the score compares this report card with the previous one for the same bot. See Trends and benchmarks for how the comparison period is chosen.
What usually drives a low score
- No escalation path at all. The bot has nowhere to send a request it cannot handle, so it closes the conversation with a generic or irrelevant reply instead.
- Escalation avoided by design. The bot is tuned to keep conversations contained even when it has nothing useful to offer, which trades a good containment rate for a bad Containment Quality score.
- Missing confirmation signals. The bot never asks whether the answer helped, so there is no positive signal to register even when the user was actually satisfied.
- Topic-specific gaps. Certain topics have no real resolution path, so every conversation on that topic ends the same unresolved way.
How to improve it
- Open Issues and Recommendations and filter to issues tagged as affecting Containment Quality.
- Start with the issue that has the highest impact and the largest number of affected conversations.
- Read several of the affected conversations in full, paying attention to how each one ends.
- Distinguish conversations that end because the user got what they needed from conversations that simply stopped.
- Fix the underlying gap — a missing answer, a missing escalation option, or a missing confirmation step — before adjusting wording.
- Re-upload a fresh dataset after the change and compare the trend arrow on the next report card.
Where to find it
- Bot Report Card — the Containment Quality tile, with score, grade, and trend. Click the tile for the breakdown.
- Bot Report Card → Benchmark comparison — your score against the average for your industry.
- Issues and Recommendations — the specific conversations behind the score.