Understanding an evaluated conversation (Quality)

Last updated: August 17, 2026

When Solidroad evaluates a conversation, it returns an overall score plus a result for every section on the scorecard. This article explains what each of those results means and how it ties back to the way your scorecard was set up.

Sections 1-2 cover how individual sections are marked. Sections 3-5 cover the three things that most often make a score look surprising. Section 6 covers what admins see on a calibrated conversation.

Before you start: Any user with access to Quality can view an evaluated conversation. Changing how sections are scored — marking type, exclusion criteria, scorecard fails or linked processes - requires Admin access and is done on the scorecard, not on the conversation.

1. Graded sections

A graded section is scored on a scale, so the conversation can land anywhere between 0 and the section's Top Score. Graded marking is normally used for softer behaviours like empathy, tone and rapport, where how well matters as much as whether.

What it means: The result tells you which performance level the conversation reached, using the same three levels defined in the section's scoring guide.

Level reference:

Poor - shown in red.
Average - shown in amber.
Good - shown in green.

How it affects the score: The section contributes the points awarded, out of that section's Top Score.

Why it matters: The levels aren't fixed percentage cut-offs - they map directly to the poor, average and good descriptions written into that section when the scorecard was built. If a result looks wrong, the fix is almost always to sharpen the wording in the scoring guide rather than to dispute the score.

2. Pass/fail sections

A pass/fail section has no middle ground. Pass/fail marking is normally used for objective, checkable actions - verifying an account, following a security step, giving the correct product information.

What it means: The conversation either met the requirement or it didn't. A pass is shown in green, a fail in red.

How it affects the score: A pass awards the section's full Top Score. A fail awards 0. There is nothing in between.

Why it matters: A failed pass/fail section costs the section's full Top Score, not a portion of it. This is why a conversation that reads well can still score lower than expected - one missed verification step on a 10-point pass/fail section removes all 10 points.

3. Excluded sections

Sometimes a scorecard section doesn't apply to a conversation at all. A refund-handling section has nothing to measure in a password-reset chat. Rather than penalise the agent, Solidroad excludes the section.

Exclusion is driven by the exclusion criteria set on that section when the scorecard was built. If the conversation matches the criteria, the section is excluded automatically.

What it means: The section was not applicable to this conversation, so it wasn't assessed. It is not a 0 and it is not a fail.

How it affects the score: The section is removed from the scoring calculation entirely - it comes out of both the points awarded and the points available.

Worked example:

Your scorecard has 10 sections, each with a Top Score of 10 - so the maximum is normally 100.
On this conversation, one section is excluded.
The maximum for this conversation is now 90, and the agent is graded out of 90.
Scoring 81 means 81/90 (90%) — not 81/100 (81%).

Why it matters: This is the single most common source of confusion when comparing conversations. Two agents can be marked out of different totals depending on what came up in their conversations, which is exactly the point - nobody is penalised for a scenario that never arose. Compare percentages rather than raw scores.

4. Scorecard fails (auto-fail / critical fail)

A scorecard fail is the most severe outcome available. When one is triggered, the entire conversation scores 0 - every other section, including sections that passed, is overridden.

You may hear this called an auto-fail or critical fail. It exists for behaviours that make the rest of the conversation irrelevant: a compliance breach, a data-protection failure, abusive conduct.

What it means: A condition your team defined as unacceptable was met, so the conversation cannot pass on the strength of anything else in it.

How it affects the score: The overall score becomes 0, regardless of how the individual sections were marked.

Configuration reference (Admin): A scorecard fail can be set up in two ways, and both produce the same outcome on the conversation.

Failure condition - A scorecard section whose only job is to define a scorecard failure. It carries no score of its own; it exists purely to catch the behaviour.
Section-level scorecard fail - A section that scores normally and also carries a scorecard fail, so failing that section in a particular way zeroes the whole scorecard.

Why it matters: Because both routes produce the same 0, it's worth checking which one is in play before you change anything. Removing a scorecard fail from a scoring section also changes how that section is scored day to day, whereas a failure condition can be adjusted on its own.

5. Process matching

A process is a reference document (a help centre article, an uploaded file, or a document written in Solidroad) linked to a scorecard section so Solidroad can assess whether the agent followed the correct workflow.

Before it can assess adherence, Solidroad has to identify which process the conversation relates to. That step is process matching.

What it means:

A process was matched - Solidroad tied the conversation to one of the processes linked to the scorecard, and assessed how closely the agent followed it. That gives you a process adherence result.
No process matched - Solidroad could not tie the conversation to any linked process, so there was no defined workflow to assess the agent against.

How it affects the score: A matched process produces an adherence result that counts towards the total in the normal way.

Why it matters: "No process matched" is not a mark against the agent - it means the conversation covered something your linked processes don't describe. If you see it often, that's usually a signal to add or broaden a process rather than to change the scorecard.

6. Calibrated conversations (Admin)

Calibration is how you correct Solidroad when you disagree with a result, and it's how it learns your standards. Once a conversation has been calibrated, admins can see both the original and the corrected score.

What it means: The original score is shown with an arrow pointing to the corrected score - for example 6 → 8 — so you can see what was changed and by how much.

How it affects the score: The corrected score is the conversation's score. The original is retained for your calibration record.

Why it matters: The before-and-after view is admin-only, so agents see the corrected score as their score without the original beside it. When you talk an agent through a calibrated conversation, lead with the corrected score.

Quick reference

Use this as a cheat sheet when you're scanning a conversation and want to know what a result means.

Result reference:

Poor / Average / Good - A graded section, marked against the three levels defined in its scoring guide. Red, amber and green respectively.
Pass - A pass/fail section that met the requirement; awards the full Top Score. Green.
Fail - A pass/fail section that didn't; awards 0. Red.
Excluded - The section didn't apply and has been removed from the total entirely. Not a 0.
0 overall - A scorecard fail was triggered; every other section is overridden.
No process matched - No linked process fitted this conversation, so there was no workflow to assess adherence against.
6 → 8 - The score was calibrated; the second number is the score that stands.

Glossary of key terms

Top Score: The maximum points available on a single scorecard section.
Graded marking: Scoring on a scale from 0 to the Top Score, against poor, average and good descriptions.
Pass/fail marking: Scoring that awards either the full Top Score or 0, with nothing in between.
Exclusion criteria: Rules on a section that remove it from scoring when the conversation doesn't call for it.
Excluded section: A section that didn't apply to this conversation and has been taken out of the total.
Scorecard fail: A triggered condition that sets the entire conversation's score to 0. Also called an auto-fail or critical fail.
Failure condition: A scorecard section that exists only to define a scorecard failure and carries no score of its own.
Process: A reference document linked to a scorecard section so Solidroad can assess workflow adherence.
Process adherence: The result reflecting how closely the agent followed a matched process.
Calibration: An admin correction to a score, used to teach Solidroad your standards.

You can now read any evaluated conversation with confidence, explain a score to an agent, and tell the difference between a genuinely low score and one that was affected by exclusion, a scorecard fail or an unmatched process.

Once you can read an evaluated conversation, it's time to shape the scorecard behind it! Click here to get started.

📚 Related article: How to create a scorecard | Solidroad Help Center

If you have any further questions or need assistance, please refer to our support resources or contact the Solidroad team via the Get Help tab within the platform.