Score calls with scorecards

Define weighted criteria, target the calls a scorecard applies to, and find the resulting scores and reasoning.

A scorecard describes what a good call sounds like for your business: a set of named criteria, each with a weight. After a call ends, the analysis scores every criterion from 0 to 100, and the scorecard’s weights combine those into one overall score. Scorecards live under Settings > Call Analysis > Call Scoring.

How a call picks a scorecard

At most one scorecard scores a call. Candidates are read in list order, top first, and the first one whose targeting matches wins — the rest are not applied. That makes the order in the table a decision, not decoration: put your most specific scorecard above the general one.

A scorecard is skipped when any of these is true:

It is paused or archived.
The call is internal. Internal calls are never scored.
The call is shorter than the scorecard’s minimum duration.
The direction does not match the scorecard’s direction setting.
The scorecard names extensions and the call’s extension is not one of them. Inbound calls are matched on the extension that answered; outbound calls on the extension that placed the call.
The scorecard names call groups and the call did not route through one of them. Call groups are an inbound concept, so this filter is not applied to outbound calls.
The scorecard includes or excludes categories and the call’s category does not qualify. Category targeting needs the call classified first, so a call with no category never matches an include filter.
The scorecard’s AI Callers setting does not match the call. A practice call placed by an AI caller carries that caller; every other call carries none. Any AI Caller scores only practice calls, Specific AI Callers scores only practice calls from the callers you name, and Exclude AI Callers scores only real calls. No filter scores both.

A scorecard set to Any AI Caller or Specific AI Callers treats practice calls differently in two ways: its direction setting is ignored, because a practice call is recorded as outbound even though it rings the employee’s extension, and an extension filter is matched on the extension that answered the practice call rather than the one that placed it.

The Specific Call Assistants field is stored with the scorecard but does not currently narrow which calls are scored. Use direction, extensions, call groups, and categories to target a scorecard.

Create a scorecard

Open Settings > Call Analysis > Call Scoring and select Add Scorecard.
Under Basic Info, name the scorecard for what it measures, such as Service Quality or Outbound Follow-up, and add a description if others will maintain it.
Set Minimum Duration. Calls shorter than this are not scored, which keeps hang-ups and wrong numbers out of your averages. New scorecards start at 30 seconds. The limit also applies to AI caller practice calls, which are often shorter than the default.
Leave Status active. A paused scorecard stays configured but stops scoring new calls.
Under Targeting, choose the calls it applies to: a category filter of All, Included, or Excluded; a direction; optionally, specific extensions or call groups; and an AI Callers setting of No filter, Any AI Caller, Specific AI Callers, or Exclude AI Callers. Included and Excluded both require at least one category, and Specific AI Callers requires at least one active AI caller. Choose Exclude AI Callers on scorecards that measure real customer conversations so practice calls do not affect their averages.
Under Criteria, add at least one criterion — a scorecard with no criteria cannot be saved.
Select Create Scorecard.

Once it is active, matching calls are scored automatically as they finish; there is nothing to run by hand.

Weight the criteria

Each criterion takes a name, a weight from 0 to 100, and an optional description that tells the analysis what the criterion is really asking about. A name can be up to 255 characters, a description up to 5,000, and each scoring guideline up to 1,000; the editor stops accepting text at the limit and shows a character count once a description or guideline reaches 80% of it.

Weights do not have to total 100. The overall score is the weighted average of the criterion scores, so weights are read relative to each other.
The total may not exceed 100. The weights chip turns red and Save is blocked until you reduce one.
Use the weights to express priority: a criterion at 30 moves the overall score three times as much as one at 10.
Advanced scoring guidelines let you describe what Excellent, Good, Average, and Needs Improvement look like for that criterion. Filling them in produces more consistent partial credit than a name alone, and the wording is what a coach can quote back to the person.

Maintain the list

The table shows each scorecard’s criteria count, minimum duration, category filter, how many analyses have used it, and its status.

Order arrows move a scorecard up or down, changing which one wins a call.
Duplicate copies a scorecard, names the copy with a (Copy) suffix, and places it last so it cannot outrank the original by accident.
Edit opens the same form. Changes affect calls scored from then on; scores already recorded are not recalculated.
Delete is permanent. Calls already scored keep their scores; new calls stop being scored against it.
An Archived badge means the scorecard was set that way by Steer Phones. It does not score new calls and cannot be reactivated from the dashboard — contact support.

Where the scores appear

Call Log can show a score column through the column picker, and the call’s detail panel has a Scoring tab.
In that tab, View detailed scores expands each criterion with its score, its weight out of the scorecard’s total, the reasoning behind the score, and quotes from the conversation. A criterion removed from the scorecard after a call was scored still appears, so old results stay readable.
Insights > Today averages the day’s scores in Avg call score.
Insights > Coaching ranks the low-scored calls and pairs them with coaching tips from the same analysis.
Alert rules can trigger on a scorecard’s score or on which scorecard was used.

Expected result

Matching calls finish with an overall score and a per-criterion breakdown, exactly one scorecard is credited per call, and the list order reflects which scorecard you want to win.

Troubleshooting

No calls are being scored: Confirm at least one scorecard is active, that its minimum duration is not longer than your typical call, and that its direction and category filters can match real traffic.
The wrong scorecard is being applied: The first match in list order wins. Move the more specific scorecard above the general one.
A scorecard with a category filter never matches: Category targeting requires the call to be classified into one of your categories first. Check that the categories exist and are active.
Save is disabled: A criterion is missing a name, a weight is outside 0–100, the weights total more than 100, an Included/Excluded category filter has no category selected, or AI Callers is set to Specific AI Callers with no caller selected.
Practice calls are showing up in a customer scorecard: A scorecard with No filter under AI Callers scores practice calls too. Set it to Exclude AI Callers, or place a practice-call scorecard above it.
Practice calls are not being scored: Check that a scorecard set to Any AI Caller or Specific AI Callers is active and above any scorecard that would otherwise catch the call, that its minimum duration is short enough for a practice call, and that the caller is named if the mode is Specific AI Callers.
The score looks wrong for one criterion: Open the detailed scores and read the reasoning and quotes. If the criterion is ambiguous, sharpen its description or fill in the scoring guidelines.
Old calls did not change after editing a scorecard: Scores are recorded when the call is analyzed and are not recalculated retroactively.