Templates

A template is the list of what the AI has to check in a call. It is your usual quality scorecard, written so that a machine can apply it to every call and not only to a sample.

It has two kinds of items:

• Scored criteria. Each one is worth a number of points, and together they make up the call’s score.
• Data to note down. Things the AI notes from each call without scoring them, such as the reason for a cancellation.

1. In Templates, click New template to start from scratch, or choose one under Start from an example and click Use this one.

2. Give it a name and add or adjust the criteria.

3. Click Create template. Until then, nothing is saved to the account.

The available examples are “Customer Support (Standard)”, “Sales Discovery”, “Technical Support”, “Cold Calling / Outreach”, “Real Estate Qualification”, “Job Interview Screening”, “Appointment Setting” and “Customer Retention”. They are a starting point: you would normally change their text so that they say what your operation requires.

A criterion has a name, an explanation of what to check, a way of scoring and a maximum number of points.

This is the field that affects the result most. Write it the way you would explain it to someone who is listening to the call for the first time and does not know your company.

Instead of	Write
Correct verification	The agent asks for the full name and the ID number before giving any account details.
Good closing	Before hanging up, the agent sums up what was agreed and asks whether the customer needs anything else.
Offers retention	When the customer says they want to cancel, the agent offers at least one alternative before processing the cancellation.

Three things always help:

• Say who and when. “The agent”, “before giving any details”, “when the customer asks to cancel”.
• Describe something that can be heard. “Is friendly” is an opinion; “greets with the company’s name and their own” can be checked.
• Say what happens in borderline cases. “If the customer has already identified themselves in the automated menu, confirming the name is enough.”

Kind	How it works	What it is for
Yes or no	All the points or none.	Steps that are either done or not done: reading a legal notice, asking for a piece of information.
Scale	Any value between zero and the maximum. You can describe levels: what has to happen to earn each number of points.	Things that are done better or worse: explaining an offer, handling an objection.
Text only	It does not score. It turns the criterion into a data field.	What you want to know about the call without evaluating it.

In a scale with levels, one level has to be worth all the criterion’s points. An example for “Active Listening”, worth 10 points:

Points	What has to happen
10	Restates what the customer asks for and does not interrupt them.
5	Understands the request, but interrupts the customer or makes them repeat information they had already given.
0	Answers something other than what the customer asked for.

Select It is mandatory for anything that must never be missing: an identity check, a legal notice. If a call earns less than half the points of a mandatory criterion, it goes into To review even if its total score is good. With Slack or Teams notices turned on, a notice also goes to the channel.

A criterion can rely on documents from Knowledge: the welcome script, the returns policy, the month’s offers. Under Documents it can consult, click Add document. When it scores, the AI reads the paragraphs of the document that have to do with the call and says which one it relied on.

Without documents, the AI knows only what you wrote in the criterion.

Try it before saving applies the criterion, exactly as you have written it, to a real call in your account. You see the points, the explanation and the cited line. It does not change the call’s result or use any of your plan’s analyses.

This is how you refine the wording: try it, read what the AI understood, fix the text and try again. Another call changes the test call. You can run up to 40 tests per hour.

A data field has a name, an instruction saying what to note down, and an answer format:

Format	Example
Text	“Why the customer wants to leave”
One option	“Product being discussed”, with a list of at least two options
Yes or no	“Is a competitor mentioned?”
List	“Products that were offered”

In each call you see them under Noted in this call. If you later want a data field to count toward the score, open it and click Make it score.

One template carries the Default tag: it is the one that scores calls with no campaign and calls in campaigns that have not chosen a template. To change it, open the menu of another template and choose Use as default.

Changes to a template apply to calls analyzed from then on. Calls that were already scored keep their results. To apply the new version to them, select them in Calls and click Analyze again.

When you delete a template, the campaigns that used it are analyzed with the default template from then on.