CONTENTS
What's in the templateWhy critical checks stay out of the averageWhere the compliance checks come fromHow to write checks that hold upSetting a pass markGetting two reviewers to agreeThe problem a better form doesn't fixRunning your checks on every recorded call- Get it: the free Excel scorecard, with 13 checks, weights adding to 100 and automatic scoring. A filled example sits on its own tab.
- Critical checks sit outside the score. A serious miss fails the call even if everything else went well.
- Every result needs evidence: a quote from the call. That's what ends arguments with agents and buyers.
Most QA scorecards were built by someone who needed one by Friday. They've got 25 questions, half of them are about tone, and the one disclosure that could cost you a buyer is worth the same 4 points as "used the caller's name".
A call center QA scorecard is a written list of checks used to grade a call, usually with a weight on each check and a short list of checks that fail the call outright. A good one lets two reviewers look at the same call and land on the same result.
This one is built for outbound sales and live-transfer floors. You can get the Excel file here. The rest of this page explains why it's laid out the way it is, so you can make it yours.
What's in the template
It opens on the scorecard. Pick Pass, Fail or N/A down the yellow column, and the score at the top updates as you go. A second tab shows a made-up call filled in, and a third explains the scoring if you want it.
| List | Checks | Points | Fail the call if missed |
|---|---|---|---|
| Must say | Call is recorded, seller named, sales call stated, offer described, your qualifying questions, answers confirmed back, caller agreed to the transfer, clear next step | 75 | Recording notice, caller's agreement |
| Conduct | Let the caller finish | 5 | None |
| Must NOT say | Kept going after a stop request, guaranteed an outcome, lied about who's calling | 20 | All three |
Every check is yours to rewrite, and the brackets are where your own details go. Your buyer's criteria get one row each, so add as many as you need and move the weights to suit. A collections floor will put more on compliance, and a sales floor more on qualifying and the offer.
Why critical checks stay out of the average
Take the example call in the file. The agent gets the recording notice in early, names the firm, asks every qualifying question and hands off cleanly. It scores 89, above the pass mark of 85.
The agent also transferred the caller without ever getting a yes. So the result reads FAIL, with the reason right under it: "Failed on: Caller agreed to the transfer".
That's the whole reason for checks that fail the call on their own. With a plain weighted score, a pleasant call can absorb a serious miss and your report tells you the floor is fine. Kaizo's guide on this puts it well: auto-fails are for failures that "are not averageable."
Keep the list short, though. Kaizo's test is a useful filter: is the check objective, provable from the transcript, unambiguous, severe, and within the agent's control? If two of your reviewers have ever disagreed on it, it shouldn't be Critical. "Showed empathy" belongs in the weighted part, however much you care about it. They suggest most teams can justify five or fewer, and that's where this template starts.
Where the compliance checks come from
The seller, sales-call and offer checks trace back to the FTC's Telemarketing Sales Rule. On an outbound sales call, 16 CFR 310.4(d) requires the telemarketer to disclose three things truthfully, promptly and clearly: who the seller is, that the purpose of the call is to sell something, and what's being offered. The FTC reads "promptly" as before any sales pitch.
The same rule covers do-not-call requests, and it bans false or misleading statements to get a sale. That's where the stop-request and misrepresentation checks come from.
The seller is the company providing the goods or services, and that may not be the company making the call. If you're running a campaign for someone, make sure your agents name the right one.
The recording notice is there because many floors need it for state law or a buyer's terms. Your rules depend on your states and your vertical. Medicare, ACA and debt relief all carry extra required language, so add your own lines and run the final version past your counsel.
How to write checks that hold up
One act per check. "Named the seller and said it's a sales call" is two checks. Split it, or a half-done call has no right answer.
Write what would be said, not a quality. "Stated the recording notice" can be checked. "Was transparent" can't, or at least not by two people in the same way.
Name your banned lines. "No promises outside the script" sends your reviewer off to find the script. "Guaranteed approval" and "a specific dollar amount" can be marked on the spot.
Say when a check applies. "If the caller asks to stop" tells a reviewer to mark N/A when it never came up. For a line that always applies, write "on every call" so it can't slide to N/A.
Make each qualifying question its own row. If your buyer pays for callers under 65 with no current attorney, those are two rows, word for word. You'll catch weak transfers before they do.
Write what you've actually been burned on. The disclosure your buyer audits. The promise that got a batch of transfers rejected. The line your new agents love to improvise. Those belong on the card ahead of anything from a generic list.
Keep it short enough to finish. SQM Group describes scorecards with up to 20 evaluator questions, and several QA guides put the working range nearer 8 to 12. Thirteen is where this one starts, and I'd cut before adding.
Use N/A properly. A call with no stop request shouldn't lose points for it. In the template, N/A takes the check out of the score altogether, which is why the example call is scored out of 90.
Setting a pass mark
The template starts at 85 to pass and 70 for review. Treat both as a starting point: some vendor templates suggest 85 as well, and there's no industry standard behind that number.
Score your last 20 calls before you announce anything. If almost everything passes, the card is too soft. If your best closer keeps failing, a check is probably wrong. Move the marks after that, and hold off on tying anyone's pay to the score until your reviewers agree with each other.
Getting two reviewers to agree
Once a month, have your reviewers score the same three calls separately, then compare. Wherever they disagree, the check is badly worded, so rewrite it. Practitioners commonly aim for scores within about five points of each other, though that's a rule of thumb.
The evidence column does a lot of work here. A reviewer who has to quote the words can't score on a general impression. An agent who sees "you'll definitely qualify" has much less to argue with than an agent who sees "-5, compliance".
The problem a better form doesn't fix
Now the awkward arithmetic. Call Centre Helper reports 3 to 5 calls per agent per month as the most common evaluation figure.
Say an agent has 60 conversations a day across 22 working days. That's 1,320 calls, and five of them is under 0.4%. However good your scorecard is, you're grading a tiny slice of that agent's month, and the call your buyer disputes is unlikely to be in it.
You've got two ways to close that. One is more reviewers, and at that volume you'd need a lot of them. The other is to have software run your scorecard on your recorded calls, and keep your reviewers on the ones it flags.
Running your checks on every recorded call
If you're on VICIdial, that's what Klariqo Call QA does. It connects with read-only access in about 10 minutes and works from your existing recordings.
Sign up, open Scorecard, and enter your checks, or start from an editable preset for outbound, SSDI, ACA, Medicare, debt relief or dental tourism. Turn QA on from Billing and your recorded calls get scored, with the line from the transcript behind what it flags.
Klariqo keeps the scoring simpler than a spreadsheet. There are two lists and no weights or pass marks: a Critical miss scores 0 and fails the call, any other miss sends it to review, and the calls that need a manager sit at the top of your Reviews list.
Each result also lands on a signed record that anyone can check, so when a buyer questions a call, you've got more than your own scorecard to show them. Try changing one word in a signed call to see how that works.
It's $0.05 a call for the first 5 minutes, then $0.01 a minute, with the signed record included. The spreadsheet stays free either way, and plenty of floors should start there: it'll tell you which checks are worth automating.
Questions people ask
What should a call center QA scorecard include?
What must be said, what must not be said, a weight on each check, and a short list of critical checks that fail the call on their own. Add Pass, Fail and N/A results, and an evidence column with a quote from the call.
How many questions should a QA scorecard have?
Most working cards land around 8 to 12 checks, and SQM Group describes up to 20 evaluator questions. Start small, because reviewers skim long forms and scores drift.
What is an auto-fail in call center QA?
A check that fails the whole call regardless of the score, such as ignoring a request to stop calling. It should be objective, provable from the transcript and within the agent's control. Most teams keep it to five or fewer.
What's a good QA pass mark?
85 is a common starting point and this template uses it, but no industry standard sets it. Score a sample of your own calls first, then move the mark to fit your floor.
How many calls should QA evaluate per agent?
Call Centre Helper reports 3 to 5 per agent per month as the most common figure. Software can score your recorded calls, which leaves reviewers free to focus on the flagged ones.
Is this template free?
Yes. Rewrite the checks and use it however you like.
