Last updated: September 12, 2026
What a call center scorecard in Excel needs to do
A call center scorecard in Excel holds three things: the rubric with its weights, a scoring sheet where one row is one call, and a rollup that averages an agent's scored calls over a period. The file below carries all three, and the auto fail rule is a formula that zeroes the card outright.
Four tabs and about eight formulas is the whole build. The one that trips people sits in the score column, because an auto fail has to beat the arithmetic sitting above it.
=IF(COUNTIF($W3:$AA3,"Yes")>0,0,SUM($F3:$V3)/SUM(Rubric!$C$2:$C$18))
That goes in AD3 on the scoring tab, formatted as a percentage and filled down. A call worth 94 points with one auto fail flag reads 0, the agent's average on the rollup drops with it, and the supervisor opens the card again instead of filing it. Everything else in the workbook is bookkeeping around that line.
Download the spreadsheet
The workbook downloads as a spreadsheet with no form in front of it. Rubric holds the 17 criteria and their point values. Scoring is the working tab, one row per call, dropdowns already set on every criterion column. Rollup carries one row per agent for the period, and change log is four columns that start empty.
If you would rather rebuild the file than trust somebody else's workbook, the rubric alone is call-center-qa-scorecard.csv and every formula below is call-center-scorecard-formulas.csv. Both sit with every template we publish, ungated.
The rubric the file ships with
Six sections, 17 criteria, 100 points, with five auto fail rows underneath that carry no points at all.
| Section | Criteria | Points |
|---|---|---|
| Opening and identification | 3 | 15 |
| Discovery and listening | 4 | 25 |
| Communication quality | 4 | 20 |
| Resolution and accuracy | 3 | 25 |
| Compliance and process | 2 | 10 |
| Closing | 1 | 5 |
| Total | 17 | 100 |
The criterion wording and the reasoning behind each weight live on the full call center QA scorecard. To move points between sections, the method is on how to build a call scoring rubric, and the workbook takes the change fine as long as the rubric tab stays the only place the totals come from.
The formulas
The scoring tab uses a two row header. Row 1 carries the section name above each criterion column, row 2 carries the criterion name, and the first scored call lands in row 3. Columns F through V are the 17 criteria, W through AA are the five auto fail flags, and AB through AD do the math.
| Formula | What it does | Where it goes | Excel and Sheets |
|---|---|---|---|
=SUM($F3:$V3) | Adds the points awarded across all 17 criteria for one call | Scoring, AB3, filled down | Identical |
=SUMIF($F$1:$V$1,"Resolution and accuracy",$F3:$V3) | Section subtotal for one call, read off the section names in row 1 | Any section summary cell | Identical |
=COUNTIF($W3:$AA3,"Yes") | Counts auto fail flags on the call | Scoring, AC3 | Identical |
=IF(COUNTIF($W3:$AA3,"Yes")>0,0,SUM($F3:$V3)/SUM(Rubric!$C$2:$C$18)) | The score, with the auto fail override in front of the division | Scoring, AD3, percentage format | Identical |
=COUNTIF(Scoring!$C:$C,$A2) | Calls scored for one agent in the period | Rollup, C2 | Identical |
=IFERROR($C2/$B2,"") | Coverage percent, scored calls over calls handled | Rollup, D2 | Identical |
=AVERAGEIF(Scoring!$C:$C,$A2,Scoring!$AD:$AD) | Average score for one agent | Rollup, E2 | Identical |
=SUMIF(Scoring!$C:$C,$A2,Scoring!$AC:$AC) | Auto fail count for the period | Rollup, F2 | Identical |
=AND($AD3<>"",$AD3<0.85) | Conditional format rule that reddens a score under the threshold | Applied to AD3:AD500 | Identical string, different menu |
The threshold in that last rule is 0.85, which is where SQM Group puts the industry average. Move it to whatever your program calls passing. The <>"" guard matters because an empty cell evaluates as zero, and without it every unscored row down to 500 turns red the moment you build the file. A second rule, =$AC3>0 applied to A3:AD500 in amber, flags the whole row when an auto fail is on the card.
Dropdowns go on per column rather than per sheet, since a 5 point criterion allows 0 or 5 and an 8 point criterion allows 0 or 8. In Excel that is Data, Data Validation, Allow: List, with 0,5 typed into the source box. Google Sheets renamed the feature: Data, Data validation, Add rule, criteria set to Dropdown, each value added as its own item. Both leak the same way, because a paste overwrites the rule silently and takes the validation off the column without saying so.
A conditional formatting formula in Excel can point at another sheet, so the pass threshold can live on the rubric tab. Google Sheets refuses a cross sheet reference inside a custom formula and wants it wrapped as INDIRECT("Rubric!C20"). Validation has the mirror of that problem, since Excel before Microsoft 365 needs a named range when the list source sits on a different tab.
How the auto fail override works
An auto fail is a flag rather than a deduction, which is the entire reason it sits outside the arithmetic. Because the IF runs before the division and returns a hard zero, nothing an agent earned in the other 16 lines changes the number.
Most people who get this wrong write the auto fail as a large negative number instead, usually something like minus 100 in a spare column, and let it drag the total down. That kind of works on a single call. It stops working the moment two auto fails land on the same card, because minus 200 against 94 points scores negative 106, and a rollup averaging negative numbers alongside 89s makes an agent's period look either catastrophic or fine depending on how many blanks sit in the range.
Double counting is the subtler problem. A missed recording disclosure already costs 5 points on the compliance criterion, so scoring it as a deduction and a minus 100 says the same thing twice and leaves two cells to argue about in a dispute instead of one.
Mark the auto fail columns before anything else on the row. Totaling first tempts a reviewer to soften a flag so the number comes out reasonable, and that habit outlives the spreadsheet.
How the tabs are laid out
| Tab | What belongs there | What does not |
|---|---|---|
| Rubric | Section, criterion, points, auto fail yes or no. One row per criterion | Scores. Nothing on this tab changes call to call |
| Scoring | One row per call, with call ID, date, agent, reviewer, account, the 17 criterion columns, the five auto fail columns, and the three math columns | Agent averages, coaching notes, anything that summarizes across rows |
| Rollup | One row per agent per period: calls handled, calls scored, coverage, average, auto fail count | Individual calls. If you find yourself adding a call column here, it belongs on Scoring |
| Change log | Date, what changed, who approved it, version number | Everything else. Four columns is the whole tab |
Calls handled in column B is the one number the file cannot calculate. It comes out of your phone system, and it decides whether the average beside it means anything, since an 88 built on four scored calls out of nine hundred handled measures nothing. The period math is covered on roll scores up per agent.
Everybody deletes the change log, and you can see why, since it holds nothing and does no arithmetic. It exists so a supervisor comparing this quarter to last can check whether the rubric was the same rubric both times. A weight moved mid quarter turns a reweighting into a coaching win, and the chart looks right, so nobody catches it.
Where a spreadsheet stops working
Version drift arrives with the second reviewer rather than the thousandth call. One person with one file stays stable more or less indefinitely. Add a supervisor keeping a local copy and the copies diverge the first time somebody widens a criterion to settle a coaching argument and tells nobody. That usually surfaces within a quarter, as two agents scored against two different rubrics in the same monthly report.
The second one bites at the first dispute. A cell holding 0 under a column headed "Restated the caller's issue accurately before acting on it" carries no route back to the 30 seconds of audio behind it. You can add a timestamp column, and people do, and it gets filled in reliably for about two weeks. Once the review itself runs four minutes, that column is the first thing to stop getting typed, and the score is then only as defensible as a reviewer's memory of a call from nine days ago.
Manual consolidation is the one you can put numbers against. Say 14 agents at four scored calls each, 56 calls a month. The file handling on top of the listening is maybe an afternoon: paste the rows, refresh the rollup, build a sheet per agent for the one to ones. Take those same 14 agents to 20 calls each and you are at 280 rows, where the consolidation and reporting starts costing more hours than the listening did. Our own line, judgment rather than a published finding, is that a workbook holds up under about 100 scored calls a month with one reviewer and gets expensive past 300.
None of that argues against the file. Manual review covers 1 to 3% of calls across the industry by most third party counts, and a program running a spreadsheet well is ahead of one running nothing. Those numbers are where the format starts charging you for itself.
Run the same rubric against calls you already have
Before deciding the spreadsheet is fine, run the arithmetic on your own reviewer hours at what call review costs today.
What replaces the file is scoring against the transcript, where every deduction carries a timestamp and a per line confidence score and a disputed criterion opens at the second it was scored. That reads cleanly on binary lines like whether a callback number was read back digit by digit, and gets less reliable on tone and empathy, which Aircall and StackAI both note about automated scoring, so the subjective lines still want a human pass. Hand over recordings straight out of whatever your system exports, and a call review scores six months of them against this rubric and walk your supervisors through their own calls on Zoom for a fixed pilot price.
Frequently asked questions
How do I make a call center scorecard in Excel?
Build the rubric tab first, one row per criterion with its point value, because every other formula reads its totals from there. Then the scoring tab, one row per call, with a two row header carrying section names above criterion names. The two formulas doing the real work are =SUM($F3:$V3) for points earned and the auto fail override in the score column. Rollup and change log take about ten minutes between them.
What formula calculates a QA score?
Points earned divided by points available, formatted as a percentage. The auto fail check runs first and returns zero, so the full version is =IF(COUNTIF($W3:$AA3,"Yes")>0,0,SUM($F3:$V3)/SUM(Rubric!$C$2:$C$18)) rather than a bare division. Pointing the denominator at the rubric tab means a reweighting flows through without anybody editing 500 rows.
Is there a free call center scorecard template in Excel?
Yes, the workbook here is free and ungated, and so are the two CSVs beside it. Several vendors put a scorecard behind an email form, which is a fair trade if you want their onboarding and slow if you just wanted a file open in front of you.
How many criteria should a call scorecard have?
The template runs 17, and most working scorecards land between 12 and 20. What matters more than the count is whether each line names something a second reviewer can point at in the recording, since eight checkable criteria beat 25 that include "showed empathy" scored out of five. That argument is made properly on how to build a call scoring rubric.
Can I use this in Google Sheets?
Every formula on this page runs unchanged in Google Sheets. Two things differ in the menus rather than the math: a conditional formatting rule in Sheets cannot reference another tab directly and needs INDIRECT("Rubric!C20") around it, and validation is called Dropdown where Excel calls it a List.
Sources
- Trinsic feature sheet and call notes, 2026-08-27 and 2026-09-03. Verified internally.
- SQM Group on the 85% industry average QA score. Third party, unverified.
- Manual QA review coverage of 1 to 3% of calls (Verint, Gistly, ViciStack). Third party, unverified. Used as setup only.
- Aircall and StackAI on rule based scoring holding up better than tone scoring. Third party, unverified.
- The 17 criteria and the 100 point total. CallSniper's own published template, editorial judgment on the weights.
- The 100 and 300 scored call thresholds for spreadsheet workability. Editorial judgment, not a published finding.
- Excel and Google Sheets behavior on cross sheet references in conditional formatting and data validation. Product behavior, pending confirmation against current versions.