In BPO (business process outsourcing), a KPI, or key performance indicator, is a measured number with an agreed target that shows whether an outsourced team delivers what the client pays for. Most BPO KPIs fall into four families: speed, resolution, customer outcome and quality. The ones that matter most to the client are written into the contract as SLA targets.
In short
- A KPI in a BPO is a metric that has a target, an owner and a review date; a number without a target is just a metric.
- The core KPIs in BPO are service level, handle time, first contact resolution, CSAT or DSAT, QA score, attrition and adherence.
- Every speed KPI needs a quality KPI next to it, or the team hits the speed target by cutting corners.
- The QA team has KPIs of its own: how much it reviews, how consistently it scores and how many of its scores survive a dispute.
- A QA score only works in a contract SLA when both sides agree on the rubric, who scores, how much is reviewed and how disputes are settled.
What does KPI mean in BPO?
KPI in BPO means key performance indicator: a metric the outsourcing provider and the client have agreed to track against a target. A KPI says how well an account is running, such as 80% of chats answered within a set time or a QA score of 90%. If the number has no target and no owner, it is a metric, not a KPI.
The word gets used loosely on the floor, so it helps to separate four terms that are often mixed up in BPO.
| Term | What it is | Example | Who usually owns it |
|---|---|---|---|
| Metric | Any number you can measure | Average handle time last week | Workforce or reporting team |
| KPI | A metric with an agreed target and review cadence | AHT under the agreed target for billing queries, reviewed weekly | Operations manager, team lead |
| SLA | A KPI written into the client contract, with consequences when it is missed | QA score of at least the agreed level each month, or a service credit applies | Account manager and client |
| KRA | Key result area: the part of a job a person is accountable for | “Quality” for a QA analyst, “productivity and quality” for an agent | The person and their manager |
So in a BPO, KRAs describe what each role is responsible for, KPIs measure how well that responsibility is met, and SLAs are the small set of KPIs the client can enforce. If you work on the vendor side and want the bigger picture of how quality is governed between client and provider, start with QA meaning in BPO. Teams that run several client accounts often move quality scoring onto a BPO QA platform, so each client’s conversations are scored against that client’s own standard.
What are the main KPIs in BPO?
The main KPIs in BPO are service level, average speed of answer, abandonment rate, average handle time, first contact resolution, CSAT or DSAT, QA score, critical error rate, schedule adherence, occupancy and attrition. Together they cover speed, resolution, the customer’s verdict, the quality of the work and the health of the team delivering it.
Here are the KPIs most BPO accounts track, with the formula each one usually uses. The client contract or the account’s KPI definitions document decides the exact rules, such as which contacts are excluded.
| KPI | Family | Formula | What it tells you |
|---|---|---|---|
| Service level | Speed | Contacts answered within the target time / contacts offered x 100 | Whether customers reach someone fast enough |
| Average speed of answer (ASA) | Speed | Total wait time of answered contacts / answered contacts | How long the average customer waits |
| Abandonment rate | Speed | Contacts abandoned before an answer / contacts offered x 100 | How many customers give up waiting |
| Average handle time (AHT) | Efficiency | (Talk or active time + hold time + after-contact work) / contacts handled | How long a contact takes, the main staffing input |
| First contact resolution (FCR) | Resolution | Contacts resolved on the first contact / total contacts x 100 | Whether the customer had to come back |
| CSAT | Outcome | Satisfied survey responses / total survey responses x 100 | How customers who answered the survey rated the contact |
| DSAT | Outcome | Dissatisfied survey responses / total survey responses x 100 | How often contacts leave customers unhappy |
| QA score | Quality | Points earned on the scorecard / points possible x 100, averaged over evaluated contacts | Whether agents follow the client’s quality standard |
| Critical error rate | Quality | Evaluations with at least one critical error / total evaluations x 100 | How often a contact breaks a rule that must never break |
| Schedule adherence | Workforce | Time spent in the scheduled activity / scheduled time x 100 | Whether agents are where the plan needs them |
| Occupancy | Workforce | Handle time / (handle time + available idle time) x 100 | How busy agents are while logged in |
| Attrition | Team | Agents who left in the period / average headcount x 100 | How fast the account loses trained people |
Each of these has a longer definition on our blog: how first contact resolution is calculated, what average handle time includes and how DSAT differs from CSAT. The QA score is often called an internal quality score, because it is measured by the team, not by the customer.
Which metrics in BPO hide each other?
Metrics in BPO hide each other in pairs: a speed metric can look healthy because a quality metric is falling, and the dashboard shows only the good half. The usual fix is to never set a target on one metric of a pair without watching the other. The table below lists the pairs that cause the most trouble.
| If you push this down or up | Watch this next to it | What goes wrong when you do not |
|---|---|---|
| AHT down | FCR and repeat contacts | Agents close fast and the customer comes back, so the account pays twice |
| Service level up | QA score | Agents rush to clear the queue and skip verification or notes |
| CSAT up | DSAT and survey response rate | A few happy replies hide many silent unhappy customers |
| QA score up | Critical error rate | The average looks fine while a small number of contacts break a hard rule |
| Occupancy up | Attrition and schedule adherence | Agents burn out and leave, and the account loses trained people |
This is also why a single “acceptable CSAT” figure is a weak way to judge an outsourced team: it only covers the customers who replied, and it says nothing about how the result was reached. The same logic applies inside any contact center, and our guide to call center QA metrics that predict CSAT goes deeper on how efficiency, quality and outcome metrics pull against each other.
What are the KPIs of QA in BPO?
The KPIs of QA in BPO come in two sets. The first is what QA measures on agents: QA score, critical error rate and compliance pass rate. The second is what the QA team itself is measured on: how many contacts it reviews, how consistently analysts score, how fast feedback reaches agents and how many scores survive a dispute.
Most guides on BPO KPIs stop at the first set. The second set is what the QA team should be judged on, and it is often never written down. One commenter on a Filipino BPO thread on Reddit passed on what a friend in QA was told: during calibration, the client blamed QA for the agents’ low scores, on the reasoning that a low QA score means QA is not coaching well. That is one second-hand account, but it shows what happens when the QA team has no KPIs of its own: it gets judged on the agents’ numbers.
What QA measures on agents
- QA score: the average scorecard result across evaluated contacts, per agent, team and account.
- Critical error rate: the share of evaluations with at least one error that fails the contact outright, such as a missed identity check. Many scorecards use an auto-fail rule for these.
- Compliance pass rate: the share of contacts that follow the mandatory steps for the client or the regulator.
- Score by criterion: where points are lost (tone, accuracy, process), which tells coaching where to start.
What the QA team is measured on
- Evaluation coverage: the share of contacts that were evaluated in the period, and how they were selected.
- Calibration variance: how far apart analysts’ scores are when they score the same contacts. A wide gap means the score depends on who did the review.
- Dispute rate and overturn rate: how often agents challenge a score, and how often the challenge wins. A high overturn rate points at unclear criteria or inconsistent scoring.
- Feedback turnaround: the time from the contact to the coaching conversation about it.
- Vendor and client score gap: the difference between the vendor’s QA score and the client’s own review of the same contacts.
How do you calculate a QA score in a BPO?
You calculate a QA score in a BPO by scoring each evaluated contact against the client’s scorecard, dividing the points earned by the points possible, and averaging the result across contacts. Critical errors are handled separately: on most scorecards that use them, the contact scores zero, and it also counts toward the critical error rate so it is not hidden in the average.
Here is a worked example with a simple weighted scorecard for a chat and email account. The weights are illustrative; every client sets its own, and our guide on how to weight QA scorecard criteria shows how to choose them.
| Criterion | Weight | Points earned on one chat |
|---|---|---|
| Greeting and identity check | 10 | 10 |
| Correct and complete resolution | 40 | 40 |
| Process followed (tags, internal note, correct macro) | 20 | 10 (no internal note) |
| Tone and empathy | 20 | 15 |
| Clear writing | 10 | 10 |
| Total | 100 | 85, so this chat scores 85% |
Now say the agent had shared account details before the identity check. If the client’s scorecard lists that as a critical error, the chat scores 0% whatever the other criteria earned.
Across a month, the arithmetic looks like this. If the QA team evaluated 200 contacts for an account, the QA score is the average of those 200 results. If 6 of them had a critical error, the critical error rate is 6 / 200 x 100 = 3%. Report both numbers. An average of 88% with a 3% critical error rate is a very different account from 88% with none.
What QA metrics belong in a BPO contract SLA?
The QA metrics that belong in a BPO contract SLA are the QA score and the critical error rate, each with a written definition. The SLA should also fix how the scores are produced: which rubric version applies, who scores, how many contacts are reviewed, how often both sides calibrate, how disputes are settled and what happens when a target is missed.
Quality disputes under an SLA are often less about the number than about how the number was produced. A clause that only says “QA score of at least X%” leaves every one of these questions open, so write them down.
- The rubric and its version. Attach the scorecard, the weights and the critical errors to the contract, and say how a change is agreed.
- Who scores. The vendor’s QA team, the client’s team, both, or an automated tool both sides can audit. If both score, say whose score counts for the SLA.
- Coverage and selection. How many contacts are evaluated per agent or per month, and how they are picked. A sample chosen by the party being measured is weak evidence; the statistics of small QA samples explain why.
- Calibration. How often client and vendor analysts score the same contacts together, and the maximum gap allowed. The steps to run calibration sessions work the same between companies as inside one.
- Critical errors as a separate target. Do not let them disappear into the average score.
- Disputes. How long agents and the vendor have to challenge a score, who decides, and whether overturned scores are recalculated. A written dispute process avoids month-end arguments.
- Reporting. The cadence, the level of detail (per agent, per queue, per client) and access to the evaluated contacts, not just the totals.
- Consequences. Service credits, an improvement plan, or an earn-back period. The target level is a commercial choice; set it after a baseline period on the agreed rubric, not before.
How a BPO can prove its KPIs to clients
A BPO proves its KPIs by showing the evidence behind them: which contacts were scored, against which client’s criteria, and why each score was given. A monthly spreadsheet of averages asks the client to trust the vendor. Scores the client can open and check turn the QA report into proof the vendor can use at renewal.
EverHelp, a BPO that has delivered more than 100 client projects, ran quality control on Google Sheets until every client’s standard became too much to enforce by hand. After moving to Kaizo, which scores each conversation against each project’s own criteria, its QA coverage rose 270% and tickets handled per hour rose 280% between 2023 and 2024, and coaching prep dropped 75% across sixteen domains. The full EverHelp story shows what changed when the quality checks moved off spreadsheets.
Score every client account against its own standard. Kaizo scores every conversation against each client’s own criteria, with the evidence attached to every result and reporting per client. Book a demo.
When a full KPI dashboard is more than you need
A full KPI dashboard is more than you need when an account is small, new or has no agreed quality standard yet. A team handling a few hundred tickets a month can track five KPIs in a shared sheet. An account without a written scorecard should write one before it buys any tool, because no software fixes an undefined target.
Be honest about the stage you are at:
- Fewer than a few hundred contacts a month: a manual sample on a shared rubric, reviewed with the client, is usually enough.
- No rubric yet: agree the criteria and critical errors with the client first. Measuring against an unwritten standard produces numbers nobody trusts.
- One client, one queue, stable volume: a helpdesk report with service level, FCR, CSAT and a sampled QA score covers most needs.
Automated QA starts to pay off when you run several clients with different standards, when the client questions your sampled scores, or when the QA team spends more time filling sheets than coaching. That is the point where teams look at Kaizo for BPOs.
Frequently asked questions
What is the full form of KPI in BPO?
KPI stands for key performance indicator. In a BPO it is a metric with an agreed target, such as a service level, handle time or QA score, that shows whether the outsourced team is delivering what the client contracted for.
What is KRA in BPO?
KRA stands for key result area, the part of a job a person is accountable for, such as quality for a QA analyst. KPIs measure how well each KRA is met. An agent might have productivity and quality as KRAs, measured by AHT and QA score.
How many KPIs should a BPO account track?
Track as many as you need to run the account, but keep the contract SLA to a short list both sides watch every month. Pair each speed KPI with a quality or resolution KPI so neither can be hit at the other’s expense.
What is a good QA score in a BPO?
There is no universal good QA score, because it depends on the scorecard, its weights and how strict the critical errors are. Set the target after a baseline period on the agreed rubric. Before you agree to a high target, check whether a 95% QA target is achievable with your own numbers.
Is AHT a fair KPI for BPO agents?
AHT is a useful planning number for staffing, but a risky individual target. Agents pushed on handle time alone tend to rush and close contacts that come back. If AHT is an agent KPI, pair it with FCR and the QA score.
How often are BPO KPIs reported to the client?
It depends on the contract. Weekly or monthly KPI reports to the client are common, often with a deeper quarterly business review (QBR). Speed KPIs are often watched daily or by the hour inside the BPO, because a bad morning can sink a monthly service level.
