← All articles
Quality

One Quality Team, Many Clients: Keeping Scores Consistent Across a BPO

One quality team, many clients

A BPO’s quality team does not run one quality programme. It runs one for every client, each with its own scorecard, its own definition of a good call and its own person on the client side who will question the results.

That is a harder job than quality in a single-brand contact centre, and it is the reason the usual approach, listening to a small sample of calls, breaks down faster in a BPO than anywhere else.

Why sampling breaks first in a BPO

A manual team can review 1–3% of calls. In a single-brand centre that sample is at least drawn from one pool. In a BPO it is divided across every client and every project, so the share that lands on any one account is small, and the share that lands on any one agent is smaller still.

The result is familiar. An agent on a small account can go a long time between reviews. A client asks why complaints rose last month and the honest answer is that nobody heard most of those calls. And when a client disputes a score, there are too few scored calls to show a pattern either way.

One scorecard for each client, one way of scoring

The fix is not a single scorecard for the whole floor. Clients differ, and their scorecards should too. What has to be the same is the way a score is reached.

In Intella Studio, Auto-QA scores every call against the scorecard of the client it belongs to. Each check is marked against the transcript and cites the line it rests on, so a reviewer, an agent or a client can see why a check passed or failed without listening to the whole call.

Calibration is where trust is built

Scoring every call only helps if people believe the scores. In a BPO, three parties have to agree on what a good call sounds like: the automatic scoring, your own reviewers, and the client.

Between Auto-QA and your reviewers

Have reviewers score a set of calls that Auto-QA has already scored, then look at where the two differ. Most disagreements trace back to a check that is worded loosely. Fix the wording of the check, not the individual score.

With the client

Invite the client to score the same calls and compare the results. A client who has taken part in calibration argues far less about the monthly quality number, because they helped define it.

What reviewers do when every call is scored

Automatic scoring does not replace the quality team. It changes what the team spends its time on.

  • Sampling plans. Decide how many calls each reviewer checks for each client and project, instead of leaving it to whoever has time.
  • The calls worth a second look. Send reviewers to the calls where a critical check failed, where the score is close to the line, or where an agent has disputed the result.
  • Disputes. Give agents a way to challenge a score, and reviewers a way to confirm or override it. A score that can be challenged is one people accept.

From a score to a coaching plan

A score is only useful if it changes what an agent does next. With every call scored, coaching can rest on an agent’s own calls: what they do well, the one or two things to work on, and whether last month’s focus area has improved.

The same data shows how quickly new hires reach your quality standard on each account, which matters when a client asks you to add seats at short notice.

What to ask of any quality tool

  • Can each client have its own scorecard?
  • Does every check show the line in the transcript it rests on?
  • Can reviewers and clients calibrate against the automatic scores?
  • Can an agent dispute a score, and is the outcome recorded?
  • Is the transcript accurate in the dialects your callers actually speak?

The last question comes first in practice. A check marked against the wrong words is a dispute waiting to happen.

See how Quality works in Intella Studio, or start a Free Pilot on sample calls and then bring your own.

← All articles Start your Free Pilot