Contact Centre Quality Assurance: Why Reviewing One Percent of Calls No Longer Works

Titelkarte: Qualitätssicherung im Contact Center – warum die Ein-Prozent-Stichprobe ausgedient hat

If you want to know how good your customer service really is, you listen in. In most contact centres that means a team leader picks out a handful of calls per agent each month, fills in a scorecard and talks the result through at the next coaching session. Measured against total contact volume, that amounts to roughly one percent of all conversations – frequently less. The remaining 99 percent are never reviewed.

The Trend Study Contact Center 2026, based on responses from 399 service professionals and managers across the German-speaking market, shows why this is now shifting. Speech and text analytics and data-driven decision support are spreading quickly, and 90 percent of respondents name customer satisfaction as their primary goal in service – ahead of pure contact deflection and cost reduction. Steering satisfaction, however, takes rather more than a handful of reviewed calls per month.

What a one-percent sample systematically misses

The weakness of traditional quality assurance is not the diligence of the reviewers but the statistics. Five calls a month is far too small a sample to support reliable conclusions – and the selection is rarely random. What gets reviewed tends to be whatever was close at hand, or whatever had already attracted attention.

  • Rare but expensive errors simply never surface in small samples – a missing mandatory disclosure, for instance, that turns into a complaint weeks later.
  • Patterns across agents stay invisible. If twelve people answer the same follow-up question incorrectly, that is not a performance issue but a knowledge issue.
  • Other channels are usually not assessed at all. Email, chat and messaging frequently run without any quality review whatsoever.
  • Selection bias: anyone who consciously or unconsciously picks the conspicuous calls is not measuring normal operations.

Full coverage means every conversation, every channel

The real difference in modern quality assurance is not better scorecards but coverage. AI-supported speech and text analytics transcribes calls automatically and evaluates them – alongside emails, chats and messaging threads – against the same criteria. A sample becomes a census.

That changes what quality assurance is for. Instead of isolated individual verdicts it produces a continuous distribution: where do deviations cluster, in which topics, at which times of day, in which teams? Conspicuous conversations are flagged and routed to a human reviewer, who then assesses not at random but where it actually pays off. In myContactCenter, telephony, chat, email, WhatsApp and ticketing already share a single interface, so evaluation does not stop at channel boundaries.

The scorecard is the real project

At the heart of any automated quality assurance sits the scorecard: the rule set every conversation is measured against. Typical criteria include greeting and identification, correct mandatory disclosures, needs clarification, solution quality, conversational control and closing. It sounds mundane, yet this is precisely where such projects succeed or fail.

A scorecard drawn up in isolation by the quality team ends up measuring formalities rather than outcomes. What works is a compact set of eight to twelve criteria, defined jointly with team leaders and experienced agents, followed by a calibration phase: the same conversations are scored manually and automatically, discrepancies are discussed and the rules refined. Only once both methods reach comparable results does the automated evaluation carry any weight.

Numbers have to turn into coaching

Full coverage generates a great deal of data very quickly – and that is exactly where the risk lies. A dashboard producing 40 metrics a month changes nothing on the service floor. Evaluation only becomes useful when it feeds a fixed rhythm: a recurring coaching session per agent covering two or three specific conversation excerpts, plus a team review for the patterns that affect everyone.

Feeding results back into other functions matters just as much. If follow-up questions about a particular tariff detail keep recurring, that belongs in the knowledge base rather than in individual coaching. If complaints about a process repeat, the responsible department needs to hear about it. Quality assurance that only ever points at agents wastes its greatest source of leverage.

Data protection and employee representation belong at the start

Call recording and analysis touch on employee data protection. In Germany, as soon as evaluations are capable of monitoring conduct or performance, the works council holds a co-determination right under Section 87 (1) no. 6 of the Works Constitution Act. Other EU member states apply comparable employee-representation and data-protection requirements, so the principle travels even where the specific statute does not. A works agreement setting out purpose, criteria, retention periods and access rights is therefore not an afterthought but the practical foundation of the project – and it reassures staff far more effectively than any announcement.

Proven practice: adopt the scorecard together with the works council, release evaluations initially only as team-level aggregates, and use individual results exclusively for coaching. Customers must also be informed that a call is being recorded. Settling these points early avoids the arguments that otherwise erupt mid-rollout.

Conclusion

The one-percent sample was never a good measuring instrument – it was simply what could be managed with reasonable effort. That constraint has gone. What matters now is not the technology but the preparation: a lean, jointly developed scorecard, an honest calibration, a fixed coaching rhythm and a clean agreement with the works council and data protection officer. Tackled in that order, the result is quality assurance that supports service teams rather than policing them.

Would you like to see how quality assurance across every channel could look in your own service operation? The specialists at ilogixx will show you how myContactCenter brings telephony, chat, email and messaging into a single evaluation. Simply book a no-obligation consultation via the online calendar at ilogixx.de.

Read more