Call Monitoring

Live monitoring modes

Most contact center platforms offer three live modes, usually from the supervisor's dashboard:

• Silent monitoring. The supervisor listens without the agent or customer hearing. Used for spot checks, compliance and assessing new agents.
• Whisper coaching. The supervisor speaks to the agent, and the customer cannot hear. Useful for prompting a step the agent missed or helping with a tricky policy question in the moment.
• Barge-in. The supervisor joins the call so both parties hear them, or takes it over. Reserved for escalations, serious errors and abusive callers.

Whisper works best in short, specific prompts ("offer the payment plan"). Long whispered instructions while the customer is talking overload the agent. Barging should be rare; done often, it undermines agents in front of customers.

Recorded review and speech analytics

Most monitoring happens after the fact. Calls are recorded, a sample is pulled and QA analysts score them on the quality form. That feeds quality assurance reporting and call coaching sessions.

Manual sampling has a limit: a QA analyst can score perhaps 25 to 40 calls a day, so most centers review only 2 to 5 calls per agent per month, a tiny share of the total. Speech analytics changes that by transcribing every call and flagging those with signs of trouble, such as cancellation language, raised voices, long silences or missed disclosures. Analysts then spend time on calls that matter rather than random ones, and leaders see trends across thousands of calls.

How much to monitor

A common starting point is four to eight evaluated calls per agent per month, more during nesting and for agents on a performance plan, fewer for consistently strong performers. For a 200-agent center at six evaluations each, that is 1,200 evaluations a month. At about 30 evaluations per analyst per day across 21 working days, that needs roughly two full-time QA analysts, before calibration, reporting and coaching time. Most centers plan one QA analyst per 30 to 50 agents.

Choose calls with a purpose. A mix of random calls (for a fair score), targeted calls (long handle time, repeat callers, low survey scores) and calls on new processes gives a far better picture than random sampling alone.

Legal rules for monitoring and recording

In the US, the federal Wiretap Act, part of the Electronic Communications Privacy Act of 1986, generally allows recording when one party to the call consents, and it includes a business extension exception for monitoring in the ordinary course of business. About a dozen states, including California, Florida, Pennsylvania and Washington, require consent from all parties. That is why most centers play a "this call may be monitored or recorded" message and treat continued participation as consent.

Other obligations apply on top. PCI DSS restricts storing card verification codes, so centers pause recording or mask audio during card capture. HIPAA covers recordings with health information. Agents should be told in writing that their calls are monitored. Monitoring personal calls and broader employee surveillance raise separate issues covered under employee monitoring. Rules vary by state and country, and GDPR sets stricter requirements in Europe, so confirm the requirements for every location where agents or customers are.

Making monitoring feel fair to agents

Agents accept monitoring when they see it used to help them, and resent it when it only produces penalties. Practical habits: share the quality form and the selection method openly, let agents hear their own recordings, include good calls in coaching, run regular call calibration so scores are consistent, and allow a simple dispute process. Avoid using live monitoring as surveillance of breaks and idle time; workforce management reports already show adherence.