Remote Team Metrics Drift After the Move Home
Harborline Health Supplies moved its 80-agent blended contact center to remote work in March. Adherence fell from 91% to 76%. Average handle time dropped 40 seconds, but transfer rate rose 9 points and first-contact resolution fell 6 points. Supervisors see more short calls and more customer-hung-up dispositions. Agents use personal headsets; recordings from about 20% of the team carry background noise that triggers speech-analytics errors. One supervisor proposes webcam checks every two hours and random screen sharing. Two agents live in states that require all-party consent for recording; the company's recording disclosure plays at the start of each call. Attrition risk is high: 14 agents have quit in eight weeks, and the engagement survey shows the lowest scores on I know what is expected of me. Overtime is up 18%.
1. Is this shirking, system friction, or both? Lay out the evidence you would gather and how you would interpret it.
Read the worked answer
The pattern is ambiguous and should not be read as shirking by default. Short calls plus rising transfers plus falling first-contact resolution can mean agents are dumping difficult calls, or it can mean agents lack the side-by-side support and knowledge tools they had in the office and are transferring to compensate. Both produce the same surface metrics.
Evidence to gather, split by tenure and by home setup:
- Compare the remote period against a matched pre-March baseline for the same agents, not against the whole prior year.
- Sample 50 calls under 90 seconds and 50 transfers. For each, code the reason: wrong number, customer rushed, agent lacked information, escalation required, or clear avoidance.
- Pull transfer destinations and repeat-contact rates within seven days. If transferred customers call back, quality is the problem, not effort.
- Check hold time, system latency and login failures, which often masquerade as adherence loss.
- Review the 20% of recordings with audio problems separately; if their scores are distorted, exclude them from quality comparisons.
Interpretation rule: a broad, tenure-spanning pattern with high repeat contacts points to friction. A pattern concentrated in a few agents, with short calls and no repeat contacts, points to avoidance. Manage those two groups differently, and do not roll out surveillance for a problem that may be equipment and knowledge gaps.
2. Evaluate the webcam and screen-sharing proposal. What monitoring governance would you put in place instead?
Read the worked answer
Reject webcam checks every two hours. They are disproportionate to the problem, they will accelerate the attrition already visible, and they measure presence rather than performance. The engagement score on knowing what is expected is a clearer signal: agents are uncertain about standards, and surveillance deepens that.
Governance to put in place instead:
- Legality: keep the recording disclosure at the start of each call, confirm the two all-party-consent states are covered, and get counsel sign-off before any new monitoring. Do not record personal devices or home environments without explicit, narrow consent.
- Proportionality: monitor work outputs and systems, not people. Adherence, CRM activity, call outcomes and quality samples are sufficient.
- Transparency: publish what is measured, how, and what is not. Give agents access to their own dashboards.
- Equipment: fund approved headsets and a small connectivity stipend. The 20% audio failure rate is a solvable technical issue and is distorting analytics.
- Coaching: replace random screen sharing with scheduled side-by-side sessions requested by the agent or triggered by a metric threshold, and a weekly 1:1.
- Adherence flexibility: build a formal exceptions process for home interruptions rather than treating every deviation as misconduct.
If targeted screen sharing is used at all, limit it to new hires in a defined onboarding window with advance notice and consent.
3. Write a 60-day improvement plan with targets. What would you fix first, and what would you not promise?
Read the worked answer
Days 1-14, fix the instruments. Issue approved headsets to the 20% with audio problems, test connectivity, and clean the disposition taxonomy so customer-hung-up is distinguishable from system drop and wrong number. Publish role expectations in one page: what good looks like on handle time, transfer, resolution and adherence. Without clean data, every later target is guesswork.
Days 15-30, attack transfers. Code transfer reasons, identify the top three drivers, and run short knowledge refreshers on them. Review 50 short calls and give individual coaching. Set a target of transfers down 4 points from the current level and first-contact resolution up 3 points.
Days 31-45, rebuild adherence. Move to 85% with an exceptions policy, visible schedules, and weekly coaching rather than discipline for first offenses. Track adherence by tenure; new hires usually need schedule structure more than pressure.
Days 46-60, stabilize. Target attrition below 8% annualized for the period and overtime down 10%. Keep handle time within 20 seconds of the pre-March baseline rather than pushing it down, because the current drop is probably a symptom, not a win.
What I would not promise: a return to 91% adherence in 60 days, or that resolution recovers fully while the team is still remote. Those depend on tooling and staffing. Report progress weekly and flag the honest constraint.