In today’s call and contact centers, agents are caught in the middle of a quiet – but constant – tug-of-war.
Leadership wants faster calls, tighter metrics, and improved efficiency. Customers want patience, empathy, and solutions. Agents are expected to deliver both: all while navigating complex systems, expanding queues, and performance metrics that rarely reflect the realities of the work.
But the issue isn’t that agents are spending too much time with customers. The real problem is where that time is being spent.
In this three-part series, I’ve explored how contact centers can reduce average handle time (AHT) without rushing either the customer or the agent.
- Part 1 focused on removing customer friction before the call.
- Part 2 examined agent performance strategies and in-call support.
- Part 3 (this article) looks at staffing, tools, and leadership choices, and brings the conversation to a close.
Staffing Strategy
Having the right number of agents with the right mix of skills plays a critical role in both handle time and customer experience (CX). As more centers move to hybrid or fully remote models, the talent pool has expanded.
1. Attitude often matters more
When hiring, it is almost always more important to select applicants who have the right attitudes than the best technical skills.
Agents who can remain calm, curious, and customer-focused during high call volumes, system outages, and/or emotionally charged interactions are far better equipped to deliver consistent CXs compared with the most highly skilled agents.
Tools and processes can support agents. But the wrong hiring decisions will undermine the best-designed operation.
2. Language needs
Many U.S. customers, and certainly those calling from other countries, speak or are more comfortable in speaking languages other than English, notably Spanish.
Spanish is a common enough skill set, but there may be insufficient numbers of agents who can speak it in the call center.
While interpretation services for this language and others are a valuable investment in accessibility to serve customers with different needs, they frequently double – or even triple – call lengths. That is due to the time required to locate and brief the interpreters and for call interpretation.
It’s not realistic to staff for every language your customers speak. However, centers can analyze call data to identify the top two/three non-English languages and hire multilingual agents accordingly.
When customers can communicate directly with agents in their preferred language, calls move faster, misunderstandings are reduced, and resolution rates improve.
While automated AI-based interpretation and real-time applications show promise and will continue to improve, they still struggle with accents, emotion, industry-specific terminology, and poor audio quality.
For now, they should be viewed as assistive tools: not replacements for multiingual talent in complex interactions, such as explaining why a medical claim won’t pay or the purpose of the charges on a person’s credit card, to name a few.
3. Accounting for demand fluctuations is vital
Even the strongest workforce management (WFM) teams can’t predict every spike in volume caused by product launches, billing cycles, outages, or seasonal trends.
Building flexibility into the staffing model – through part-time agents, temporary staff, or vetted gig workers – helps centers scale up without overcommitting long term.
Tools and processes can support agents. But the wrong hiring decisions will undermine the best-designed operation.
A flexible staffing model can help organizations respond to unpredictable call volumes without compromising service levels. I discuss these strategies more extensively in “Tapping Into Staffing Alternatives”.
I believe there is a balance to strike. Overstaffing leads to disengagement and inefficiency. Understaffing is far worse: driving longer queues, higher handle times, rushed interactions, and agent burnout.
Sustainable efficiency comes from staffing to demand while preserving the agent’s ability to focus, think, and resolve issues well.
AI Note-Taking and Call Summaries
Automated transcription and summarization can save agents valuable wrap-up time and improve documentation quality.
However, these tools are not without risk. When AI mishears words, misunderstands context, is affected by background noise or a poor phone connection, or generates incomplete summaries, agents and QA teams are left correcting errors after the fact.
Here are two posts that caught my eye, and I decided to share:
- “The customer said they were ‘upset about late fees.’” The AI summary said they were ‘excited about fee opportunities.’”
- “Customer name: ‘Juan Martinez.’ AI notes: ‘One Martian is calling claim filing.’”
Used responsibly, AI note-taking reduces cognitive load. But if used carelessly, it creates downstream work. Further, I would suggest using such tools to help agents but ensure they are always able to review, edit, and correct AI-generated notes.
Real-Time Prompts and Whispers
Guided AI-driven real-time prompts and agent whispers hold real promise: especially for newer agents or complex, high-risk interactions.
In theory, having guidance appear at the right moment can reduce hesitation, prevent errors, and shorten handle times. But in practice, agent feedback tells a more nuanced story.
Across online agent communities, one theme repeatedly emerges: timing matters.
When prompts appear at the wrong moment – mid-sentence or during active listening – they don’t help. They distract.
As one agent humorously put it, “it can feel like being coached in real time while already on stage.”
The challenge isn’t the idea of prompts: it’s how they’re delivered.
Used thoughtfully, prompts and whispers can be valuable, by:
- Guiding newer agents through unfamiliar scenarios.
- Supporting complex or regulated interactions.
- Bringing up reminders that prevent compliance issues.
But prompts and whispers get in the way when they:
- Appear too frequently or at the wrong moments.
- State the obvious (“Acknowledge customer frustration”).
- Conflict with policy or performance expectations.
- Remove the agents’ sense of control.
The most effective implementations treat AI prompts and whispers as assistive guides, not as supervisors micro-managing agents. In that use they:
- Allow agents to mute or turn off prompts when needed.
- Adjust prompt frequencies and volume based on experience level.
- Focus on clarity, not constant instruction.
The goal isn’t to tell agents how to think: it’s to give them the information they need faster, while also providing gentle nudges to improve call quality.
Conclusion: AHT Is a Measure: Not the Mission
AHT should never be viewed in isolation. When leaders focus on a single metric, something else almost always pays the price. A narrow, laser-focused push to reduce AHT can unintentionally lower FCR, drive repeat calls, and ultimately damage CSAT. Faster calls don’t always mean better calls.
The real opportunity lies in looking at performance holistically. Balanced metric analysis is essential when improving AHT.
In “Using Data to Improve Performance”, I discuss how data interpretation can uncover hidden drivers behind service inefficiencies. Because when metrics are reviewed together – and supported by clean, reliable data – patterns begin to emerge.
Building flexibility into the staffing model…helps centers scale up without overcommitting long term.
Analytics tools like Power BI can help visualize trends, while AI can quickly and cost-effectively identify recurring issues or friction points. Outliers and data errors – such as days when AHT spikes to unrealistic levels – must be addressed first.
For organizations with the budget, an external consultant can provide a valuable, objective review and uncover blind spots internal teams may miss.
And when used correctly, data becomes a powerful driver of improvement. It can:
- Highlight frequently asked questions that belong in FAQs, scripts, or self-service channels.
- Reveal gaps in training.
- Inform better knowledge article creation.
- Strengthen AI chatbot responses.
- Improve WFM forecasts.
- And even influence smarter hiring decisions.
In short, data should guide design: not just for reporting purposes.
Coaching also plays a critical role in sustainably improving handle time. I’ve seen firsthand how minor, personalized adjustments can make a meaningful difference.
In one case, an agent struggling with AHT wasn’t lacking skill: she needed better organization. By helping her prep systems in advance, use her monitors more effectively, and organize her most-used resources, her handle time improved naturally.
…AHT doesn’t need to be forced; it improves naturally as a byproduct of doing the work better.
In another case, an agent needed support with call control when working with particularly talkative callers. With targeted coaching, practical techniques, and thoughtful short videos and team chat tips, she didn’t just improve; she became a resource for her colleagues.
These experiences reinforce an essential truth: AHT improves when agents are supported, not pressured.
Ultimately, reducing handle time isn’t about pushing agents to work faster. It’s about removing friction, designing smarter processes, providing the right tools, and coaching with intention.
When centers strike the right balance between efficiency, CX and agent experience (AX), AHT doesn’t need to be forced; it improves naturally as a byproduct of doing the work better.
Efficiency and empathy are not opposites. When done right, they move forward together.
