Skip to content
EN

Telephony

Remote agents: a contact center in the browser

An agent working from home needs a computer with a browser, a headset, and a stable internet connection – calls run right in the browser, with nothing to install. To make the move without losses, start with a pilot group and agree in advance on what to do when the connection fails.

Pavel Ivanov, founder of CONTACT.WORK · · 7 min read

Remote agents: a contact center in the browser

Agents can work from home without a new system or software installed on home computers. But a remote contact center only works well if you plan equipment, network, security, and quality assurance in advance. Here is what an agent needs at home, how browser-based calling works, and how to move your team without hurting service level.

What an agent needs at home

  • A computer with a modern, up-to-date browser.
  • A headset – ideally wired, with a noise-canceling microphone.
  • A stable internet connection – over Ethernet rather than Wi-Fi when possible.
  • A quiet spot with no TV or other conversations in the background.
  • A backup connection, such as a mobile hotspot, in case the internet provider goes down.

There is no softphone to install: agents make and take calls in the web app. Onboarding a new hire is simple – create an account and test the headset.

The headset is the most common source of audio complaints. A laptop's built-in mic picks up household noise and echo from the speakers, and a wireless headset can run out of battery mid-shift or drop the connection. Give everyone the same tested model – it makes troubleshooting and replacing equipment easier.

How browser-based calling works

Calls run over WebRTC, a technology built into modern browsers, so there is nothing to install. Voice and signaling are encrypted (using SRTP, TLS, and DTLS) all the way from the browser to the server. Every call can be recorded and attached to the customer record in your CRM.

The agent signs in to the web app with their own account, allows the browser to use the microphone, and the workspace is ready. The same window holds website chat and messaging inquiries, the knowledge base, and AI assistant suggestions. Access is role-based: agents see what they need for their work, and managers see their team's reports and recordings.

The softphone has a call quality indicator. Agents see the connection getting worse before the customer starts asking them to repeat themselves, and can switch to the backup connection.

Network requirements

For voice, internet stability matters more than speed. A call needs little bandwidth, but it handles delays, jitter, and data loss poorly. The customer hears it as echo, clipped words, and a robotic voice.

Why this happens. The browser cuts voice into small packets – usually 20 milliseconds each, about 50 per second – and sends them over the network. Lose a packet, and a syllable disappears. When packets arrive unevenly, the browser waits for the stragglers, and the delay grows. Once one-way delay goes above about 150 milliseconds, people start talking over each other. Bandwidth barely matters here: a call needs about a tenth of a megabit, but a spike from a video stream or a backup is audible right away.

  • Connect the computer over Ethernet: home Wi-Fi often causes delay spikes.
  • Keep the connection free during shifts: HD video, large uploads, and cloud backups interfere with voice.
  • If agents work through a corporate VPN, test calls over that VPN – it can add delay.
  • Make sure the home router and corporate network rules don't block WebRTC traffic.
  • Before the first shift, make test calls and listen to the recordings – that way you catch problems before customer calls.

Call quality often drops only at certain hours – for example, in the evening, when home networks are busy. One good test call is not enough: have agents note the times customers complained they couldn't hear, and compare them with the indicator and the recordings.

Poor call quality costs money even if no call drops. The customer asks the agent to repeat, the agent repeats, and the call gets longer. A line with 120 calls an hour, a handle time of 240 seconds, and an 80/20 target needs 11 agents. If repeating adds just 15 seconds to each call, the Erlang C formula says you need 12 agents, and with the same 11, service level falls from 81% to 73%. For a line open 9:00 AM-9:00 PM, seven days a week, that means 51 people on staff instead of 45 – six extra salaries every month. Run your own numbers in the Erlang C calculator.

Cloud, on-premises, or hybrid

For a distributed team, the cloud is simplest: we run the platform, setup takes 1 day, and you add seats as you grow without buying licenses. If your data has to stay with you, we install the platform on your servers in 2-7 days, with the same features. The server calculator shows how many servers you'll need.

There is also hybrid deployment: some components in the cloud, some on your side. The choice depends on your data storage requirements and on who will maintain the infrastructure. You can move from the cloud to your own servers later without changing platforms.

Customer records and inquiry history open in the browser, so agents never need to save customer data to a home computer. That makes it easier to meet data protection requirements, such as the CCPA/CPRA in California and other state privacy laws.

How managers run a remote team

Remote and in-office agents work in the same queues and show up in the same reports. Managers run one team, not two.

  • Real-time reports: queues, agent statuses, service level, and missed calls.
  • Speech analytics: remote agents' call quality is scored the same way as in-office agents'.
  • Workforce management (WFM): schedules, breaks, and schedule adherence – especially important when you can't see the team in the office.
  • Knowledge base and suggested replies: they replace asking the person at the next desk, which no one can do at home.
  • BI export: remote team data goes into the same analytics as office data.

Plan feedback separately. In the office, a supervisor hears a call and walks over right away; at home, the agent is on their own. Regular call reviews, short video check-ins, and clear scoring criteria help keep one standard.

Common mistakes

  • Moving the whole team in one day without a pilot group.
  • Allowing Wi-Fi and the laptop's built-in mic – customers hear echo and noise.
  • Not planning for outages: an agent drops offline, but calls keep routing to them.
  • Judging remote agents only by call count, with no quality review.
  • Forgetting new hires: at home, there's no one to ask, so the knowledge base has to stay current.

How to measure results

Compare remote and office groups on the same metrics over the same intervals. If the remote group falls behind, find out where: connection, schedule, or knowledge. Each cause has its own fix, and you usually don't have to give up remote work.

  • Service level and abandonment rate by queue.
  • Average handle time (AHT) and after-call work time.
  • Quality scores from speech analytics and the share of calls with complaints about audio.
  • Schedule adherence and the share of time in productive statuses.
  • Turnover and how fast new hires reach target metrics.

Where to start: a transition checklist

  1. Decide who goes first: start with a pilot group of experienced agents.
  2. Check each agent's computer, browser, headset, and internet, and record their backup connection.
  3. Create accounts with the right roles and check access to the CRM and knowledge base.
  4. Make test calls and listen to the recordings together with the agent.
  5. Agree on the rules: statuses, breaks, how to reach a manager fast, and what to do when the connection fails – the agent switches to the backup connection or tells their manager, and their status changes so calls go to teammates.
  6. Set up reports and schedules so you see remote and in-office agents together.
  7. Compare the pilot group's metrics with the office, and only then move everyone else. If the pilot group's talk time went up, calculate what it costs, as in the example above, and fix the cause first.

Moving agents to remote work?

Tell us how many agents you have and where your data is stored – we'll show you browser-based calling and reports for a distributed team.

Discuss your project

Newsletter

Get new articles by email

Articles on contact center operations, AI agents, and quality assurance.