Skip to content
Back to blog
10 min readBy The dialque Team

Mobile phone dialer for cloud contact centres: when agent phone mode beats the browser

We shipped mobile-phone dialing because field-lending officers, insurance surveyors, and small tier-3 teams told us their agents work off a phone, not a desk. Here's how the mode works, when it beats the browser softphone, what breaks (predictive autodial), what doesn't (warm transfer, dispositions), and a concrete cost breakdown for 10 field agents on BYOC.

Product updateCloud dialerField operationsCost

We shipped mobile-phone dialing this week because the pattern kept repeating in customer calls: "our agents don't sit at a desk." Home-loan field officers driving between Andheri and Vashi. Insurance surveyors doing spot inspections in Meerut. A three-person collections team in Nagpur that never bought USB headsets. A rural micro-lending desk where BSNL fiber drops for four hours every afternoon and WebRTC is a coin toss. For all of them, the browser softphone was the wrong shape. So we made the agent's mobile phone a first-class endpoint. The platform now dials the agent's number through the trunk, bridges the customer in, and the agent runs the rest of the call — status, disposition, notes, campaign progress — from the web app on whatever screen they have. Only the audio moves to the phone.

This post is a straight walk-through of when to use it, when not to, how the switch works, and what it costs.

Why the browser softphone isn't always the right answer

WebRTC is our default for a reason. Audio quality is better, DTMF is more reliable, recording is cleaner, autodial pace stays tight, and per-minute cost is lower because you're only billing one leg. For an office-based inside-sales team on a wired 20 Mbps line with a decent headset, WebRTC wins every category.

But India's contact centre workload is not uniformly office-based. Rough segmentation from our own book of business:

| Team shape | Share of our tenants | Practical constraint | |---|---|---| | Office inside-sales / support | ~55% | None — WebRTC is ideal | | Hybrid (some at desk, some in field) | ~25% | Field cohort needs mobile | | Fully field (surveyors, DSAs, recovery officers) | ~12% | Cannot use WebRTC | | Small teams (<10 agents) with no AV budget | ~8% | Would rather use own phone |

That's roughly 45% of our tenants for whom "just open the Dialer tab" is either impossible or annoying enough that agents start using WhatsApp and personal phones instead — which then breaks DLT hygiene, kills recording coverage, and makes RBI FPC audits ugly.

Mobile mode fixes that without breaking the parts of the platform teams already rely on.

What mobile mode actually does

The mechanics are boring, which is the point.

  1. Agent logs into the dialque web app as usual. Status, campaign, disposition tree — same UI.
  2. Agent clicks a lead, or a predictive miss lands, or an IVR routes an inbound call to them.
  3. Platform originates the call through the trunk — a leg to the agent's registered mobile number and a leg to the customer — and bridges the two.
  4. Agent's phone rings. They answer. Customer hears them.
  5. Agent taps the disposition in the browser when done. Recording, timers, wrap-up all land in the CRM the same way.

Applies to manual dial, Click-to-Call from the CRM, inbound queues, and IVR transfers. Agents without a Phone Number filled in on their profile silently stay on the browser softphone even when the tenant is in Phone mode — which is the safety net for the mixed-shape team above.

The one thing that does not move to mobile is predictive and progressive autodial. That stays on the browser. The reason is timing: autodial pace assumes a sub-500 ms answer bridge, and originating a leg to an Indian mobile network adds 3–7 seconds of ring + PDD (post-dial delay). Push that in front of a predictive engine and pacing collapses — agents get customer calls before they've finished the previous wrap, or worse, the customer says "hello" to silence for four seconds. So we don't try.

Five scenarios where mobile beats WebRTC

1. Field-lending officers on the road. A DSA doing four unsecured-loan pickups a day in Pune cannot lug a laptop into a customer's living room. But they can take the branch's follow-up calls from the Bolt cab between visits. On mobile mode, they answer the ring, the customer hears them, and the call is recorded and disposed like any other. RBI FPC coverage stays intact because the platform trunk carried the leg.

2. Insurance spot inspections and claim visits. Surveyors at accident sites or hospital wards need to answer the underwriter, not fumble with a Chrome tab on tethered 4G. Mobile mode also solves the compliance angle: the surveyor's personal phone number never leaks to the claimant, because the customer sees the tenant's DLT-registered CLI.

3. Tier-3 city offices on flaky ISP fiber. BSNL fiber outage patterns in a small-town branch are predictable — 2 pm to 4 pm, most Wednesdays. Airtel 4G in the same building is fine. Mobile mode lets a branch keep dialing through the outage window instead of losing 30 productive agent-hours a month.

4. Small teams that will not buy headsets. A five-agent recovery team billing on Starter (₹1,500/agent) genuinely does not want to spend ₹15,000 on USB headsets before they've closed their first month of collections. Mobile mode ships them working from day one on the phones they already own. When they scale, they buy headsets and flip the tenant setting.

5. Legacy Windows machines that choke on WebRTC. A large NBFC we onboarded last quarter still runs 2 GB RAM Windows 10 desktops in tier-2 branches. Chrome + Opus codec + a call recording upstream on those boxes wedges the machine within an hour. Mobile mode moves the codec load off the desktop entirely; the browser only carries UI traffic, which is ~40 KB/s. Same hardware, no wedge.

A sixth honorable mention: agent onboarding week. New joiners are still learning the CRM. Letting them take the first week's calls on their phone while they learn the UI cuts training friction — no "why can't the customer hear me" mic-permission moments.

What breaks, what doesn't

Straight table, no hedging:

| Feature | WebRTC | Mobile mode | |---|---|---| | Manual dial | Yes | Yes | | Click-to-Call from CRM | Yes | Yes | | Inbound queue / ACD | Yes | Yes | | IVR transfer to agent | Yes | Yes | | Warm transfer to another agent | Yes | Yes (both agents on their phones) | | Cold transfer | Yes | Yes | | Conference (3-way) | Yes | Yes | | DTMF (IVR navigation, OTP entry) | Yes, in-band | Yes, via trunk RFC 2833 | | Call recording (dual-channel) | Yes | Yes (mono; recorded at the trunk bridge) | | Whisper coaching (supervisor) | Yes | Yes | | Barge-in | Yes | Yes | | Predictive autodial | Yes | No — falls back to browser for that agent | | Progressive autodial | Yes | No — same fallback | | Preview dialing | Yes | Yes | | Live disposition + notes | Yes | Yes (in browser) | | Post-call survey (IVR) | Yes | Yes |

The one hard "no" is predictive/progressive. If a tenant on Phone mode starts a predictive campaign, the platform silently routes those specific calls through the browser softphone for the campaign's duration. Agents will need a headset for that campaign window; everything else stays on their phone.

Configuration — tenant switch and per-agent exemption

Two levers, both live in System Settings → Agent Endpoint:

  • Tenant default. Browser (WebRTC) or Phone. This is where you set the mode for the whole account. Changes take effect on the agent's next login — active calls are not interrupted.
  • Per-agent Phone Number field. Set on the agent's profile. Any agent with a blank Phone Number stays on the browser softphone regardless of tenant setting. That's how you run a mixed shape: tenant set to Phone, office-based supervisors and predictive-campaign agents left with a blank number.

For teams that want a scheduled switch — mobile during outage windows, browser at other times — we expose the tenant setting via API so you can script it. Most teams don't need this; they pick one mode and stick with it.

A concrete cost example: 10 field agents, 200 minutes/day, BYOC

The math changes with mobile mode because the platform is bridging two legs — one to the agent's phone, one to the customer. On our own trunks (Tata, Airtel wholesale) we bill both legs, which we call it out as 2× minutes in the pricing page. On BYOC (bring-your-own-carrier), you pay your own carrier for both legs at whatever your negotiated rate is, and dialque bills nothing for telecom.

Take a 10-agent field-lending team on Growth (₹2,000/agent/month), each doing 200 minutes of talk time/day, 22 working days.

  • Customer-side minutes: 10 × 200 × 22 = 44,000 min/month
  • Agent-side minutes (mobile leg): same volume = 44,000 min/month
  • Total carrier minutes: 88,000 min/month

| Deployment | Platform licence | Telecom cost | Monthly total | |---|---|---|---| | Mobile mode, dialque trunk (Tata @ ₹0.70/min, 2× billing) | ₹20,000 | 44,000 × 2 × ₹0.70 = ₹61,600 | ₹81,600 | | Mobile mode, BYOC (Airtel enterprise SIP @ ₹0.45/min flat) | ₹20,000 | 88,000 × ₹0.45 = ₹39,600 | ₹59,600 | | Hypothetical WebRTC on same team (if they had headsets + wired net) | ₹20,000 | 44,000 × ₹0.70 = ₹30,800 | ₹50,800 |

Two things fall out of that table. First, BYOC saves this team ₹22,000/month over the platform trunk — enough to pay for the Enterprise-tier upgrade (₹2,500/agent) if they want the extra recording retention. Second, the "if only they had headsets" WebRTC number is roughly ₹9,000/month cheaper than BYOC mobile — but that gap disappears the moment you count the productive hours lost to a single BSNL outage or a laptop wedge, let alone the DSA who literally cannot sit at a desk.

When to stay on WebRTC anyway

The honest list:

  • Any predictive or progressive dialer campaign, always.
  • Inside-sales teams closing on price-sensitive SKUs where a 200 ms audio quality difference actually loses deals.
  • Any team where supervisors depend heavily on barge/whisper — quality is measurably better on WebRTC.
  • Multi-language support desks routing between Hindi, Tamil, Telugu, Bengali queues where lower latency helps agents catch language cues faster.
  • Regulated flows (KYC verification, RBI-mandated recorded consent) where dual-channel recording is a legal requirement.

If your team doesn't hit any of the above and half of them work off a phone anyway — flip the tenant switch, let the office agents keep blank Phone Numbers, and stop pretending everyone works at a desk.

FAQ

Does mobile mode work on 4G? Yes, and it's the point. The agent leg is a normal PSTN call on their carrier — Jio, Airtel, Vi, BSNL. As long as they have a voice signal (not just data), the call rings and connects. We don't care whether the phone is on 4G, 5G, 3G fallback, or a 2G edge network — we're only originating a voice leg.

Does it work over WhatsApp calls? No. Mobile mode originates a regular PSTN call to the agent's number. WhatsApp is a data-side VoIP channel and doesn't accept originated calls from a SIP gateway. If you want the agent to take calls in WhatsApp, that's a different product — a WhatsApp Business API bridge — and it comes with its own DLT and consent constraints.

Can I mix modes across agents in one tenant? Yes. Set the tenant default to whichever mode covers the majority. Then either fill in or blank out the Phone Number field on individual agent profiles to exempt them. A blank Phone Number always means "use the browser softphone for this agent," regardless of tenant setting.

Does mobile mode count against DLT / TCCCPR compliance the same way? Yes. The customer-facing CLI is still your DLT-registered header, the call is still logged against your registered template, and NDNC scrubbing still runs at the trunk. The agent leg is invisible to the customer — they never see the agent's mobile number.

What happens to recordings? Recorded at the trunk bridge, same as WebRTC. On Starter you get 3 months retention, Growth 6 months, Enterprise 9 months. Mobile-mode recordings are mono (single channel from the bridge) rather than the dual-channel stereo we produce on WebRTC — supervisors can still use them for QA but noise-suppression scoring is less accurate.

If my agent's phone battery dies mid-call, what happens? The customer leg stays up for our normal abandon-call window (default 8 seconds), then drops with a "dropped by agent side" disposition auto-logged. Same behavior as an office agent unplugging their headset.

Want to see it running with your own trunk and a couple of test agents? [Book a 20-minute walkthrough](/contact?source=demo&topic=mobile-phone-dialer-when-to-use-agent-phone-mode) — we'll set up a sandbox on your DLT header and let your field team try it in a live pilot before you flip the tenant setting.