Cloud Telephony Solutions India: What You Actually Pay Per Seat

The ₹18,000-a-month bill that should have been ₹6,400

A 42-person logistics firm in Peenya called us in February 2026 because their cloud telephony bill had crept from ₹9,200 to ₹18,400 a month over eleven months. Nobody had changed anything deliberate. What had happened was ordinary: two sales hires, a customer support line added for a new contract, and a per-seat plan that charged ₹399 per user per month regardless of whether the seat made four calls or four hundred.

We pulled their call detail records. Twenty-nine of their forty-two provisioned seats had not placed a single outbound call in the previous ninety days. They were paying for voicemail boxes and a dialler login that nobody used. We cut them to fourteen seats, moved outbound sales to a per-minute trunk, and their monthly cost settled at ₹6,380 with better call quality than before. Same carrier. Same numbers. Different commercial model.

That story repeats itself across Bangalore and every other Indian city, because cloud telephony in India is sold almost entirely on per-seat pricing while a large chunk of Indian SME call traffic is outbound, bursty, and seasonal. If you are evaluating cloud telephony solutions right now, the decision you are actually making is not "cloud versus PRI". It is "which pricing model matches my call pattern", followed by "can I get through DLT registration without losing a week".

This article works through the money, the compliance, the analytics you should demand, and the specific cases where we tell clients to keep their PRI line.

What cloud telephony actually means in the Indian market

Strip away the marketing and there are four distinct products being sold under the same phrase.

Hosted PBX / virtual PBX. Your extensions live on a vendor's platform. You get a softphone on laptops, mobile apps, and desk phones that register over SIP. This is what most people mean when they say cloud telephony. Examples in the Indian market: Ozonetel, Exotel, Knowlarity (now part of Gupshup), Tata Tele Smartflo, Vodafone Business Cloud, Airtel IQ, and self-hosted options running on FreePBX or 3CX on an Indian cloud region.

Cloud contact centre (CCaaS). Queueing, IVR trees, supervisor dashboards, call recording, workforce management, CRM screen-pop. Sold per agent per month. Examples: Freshcaller, Ozonetel CloudAgent, Ameyo (on-prem historically, now hybrid), Servetel, Exotel's contact centre tier.

SIP trunking. You keep your existing PBX — a physical IP-PBX in your server room — and replace the PRI line with a SIP trunk delivered over your internet connection. Pricing is per channel or per minute. This is the cheapest path when you already own a working PBX.

Unified communications (UCaaS). Telephony bundled with video meetings, chat, and file sharing. Microsoft Teams Phone, Zoom Phone, RingCentral. Usually the most expensive per seat, and usually overkill if you only want to make and receive calls.

Confusing these four is where most procurement mistakes start. A 30-person manufacturing office in Bommasandra that needs a reception number, six extensions, and a hunt group does not need a CCaaS platform with supervisor barge-in. Conversely, a 60-agent outbound collections floor does not need a UCaaS suite where half the licence value sits in video meetings nobody joins.

Per-seat versus per-minute: the arithmetic

Here is where we earn our fee. The two pricing models behave completely differently depending on your call volume and call duration.

Per-seat pricing

You pay a flat monthly fee per provisioned user. In 2026, mainstream Indian cloud telephony per-seat pricing sits in these bands:

• Basic hosted PBX seat with a local number and mobile app: ₹150-₹350 per user per month
• Mid-tier seat with call recording, IVR, and CRM integration: ₹350-₹700 per user per month
• CCaaS agent seat with queueing, supervisor tools, and wallboards: ₹900-₹2,200 per agent per month
• UCaaS seat (Teams Phone, Zoom Phone India): ₹550-₹1,400 per user per month, plus calling plan minutes on top for outbound PSTN

Per-seat is predictable. It is also indifferent to usage, which is precisely why vendors push it and precisely why it quietly wastes money when seat utilisation is uneven. A receptionist making 200 calls a month and a field engineer who logs in twice a quarter cost the same.

Per-minute pricing

You pay for talk time, usually with a small platform fee. Typical 2026 rates:

• Inbound calls to a virtual number: ₹0.40-₹1.20 per minute depending on volume commitment
• Outbound calls to Indian mobile numbers: ₹0.35-₹0.90 per minute
• Outbound to Indian landlines: ₹0.25-₹0.70 per minute
• International outbound to US/UK: ₹1.60-₹4.50 per minute (varies sharply by carrier route)
• Platform or trunk access fee: ₹1,500-₹6,000 per month, sometimes waived above a commit

Worked example: 30-person inside sales team

Assume 30 agents, each making 70 outbound calls a day, average call duration 3.5 minutes, 22 working days a month.

Monthly minutes = 30 × 70 × 3.5 × 22 = 1,61,700 minutes

In this scenario per-seat wins by a mile — provided the fair-use cap is real. This is the catch. Most Indian vendors advertise "unlimited calling" and then apply a fair-use policy that triggers at 2,000-3,000 minutes per seat per month. An agent doing 70 calls a day at 3.5 minutes averages 5,390 minutes a month. That is well past the cap, and the overage at ₹0.85 per minute turns the ₹42,000 bill into something closer to ₹1,00,000 — worse than per-minute and with a surprise attached.

Always ask for the fair-use threshold in writing, in minutes per seat per month. If the salesperson cannot give you a number, treat it as 1,500 minutes and model accordingly.

Worked example: 12-person inbound support desk

Now flip the profile. Twelve support agents, each handling 35 calls a day, average duration 5 minutes.

Monthly minutes = 12 × 35 × 5 × 22 = 46,200 minutes

Per-seat again, and comfortably inside most fair-use caps at 3,850 minutes per agent per month. Notice the pattern: per-seat wins when seats are fully utilised and call durations are moderate.

Where per-minute wins

The arithmetic flips in three situations, and they are common in Indian SMEs:

1. Low-utilisation seats. A 40-seat licence where only 15 people call regularly. You are paying 25 seats of pure waste. Per-minute with a trunk fee is almost always cheaper.
2. Highly seasonal businesses. A tax filing service with a March-April peak and near-zero activity in June. Per-seat makes you pay for July's empty desks.
3. Short, high-volume calls. Appointment reminders, delivery confirmations, OTP-adjacent outbound at 40-60 seconds a call. Per-minute at ₹0.40 for 45 seconds is ₹0.30 a call. A ₹1,200 seat that makes 200 such calls a month costs ₹6 a call.

There is a fourth case, less obvious: burst outbound campaigns. If you run a 3-day campaign with 20 tele-callers pulled off other work, per-minute is the only sane model.

PRI versus cloud: when the old line still wins

We still install and maintain PRI lines. Not because we are nostalgic, but because there are real cases where cloud telephony is the wrong answer in India, and pretending otherwise costs clients money.

Here is the honest comparison for a 25-user Bangalore office in 2026.

The case for PRI survives on two pillars: deterministic quality and independence from your internet link. If you run a hospital front desk, a diagnostic lab booking line, or a customer care number where a dropped call has a real cost, and your broadband is a single ₹2,500-a-month connection, a PRI line gives you a call path that does not care what your ISP is doing. We have seen enough monsoon-season fibre faults in Bangalore's outer ring to take this seriously.

The case against PRI is equally real. It is inflexible, undeniably more expensive per seat once you factor in PBX maintenance, and hostile to work-from-home. A hybrid model — PRI for the main inbound number, cloud for outbound and remote staff — is a legitimate answer, and it is what roughly one in five of our clients ends up running. You can read more about how we handle these mixed deployments on our communications services page.

The failure that taught us to always ask about the internet link

In July 2025, during a heavy monsoon week, a 60-seat BPO in Electronic City running fully cloud-based telephony lost their primary ISP for four hours and eleven minutes. They had a backup broadband line from a different provider, but nobody had configured it for SIP failover. Their cloud telephony vendor's platform was up and healthy the whole time. The phones were dead because the building's last mile was down.

Direct cost: roughly ₹1,80,000 in abandoned calls across two client accounts, plus a service credit they had to issue. Indirect cost was worse — one client invoked a penalty clause.

The fix cost ₹34,000: a dual-WAN router with automatic SIP-aware failover, a second ISP (Jio Business fibre, ₹3,999 a month for 300 Mbps), and a half-day of configuration. Four hours of provisioning work. The lesson is not that cloud telephony is unreliable. It is that cloud telephony moves the reliability question from the exchange to your broadband, and most buyers never ask the question at all.

If your office runs on a single internet connection, budget for a second one before you cut over to cloud telephony. Not after.

TRAI, DLT registration, and the compliance work nobody mentions

This is the part that delays deployments, and the part vendors mention last.

If you make any outbound calls to Indian mobile numbers at volume — marketing, transactional, or service — you fall under TRAI's Telecom Commercial Communications Customer Preference Regulations. That means DLT (Distributed Ledger Technology) registration.

The chain looks like this:

1. Entity registration. Your company registers on a DLT portal. There are multiple: Vodafone Idea, Airtel, Jio, BSNL, and the aggregated platforms like pingConnect and Kaleyra. Registration is usually free but requires your GST certificate, PAN, and a signed authorisation letter.
2. Header registration. Each SMS sender ID (the six-character alphabetic header) and each voice calling line identity must be registered. Per-entity header fees run ₹5,000-₹10,000 depending on the portal and the number of headers.
3. Template registration. Every SMS template and, for voice, every call-flow script category must be approved. Voice templates are less rigid than SMS but still require consent documentation.
4. Consent management. You must maintain proof of consent for every number you call commercially. For voice outbound, this is where most Indian SMEs get caught — they have a CRM field called "opted in" and no audit trail.

Timelines in 2026: entity registration is typically 3-7 working days if your paperwork is clean. Header and template approval add another 5-12 working days. In practice, budget three weeks from decision to first compliant outbound campaign.

Costs: registration itself is often free or nominal. The real cost is the per-call/per-SMS DLT lookup charge, which for voice varies by vendor but commonly lands at ₹0.03-₹0.10 per call for consent and preference checks, plus your platform's compliance tooling fee if they charge for it separately.

What happens if you skip DLT

Your outbound calls to mobile numbers get flagged, your calling line identity gets blocked by the telcos, and the numbers you spent months building recognition on become useless. We have watched a Bangalore fintech lose their primary outbound number for eleven days in 2025 because their dialler was calling numbers with no recorded consent. Re-registering a burnt number is not a fast process.

If you are already running outbound at scale and have never done the consent documentation properly, this is the single highest-priority item on your list. It outranks any feature comparison between vendors.

DPDP Act obligations that touch call recordings

India's Digital Personal Data Protection Act, 2023 has been progressively coming into force, and it directly affects cloud telephony because you are storing voice recordings and caller data. Two practical requirements: you need a lawful basis and, where consent is the basis, a clear notice to the caller; and you need the ability to delete a person's data on request.

Cloud telephony vendors that store recordings outside India create a complication — cross-border transfer is permitted under DPDP but subject to conditions. Ask your vendor where the recordings physically live. Indian-region storage (Mumbai or Hyderabad data centres) removes a layer of headache. Exotel, Ozonetel, and Knowlarity all offer Indian data residency. Microsoft Teams Phone recordings land in your Microsoft 365 tenant region, which for Indian tenants is typically India or Singapore depending on your licence configuration — check this, do not assume it.

Call analytics: the features worth paying for, and the ones that are noise

Every Indian cloud telephony pitch deck has a slide with a wall of charts. Most of it is decoration. Here is what we actually configure for clients, and what we tell them to ignore.

Worth paying for

Call disposition tagging. Agents mark outcomes — connected, not interested, callback, wrong number. Without this, every other metric is guesswork. This is the single highest-value analytics feature and it costs nothing to enable.

Live queue monitoring and abandoned call rate. For inbound operations, the abandoned call rate is your customer experience in one number. A support line abandoning more than 8% of calls in peak hours needs another agent, not a better IVR.

Talk-time distribution by agent. Not to rank people, but to spot outliers. An agent averaging 2 minutes against a team average of 5 is either excellent at quick resolutions or rushing people off the line. You cannot tell without listening, but the analytics tell you who to listen to.

First response time on inbound. How long before a human answers after the IVR. This is the metric your customers feel.

Missed-call-to-SMS or missed-call-to-WhatsApp. In India, missed call triggers are disproportionately effective. A missed call on your booking line triggering a WhatsApp message with a callback link recovers a meaningful share of lost enquiries. Several Indian platforms support this natively now.

CRM-integrated call logging. If your sales team is entering call notes manually, half of them are not doing it. Automatic logging with recording links attached to the CRM record is worth the integration effort.

Usually noise

Sentiment analysis on Indian-language calls. Hindi works passably. Kannada, Tamil, and mixed-language code-switching still produces garbage in most vendor engines. Do not buy a platform on this feature.

Speechanalytics-driven "agent coaching scores". Interesting in a 500-seat contact centre with a dedicated quality team. Useless in a 15-person setup.

Real-time wallboards on a 12-agent floor. The supervisor can see the room. Wallboards matter at 40+ seats.

Cost modelling: three real Indian SME profiles

Abstract pricing tables are hard to act on. Here are three profiles we have priced in the last year, with the actual product stacks and 2026 figures.

Profile A: 18-person professional services firm, MG Road

Two inbound numbers (reception and a partner line), 18 extensions, heavy email, light telephony, half the team travels. They had a 6-year-old Tata PRI line and a failing PBX.

Chosen stack: hosted PBX with mobile apps, per-seat pricing, 18 seats at ₹280 per user per month, two additional virtual numbers at ₹250 each per month, call recording enabled.

Monthly: ₹5,040 + ₹500 = ₹5,540 plus GST. Setup: ₹8,500 for configuration and number porting, done in four working days.

Removed: ₹9,200 monthly PRI rental and call charges, plus an AMC on the ageing PBX of ₹14,000 a year. Net saving roughly ₹52,000 a year with better remote-working support.

Profile B: 45-agent outbound sales floor, Koramangala

High-volume outbound to Indian mobiles, 45 agents, dialler required, CRM is Zoho. This is the case where per-seat "unlimited" plans collapse.

Chosen stack: cloud contact centre with per-minute outbound, 45 concurrent channels, Zoho CRM integration, call recording with 90-day retention, DLT entity and header registration handled as part of deployment.

Fixed monthly: ₹14,000 platform and trunk fee. Variable: average 3,10,000 outbound minutes a month at ₹0.52 = ₹1,61,200. Recording storage add-on: ₹3,500. Total: ₹1,78,700 per month plus GST.

A per-seat "unlimited" quote they had received from another vendor was ₹1,350 per agent = ₹60,750 a month. When we asked for the fair-use cap in writing, it came back at 2,500 minutes per agent. Their agents average 6,890 minutes. The overage would have been 4,390 minutes × 45 agents × ₹0.90 = ₹1,77,795, bringing the real total to ₹2,38,545. The "cheap" per-seat plan was ₹60,000 a month more expensive than the per-minute plan it looked like it beat.

This is the most common trap in Indian cloud telephony procurement. Get the fair-use number.

Profile C: 8-person clinic, Jayanagar

One number, appointments, low volume, and a hard requirement that calls never go unanswered during consulting hours.

Chosen stack: cloud telephony, 8 seats at ₹199, IVR with two options, missed-call-to-WhatsApp for after-hours, mobile app on the front desk and one doctor's phone.

Monthly: ₹1,592 + ₹300 number = ₹1,892 plus GST. Setup: ₹4,500.

There is no PRI argument at this scale. A business line with call hunting would cost ₹1,200-₹1,800 a month and would fail the moment the front desk steps away. Cloud wins outright, and the mobile app fallback is the whole reason.

What to demand before you sign

When you evaluate cloud telephony solutions in India, most of the risk sits in the contract and the fine print rather than the platform. Insist on these in writing.

Fair-use minutes per seat, in numbers. If it is not a number, it does not exist.

Overage rate per minute. Should be quoted clearly. Anything above ₹1.20 per outbound minute to Indian mobiles is expensive in 2026.

Number ownership. Your DLT-registered numbers and virtual numbers should be portable away from the vendor. Ask what happens to your numbers if you leave. The answer should be "you keep them and we release them within X days". If the answer involves the word "platform", walk away.

Recording retention and storage location. How many days, where physically, and what the retrieval path is if you leave.

Uptime commitment with teeth. A 99.9% SLA with no service credit is marketing. Ask what the credit is, and how you claim it.

SIP compatibility. Even if you go fully hosted, ensure your desk phones are standard SIP so you are not locked to one vendor's hardware. Poly, Yealink, Grandstream, and Snom all work with major Indian platforms. A ₹4,500 Yealink T33G should register on a new platform with a config change.

API access and CRM integration. If they charge separately for a Zoho or Freshsales integration, know the cost up front. Some Indian vendors bundle it, some charge ₹8,000-₹25,000 as a one-time integration fee.

Network readiness: this is where deployments actually fail

Cloud telephony quality is a network problem before it is a telephony problem. We will not cut a client over to cloud calling until their network passes three checks.

Jitter under 30 ms, packet loss under 1%, latency under 150 ms to the vendor's SIP endpoint. Measure this over a full working week, not a single ping test. Freeware tools like PingPlotter or a simple iPerf setup will do it. If your outbound firewall or ISP routing is unstable, SIP registration will drop mid-call and you will blame the telephony vendor for a network fault.

QoS on the router. Voice traffic should have priority over general browsing and, critically, over Windows Updates and cloud backup jobs. We have seen an office where 30 laptops pulling Windows patches at 10 a.m. turned every call into a garbled mess. A single QoS rule fixed it.

Separate VLAN for voice where the LAN supports it. Not always necessary in a 20-person flat network, but at 40+ endpoints it stops a broadcast storm or a looping cable from taking down the phones.

Backup internet with SIP-aware failover. Covered earlier, but it belongs on this list too. Two ISPs from two different providers, ideally on two different last-mile technologies (one fibre, one something else).

If you are outsourcing this, this is exactly the sort of pre-deployment work that sits inside a managed communications engagement. Our team handles the network readiness, the DLT paperwork, and the carrier liaison as a single scope — details on the communications services page.

Choosing between Indian vendors and global platforms

Indian platforms (Exotel, Ozonetel, Knowlarity, Servetel, Tata Smartflo, Airtel IQ) understand DLT, offer Indian data residency, price in rupees, and their support teams are in Indian time zones. Their dashboards are functional rather than beautiful, and their API documentation quality varies.

Global platforms (Microsoft Teams Phone, Zoom Phone, RingCentral) integrate beautifully with the Microsoft or Google ecosystem you already run, and their reliability engineering is excellent. They are also more expensive per seat, their India DLT handling is often through a partner and less smooth, and their support escalation paths can be slow for Indian-specific issues.

A practical rule: if telephony is a utility for your business, an Indian platform is usually the right call on cost and compliance. If you are deeply standardised on Microsoft 365 and want calling inside Teams, the convenience of Teams Phone may be worth the premium — but check the calling plan minutes carefully. Teams Phone calling plans in India have had availability quirks, and many organisations run Teams Phone with a third-party SIP trunk instead, which lands you back in the Indian vendor ecosystem anyway.

FAQ

What does cloud telephony cost per month in India in 2026?

For a basic hosted PBX, budget ₹150-₹350 per user per month for the seat plus ₹250-₹500 per virtual number. Contact centre seats run ₹900-₹2,200 per agent per month. Outbound-heavy operations should expect ₹0.35-₹0.90 per outbound minute to Indian mobiles instead of, or alongside, seat fees. A 25-person office with moderate usage typically lands between ₹6,000 and ₹15,000 a month plus GST.

Is DLT registration mandatory for cloud telephony in India?

DLT registration is mandatory for commercial outbound calling and SMS to Indian mobile numbers under TRAI regulations. You need entity registration, header or calling-line-identity registration, and template approval. Budget three weeks end to end. Skipping it risks your numbers being blocked by the telcos, and re-registering a blocked number is slow and painful.

Can I keep my existing landline or PRI number when moving to cloud telephony?

Yes, in most cases. Number porting from a PRI or landline to a cloud platform is a standard process, typically taking 7-15 working days. Your vendor handles the porting request with the incumbent operator. Keep the old line running until the port completes — do not cancel it early.

When should I keep my PRI line instead of moving to cloud?

Keep the PRI if your call quality requirements are absolute and your internet connection is a single consumer-grade line, or if you have a high-volume inbound number where any outage has a direct revenue cost. A hybrid setup — PRI for the main inbound line, cloud for outbound and remote staff — is a common and sensible middle path.

Does cloud telephony work during a power cut?

Partly. Your desk phones will die if they are powered from mains without a UPS or PoE backup. Your mobile app seats keep working because they use the mobile network. If call continuity during power cuts matters, put your router and PoE switch on a UPS, and ensure key staff have the mobile app installed and logged in. Bangalore's power reliability is decent but not perfect, and a 500 VA UPS costs ₹3,500.

What internet speed do I need for cloud telephony?

Bandwidth is rarely the constraint — a compressed voice call uses roughly 85-100 Kbps each way. Stability is the constraint. You need low jitter and low packet loss more than raw speed. A stable 20 Mbps connection with good QoS handles 25 concurrent calls comfortably. A jittery 100 Mbps connection will still produce bad calls.

A concrete next step

Pull your last three months of call detail records and your current telephony bill. Count two things: how many provisioned seats placed fewer than 50 calls in that period, and what your average call duration is. Those two numbers tell you whether you are on the wrong pricing model almost immediately.

If more than a quarter of your seats are underused, or your average call is under 90 seconds, you are probably overpaying on a per-seat plan. If your agents exceed 3,000 minutes a month each, check whether you are inside a fair-use cap — because if you are not, the next invoice will tell you.

Then get a second opinion on the commercial model before you renew. Bring us your CDRs and current contract terms and we will model the alternatives free of charge. Start by getting in touch with our team.