Ask what the cost to add video calls to an app is and you will get a number that turns out to be half the answer, because video calling is the clearest example of a feature where the build price and the running price are genuinely separate businesses. There is the engineering work to put calling inside your product, which for most apps lands between £6,000 and £45,000. And there is the per participant minute fee that starts the day you launch and never stops, which is small at first and becomes the dominant number if your product succeeds.
Founders who budget only for the first are fine through beta and surprised in month eight. Founders who assume video is a solved commodity underestimate how much sits between a working demo on office wifi and a call that survives a train journey through Reading. This guide separates the two costs cleanly, compares the realistic options in 2026, and shows where each one stops making sense.
What does the cost to add video calls to an app actually come to?
The cost to add video calls to an app is a one off build plus an ongoing usage fee. A one to one calling feature typically costs £6,000 to £15,000 to build. Multi participant rooms with screen sharing, recording, and a proper lobby usually cost £20,000 to £45,000. The usage fee is charged per participant minute by whichever provider carries the media, and most providers include a monthly free allowance that covers early stage volume entirely.
That shape matters more than the exact figures, because it changes how you should plan. A feature whose cost scales with success needs a unit economic answer before launch, not after. If each user costs you money every minute they spend in your product, you need to know what a user is worth per minute, and most teams building video have never done that sum.
The build, by scope
| Scope | Typical build cost | What it involves |
|---|---|---|
| One to one calling | £6,000 to £15,000 | Provider integration, call setup and teardown, permissions, mute and camera controls, reconnection handling, in call UI |
| Small group rooms | £15,000 to £28,000 | Everything above plus multi participant layout, active speaker detection, join and leave states, device selection |
| Full room product | £28,000 to £45,000 | Everything above plus screen sharing, recording, waiting rooms, moderation controls, network quality indicators, chat alongside video |
| Raw WebRTC, self operated | £40,000 upwards | All of the above plus signalling servers, relay infrastructure, scaling, and permanent operational ownership |
The jump between the first two rows surprises people, and the reason is layout. One to one video is two boxes. Four people is a layout problem that changes on every join, leave, and network wobble, and doing it without the interface jumping around is real work.
The four realistic options in 2026
Daily
Daily is the option we reach for most often for product teams who want video working quickly and predictably. The pricing is unusually transparent: a free allowance of 10,000 participant minutes a month, then 0.004 US dollars per additional participant minute, with graduated volume discounts that fall as far as 0.0015 US dollars per minute at very large scale. Add ons are priced separately, including cloud recording at 0.01349 US dollars a minute, live streaming over RTMP at 0.015 US dollars, and HLS at 0.03 US dollars.
That free tier is more generous than it sounds for an early stage product. Ten thousand participant minutes is, for example, roughly 250 four person calls of ten minutes each per month, which is more than most apps do in their first year.
Agora
Agora is the other mature specialist and prices similarly in structure. There is a free monthly allowance of 10,000 minutes, and the published standard rate for high definition video is 3.99 US dollars per 1,000 minutes, with committed subscription tiers and credit packs bringing the effective rate down substantially for teams willing to commit in advance. Different video resolutions convert differently against that rate, so your real cost depends on the quality you actually stream, not on the headline number.
Agora suits teams who expect volume and are prepared to negotiate or commit, and it has strong coverage in regions where other providers are weaker.
Twilio Video
Twilio’s video product has had a complicated few years. Twilio announced an end of life for Programmable Video in 2024, then extended the retirement date by two years to December 2026 after customer pushback, and subsequently reversed the decision entirely, confirming that Twilio Video remains a standalone product with continued investment.
That is a genuinely good outcome for existing customers and it is also a piece of history worth weighing if you are choosing today. Twilio remains a credible option, particularly if you already use Twilio for messaging and want one supplier relationship. Confirm the current product status and pricing directly with Twilio before you commit, because this is the one area of the comparison where the ground has moved recently.
Raw WebRTC, operated yourself
WebRTC is the underlying browser and mobile technology all of the above are built on, and yes, you can use it directly and pay no per minute fee to anyone. The catch is that WebRTC gives you the media transport and almost nothing else. You still need signalling to introduce two peers to each other, relay servers for the substantial share of users behind restrictive networks who cannot connect directly, and a selective forwarding unit as soon as you go beyond two or three participants, because sending everyone’s video to everyone else stops working almost immediately.
All of that is servers you provision, monitor, scale, and get paged about. The per minute fee disappears and reappears as infrastructure cost and engineering time. For a team with real time infrastructure experience and serious volume, it can be the right answer. For everyone else, it is a way of converting a predictable variable cost into an unpredictable fixed one.
How to work out your real running cost
Do this before you pick a provider, because it usually decides the answer.
Start with participant minutes rather than call minutes, since that is what you are billed on. A four person call lasting ten minutes is forty participant minutes. Then estimate three numbers: how many calls a month, average participants, and average duration.
Take a telehealth app doing 800 consultations a month, two participants, twenty minutes average. That is 32,000 participant minutes. On a rate of 0.004 US dollars per minute with 10,000 free, you pay for 22,000 minutes, which is 88 US dollars a month. Not a business problem.
Now take a group coaching app doing 400 sessions a month with six participants for sixty minutes. That is 144,000 participant minutes, so 134,000 billable, which is 536 US dollars a month, and it climbs linearly as you grow. Add recording on every session and you are adding roughly another 1,940 US dollars a month at the rates above. Suddenly the unit economics of your pricing tier matter a great deal.
The lesson is not that one of these is expensive. It is that the same feature has wildly different economics depending on participants and duration, and you can work out which situation you are in on the back of an envelope in ten minutes.
Where the build cost actually goes
Teams estimating this themselves usually cost the happy path and miss four things that consistently take longer than expected.
Reconnection and network transitions. A user walking out of the office onto mobile data should stay in the call. Handling that gracefully, including the state your interface shows during the gap, is a meaningful chunk of the work and it is what separates a demo from a product.
Permissions and device selection. Camera and microphone access, choosing between a headset and built in speakers, and the case where the user denied permission months ago and cannot work out why nobody can hear them. Each of these is small, and together they are a fortnight.
Layout under change. As covered above, multi participant layout that stays calm while people join and leave is real engineering.
Recording, if you need it. Recording sounds like a toggle and is not, because it brings storage, retention policy, access control, and, for anything involving UK users, a consent and data protection question. If your recordings contain personal data, and a video of a person always does, you need a lawful basis and a retention period under UK GDPR, and the Information Commissioner’s Office expects you to be able to explain both.
When video calling is the wrong feature
Worth saying plainly, because we have talked several clients out of this. Video is often requested when the underlying need is presence and responsiveness, not moving pictures. If your users mainly need quick answers, asynchronous messaging or voice notes solve the problem at a fraction of the build cost and none of the per minute cost. Our breakdown of the cost to add in app chat covers that comparison, and chat plus well timed push notifications covers a surprising share of what founders originally described as needing video.
Video earns its cost when the interaction genuinely requires seeing someone: consultations, teaching, assessment, interviews, anything where facial expression or a physical demonstration is part of the value. If you cannot say which of those you are, the answer is probably not video yet.
What we would recommend
For a first version, use a managed provider, not raw WebRTC, almost regardless of your engineering strength. The free allowances mean you will pay nothing while you find out whether people use the feature, and the build is a fraction of the self operated route. Revisit the decision when your monthly usage bill is large enough that a dedicated infrastructure engineer would be cheaper, which is a threshold most products never reach and the ones that do reach it can afford.
Scope the first version to the smallest call that delivers your actual use case. One to one before groups. No recording until someone asks twice. Get it in front of users, watch the participant minutes, and let the real numbers decide what comes next.
For where video sits against other features in a UK build budget, our 2026 app development cost guide puts these figures in context. And if you want us to run your specific call volumes and tell you what this will cost you in year two rather than month one, send the numbers through our project brief form.
.webp)
.webp)
