State of Playback

# How video is watched on the Bradmax Analytics network

Bradmax Analytics collects anonymised playback statistics from the players that enable it. This report is the aggregate: how that traffic is watched, in numbers, with the method and its limits stated.

It covers the accounts with analytics enabled, which is a per-player setting, and not the whole Bradmax client base. It isn't a picture of the internet as a whole either, and it says so wherever that distinction changes how a figure should be read.

## What this is

Bradmax Analytics is the measurement layer inside the player. Clients switch it on per player, so this report covers the accounts that have it enabled for September 2026; the sample size and the coverage behind each figure are stated in the method section at the end.

That boundary matters. Traffic from clients that run the player without analytics is not measured anywhere, so these numbers are a sample of the network and not its total. They also describe the Bradmax Analytics sample rather than the internet as a whole: a handful of large live-streaming clients account for most of the volume, so the mix below reflects their audiences as much as any general trend. Where that matters, it is said next to the number.

## Live is most of what is watched

Why so much live? Because the client base is built around always-on channels, the kind that run all day: sport, racing, worship, news. Of the 2,543,687 views that carried content metadata, 96.1% were live streams and 3.9% were on-demand. That is not a claim about the wider industry, it is who the sample is.

It is also why the watch-depth figure below covers on-demand content only. Live streams have no fixed duration to measure a percentage against.

## Playback volume over the last 13 months

Total views per month across the accounts that report to Bradmax Analytics.

| Month | Views |
| --- | --- |
| Sep 2025 | 3.69 M |
| Oct 2025 | 4.05 M |
| Nov 2025 | 4.58 M |
| Dec 2025 | 5.14 M |
| Jan 2026 | 5.35 M |
| Feb 2026 | 5.44 M |
| Mar 2026 | 5.68 M |
| Apr 2026 | 4.25 M |
| May 2026 | 4.70 M |
| Jun 2026 | 4.79 M |
| Jul 2026 | 3.98 M |
| Aug 2026 | 3.60 M |
| Sep 2026 | 3.64 M |

## How it is watched

Share of views by device group.

| Device group | Share of views |
| --- | --- |
| Mobile | 73.1% |
| PC | 25.2% |
| Other | 1.7% |

## Device types behind those views

| Device type | Share of views | Views |
| --- | --- | --- |
| Smartphone | 72.1% | 2,256,261 |
| Desktop | 25.2% | 789,307 |
| Smart TV | 1.4% | 44,427 |
| Tablet | 1.0% | 30,712 |
| Portable media player | 0.2% | 6,570 |
| Camera | 0.0% | 760 |
| Car browser | 0.0% | 231 |
| Game console | 0.0% | 121 |
| Smart display | 0.0% | 7 |

## Operating systems

| Operating system | Share of views | Views |
| --- | --- | --- |
| Android | 63.1% | 1,975,355 |
| Windows | 18.2% | 569,006 |
| iOS | 13.1% | 410,834 |
| macOS | 2.9% | 90,005 |
| Linux | 1.6% | 51,482 |
| (unknown) | 0.8% | 24,071 |
| Chrome OS | 0.2% | 7,643 |

## Browsers

| Browser | Share of views | Views |
| --- | --- | --- |
| Chrome | 71.8% | 2,246,721 |
| Safari | 11.0% | 344,310 |
| Samsung Internet | 4.5% | 141,968 |
| Edge | 4.2% | 132,446 |
| (unknown) | 4.2% | 132,150 |
| Firefox | 2.6% | 82,868 |
| Opera | 0.6% | 19,059 |
| Android | 0.5% | 14,825 |

## How it is connected

The player reads the connection the browser reports. Across the tagged views 48.0% were on Wi-Fi and 23.7% on cellular, and 79.1% reported 4G or better. That split is what makes adaptive bitrate and small startup buffers worth the engineering on a mobile-heavy audience.

These figures cover 50.5% of views, because only Chromium-based browsers expose the Network Information API. Safari and most TV browsers report nothing, so read this as the shape of the connected web audience, not of every screen.

## Connection type

Share of tagged views by connection type.

| Connection | Share of tagged views |
| --- | --- |
| Wi-Fi | 48.0% |
| Cellular | 23.7% |
| Not available | 15.1% |
| Not reported | 12.8% |
| Ethernet | 0.3% |
| Unknown | 0.1% |
| No connection | 0.0% |

Tagged views are 50.5% of all views in the period.

## Connection speed

Share of tagged views by reported connection speed.

| Speed | Share of tagged views |
| --- | --- |
| 4G | 79.1% |
| Not available | 15.1% |
| 3G | 5.6% |
| Slow 2G | 0.1% |
| 2G | 0.1% |

## How much of a video people actually watch

Across the on-demand views that carry engagement data, the average session reached 63.6% of the item. That number comes from 97.5% of on-demand views, so it is a solid reading of the on-demand audience rather than a partial sample.

It is the metric that matters most for content decisions. A view count tells you a player started; watch depth tells you whether the video was worth starting.

## Where the views come from

| Region | Share of views | Views |
| --- | --- | --- |
| Africa | 42.8% | 1,507,433 |
| North America | 27.9% | 983,558 |
| Europe | 22.8% | 801,833 |
| Asia | 3.8% | 133,357 |
| South America | 1.4% | 49,455 |
| Oceania | 1.3% | 45,274 |

## A note on the regional table

This table is the clearest example of the caveat at the top of the page. It follows where Bradmax clients stream to, and one client can move a region by tens of percentage points between editions. It is useful for capacity and CDN planning, not as a market-size estimate.

It stops at region on purpose. Countries are not broken out because on this traffic a single country can be most of a region, which would point at individual clients.

## What is not in this report

Start-up time, rebuffering and playback error rates are not here, because the analytics pipeline does not aggregate them. Neither is anything per client, per video, per country or below, and no demographic data is published. Traffic from players that do not run Bradmax Analytics is outside the report entirely.

Those limits are deliberate: the report is built from aggregate counters that cannot identify a person or a customer.

## Frequently asked questions

### Is this a survey or a measurement?

A measurement. The figures are read from the same aggregate counters the Bradmax analytics panel uses, not from a survey or a third-party estimate. The period, the number of reporting accounts and the coverage behind each figure are stated on the page so the sample can be judged.

### Does this cover all Bradmax clients?

No. It covers the clients that have Bradmax Analytics enabled, which is a per-player setting. Clients that run the player without analytics aren't measured, so the report is a sample of the network rather than its total. Enabling analytics is a configuration change, and it's what puts a client into the next edition.

### Does this describe how the whole internet watches video?

No. It describes the Bradmax Analytics sample. A few large live-streaming clients dominate the volume, so the device and regional mixes reflect their audiences. The page says so wherever it matters, and the regional table should be read as where Bradmax traffic goes, not as a market-size estimate.

### Why is most of the traffic live rather than on-demand?

Because the client base is built around always-on channels such as sport, racing, worship and news, rather than libraries of on-demand titles. Live was 96 percent of the views that carried content metadata in the reported month.

### Why are there no start-up time or error rate figures?

Because the analytics pipeline doesn't aggregate them. The player reports errors through a separate opt-in beacon that isn't part of these counters. Publishing a start-up or rebuffering number would mean inventing it, so it's left out and the gap is named on the page.

### How often is the report refreshed?

Quarterly, for the last complete month, with a twelve-month trend. Each edition is regenerated from the store rather than edited by hand, so the numbers stay reproducible and comparable between editions.

## Sources and method

- Source. Bradmax Analytics (VideoStats), aggregate counters from the players that have analytics enabled. Generated 2026-10-19T09:37:45Z.

- Sample. 3,640,989 views in September 2026 across 171 client accounts that reported to Bradmax Analytics.

- Scope. Only players with Bradmax Analytics enabled are counted. Clients that run the player without it aren't measured, so this is a sample of the network rather than its total.

- Method. Counters are read straight from the per-account statistics stores. Device, OS and browser figures come from the player user agent, and geography from an IP database resolved at collection time.

- Coverage. Connection figures cover 50.5% of views; watch depth covers 97.5% of on-demand views.

### Measure your own streams

These figures only cover players with Bradmax Analytics enabled. If yours isn't in them, switching it on is a configuration change, and the same numbers then appear for your own content.