Live stream analytics for corporate communications, what to measure and how to check a platform can measure it

After a town hall, the questions from leadership are always the same. Who watched, how long did they stay, and did it reach them properly? Most live streaming platforms answer the first one with a viewer count and leave the rest to guesswork.

What should live stream analytics tell a corporate communications team?

Strong analytics for internal video answer five questions, and a live viewer count answers none of them on its own. A peak of 4,200 viewers on a 6,000-person all-hands sounds fine until you learn that 1,800 of them left before the CEO reached the restructure, or that the Frankfurt office buffered for the whole hour.

Who watched, by identity

For internal communications, attendance means names, not numbers. A count of 4,200 can’t tell you whether the sales team showed up or a regional office missed the message. Names only exist if viewers have to sign in. When they sign in with their work account through single sign-on, each viewing is recorded against their name. Outside guests admitted through viewer management are recorded by email. On StreamShark both lists can be searched and downloaded as a spreadsheet, with each viewer’s device and when they joined and left, and the single sign-on list shows who came back for more than one session.

The trade-off is that identity only exists where access is gated. A public stream with no sign-in can report a count and a country, nothing more. And per-person viewing data is personal data, so in some countries it needs a conversation with HR or the works council before the first report is pulled. Decide what you’ll do with the names before you collect them.

How long they stayed, and when they left

Watch time turns a viewer count into a judgment about the content. Four minutes of a 45-minute broadcast is a different outcome from the full hour, and the moment people leave tells you which segment lost them. Per-viewer start and end times, plotted across the audience, give you the drop-off curve, the one chart worth putting in front of the executive who spoke.

Whether the stream reached them in good shape

A viewer who buffered for ten minutes and gave up counts as a viewer in most reports. What separates “they didn’t watch” from “they couldn’t watch” is how the stream played on each viewer’s screen, and it’s the question IT gets asked after every event. The way to see it is analytics built on CMCD, an industry standard in which the video player itself reports when it stalled or was slow to start. StreamShark’s CMCD analytics also show which caption and dubbing languages viewers chose, so you can see how the broadcast went in each language.

Where offices watch over peer-to-peer delivery, a report shows how well the stream played in each office, down to a single viewer. That’s the report that settles the “was it the stream or was it our network” argument.

The view across every event

One town hall is an anecdote. Twelve months of them is a trend, and the trend is what a communications team is judged on. You need one report across every event over a period you choose, with events run, viewership and peak. That lets you compare the quarterly results broadcast against the last three, or a Tuesday town hall against a Thursday one, then drill into any single event. The recording belongs in the same report. For a global workforce, the people who watch it the next morning often outnumber the live audience, and a report that stops at the live number undercounts the event.

What the operator did

The last question is for the production team. When something went wrong at 10:14, what was happening on the incoming feed, and what did the operator do about it? A record of the video signal coming in from the venue, plus a log of every action the operator took, lets you reconstruct the run of the show. Without it, the post-mortem is memory against memory.

Which metrics matter during the event, and which matter after it?

Split the metrics by when you can act on them. During the event, the only numbers worth watching are the ones that let someone fix a problem while the audience is still there. Afterwards, the useful numbers are the ones that change the next event.

Metric	What it tells you	When it’s useful
Signal from the venue	Whether the feed from the room is healthy	During, as an alert
Concurrent viewers	Whether the audience arrived, and when they leave	During
Buffering and startup time by region or office	Where delivery is struggling	During, to decide whether to switch to the backup stream; after, to fix the network
Per-viewer start and end times	Attendance by name, and the drop-off curve	After
Returning viewers	Who came back to the recording or a second session	After
Caption and dubbing language use	Which languages the audience actually chose	After, to plan the next event’s languages
Events run, viewership and peak across the period	The trend leadership asks about	Quarterly
Operator action log	What was done and when, for the post-mortem	After

StreamShark’s real-time stats go down to the individual viewer, so a producer can see the audience thinning during a segment and shorten the next one. The analytics and insights page lists each report and what it exports.

How do you check a platform can measure these before you buy?

Ask for the report, not the feature. Every feature list says “analytics”, and the word covers everything from a viewer count to a per-person export. The test is whether the vendor can hand you the file you’d send to leadership after a town hall.

1. Write the questions first. List what leadership asked after your last event. Attendance by department, watch time, offices that struggled, how the recording performed. If a report can’t answer one of them, it’s a gap, however good the dashboard looks.
2. Run a gated test event. Stream to a small group signed in through your SSO or a viewer list, with at least one person in each region and one outside the corporate network. Twenty people are enough.
3. Pull every report the next morning. Export them, open them in a spreadsheet, and check the columns against your questions. Look for identity per session, start and end times, device, and whether the on-demand views of the recording are in the same report as the live ones.
4. Break something on purpose. Have one viewer watch on a deliberately slow connection, such as a phone hotspot with weak signal. Then check whether the platform saw it, from the viewer’s side, and where that shows up.
5. Check the period view. Ask for a report across the last three test events together. If each event is an island, the quarterly trend will be assembled by hand forever.

What should you ask a vendor about analytics?

Put these to each vendor in writing, and keep the answers next to the reports from the test event. The two should agree.

• Can we see which individual viewers watched, by SSO identity or by email, and export that as a file?
• Are the analytics available while the stream is running, and down to what level?
• Can you see how the stream played on each viewer’s screen, or only what your servers sent?
• Can we report per office or per network, not just per country?
• Is the incoming stream’s health recorded, and are operator actions logged?
• How long is per-viewer data retained, and who at the vendor can see it?

The last question matters more than it looks. A report of who watched the restructure announcement is sensitive, and the retention and access answers are the ones your security team will ask you for later.

Questions communications teams ask about live stream analytics

Can live stream analytics show who watched a town hall?

Yes, when viewers sign in through single sign-on or a managed viewer list. Each session is then logged against that identity and can be exported. An open stream reports counts and locations only.

What is a good watch time for an internal live stream?

There is no industry number worth quoting, because it depends on the length and the subject. The useful comparison is your own events against each other. Track the share of viewers who stayed to the end and the point where most of the leavers left, and watch whether both improve as you change the running order.

What is CMCD analytics?

CMCD (Common Media Client Data) is a CTA WAVE standard in which the player reports its own buffer level and throughput with each request. Analytics built on it show playback quality as the viewer experienced it, not as the server assumed it.

Do the analytics cover the recording as well as the live event?

They should. Ask for the per-viewer and account-wide reports to include on-demand views, so people who watched the recording the next day are counted with the live audience.

Want to see the reports your team would actually send after a town hall? Talk to our team and we’ll walk through them with your own test event.