How to check that your ad actually ran in a live stream

Imagine an invoice for 300,000 views (a hypothetical figure). What you actually need to know is narrower: was your creative on screen, in full, in front of real people, on the channels you approved and at the times you paid for? A live stream can't be rewound for an audit, so the proof has to be collected at the moment of display.

Why “the ad was served” is not proof

The Media Rating Council's Viewable Ad Impression Measurement Guidelines (version 2.0, 2015, written for desktop browsers) note that served impressions are counted regardless of whether the ad appeared in the viewable space of the page, so not every served impression had a chance to be seen. A viewable display impression needs at least 50% of the ad's pixels in view for at least one continuous second; a video ad needs the same 50% of pixels in view for two continuous seconds of play.

Those rules were written for display ads on web pages and video ads in web players. An ad rendered into the picture of a live stream is a different object, and the MRC thresholds don't transfer to it one-to-one. The principle does: separate “delivered” from “seen”. In a live stream, failures are specific:

the streamer hides or switches off the ad layer, or changes the scene so that something covers it;
the creative doesn't load, renders broken or sits in the wrong place;
the chat link is never posted, or is posted but not pinned;
the audience is inflated by bots.

Each needs its own check; a total on an invoice covers none of them.

Six layers of proof behind one view

At APIQ, a view is verified only if the creative was shown correctly, was fully visible to the viewer and met the campaign requirements. Only verified views count toward delivery and appear on the invoice.

Layer	What it proves	What you see
Screenshot at the moment of display	The creative was in the stream	A screenshot of every view, with streamer, time and audience size
Computer vision	The creative was there, intact, in the right position	Hidden overlays, broken creatives and wrong positions are caught; clickable elements are checked
Chat-link check	Viewers had something to click	On Twitch and Kick, the system checks that the link was posted and pinned
Anti-fraud, 15 parameters	The audience is real	Suspicious traffic is excluded from the report and the invoice
Online report	Delivery is on track	Verified views, CTR, QR scans, chat clicks, results by creative and by streamer
Your own attribution	Results don't rest on the seller's numbers	AppsFlyer, Adjust, tracking links, UTM tags, pixels

The anti-fraud check looks at signals such as chat activity, audience dynamics and stability, sudden spikes in viewers, user behavior, clicks and conversions. Examples: the ratio of viewers to chat participants, how regular and how long a channel's streams are, and whether viewer profiles have an avatar and a description. The full list of parameters is available on request.

When a check fails, the view doesn't reach the invoice. The streamer can't change the creative, and if they switch off the layer, alter the settings or remove the link, the system sees it right away: the streamer is penalized and those views aren't billed. If bots are found, those views aren't paid for and the streamer is removed from the platform. How screenshots also prove the context around the ad is covered in Brand safety in live streams.

How to read the report in ten minutes

You don't need to audit every view. A short routine is enough:

Volume against the plan. Compare verified views with the volume fixed in the media plan. Views are spread evenly across the campaign days, and the daily plan is recalculated every night, so a lag shows early, not on the last day.
Channels against your list. Every streamer in the report should be on the list you approved before launch. You can remove any streamer during the campaign, too.
Spot checks of screenshots. Pick views from different streamers and hours. Check that the creative is the right one, sits where it should and that the recorded audience size looks plausible for that channel.
Outliers by streamer. Each channel gets its own link by default, so a channel whose views bring almost no clicks stands out. CTR is tracked online on every clickable element, and a streamer who drops below the threshold agreed with you is removed from the campaign automatically.
Live only. Creatives stay in the recording of the stream, but streamers are paid only for live views; views from recordings come free and add only a few percent to a campaign's volume.

How to break CTR down by chat link, QR code and on-screen click is explained in Where clicks come from.

Proof of a view is not a total on an invoice. It is a screenshot, a channel, a time and an audience you can check one by one.

Want to see a media plan for your brand?

Get a media plan

Reconciling the report with your own analytics

Your numbers and the seller's will never match exactly. The point is to know which gaps are normal and which are a signal.

Clicks are not sessions. Google's own troubleshooting guide for Google Ads and Analytics lists, alongside Google-specific ones, causes that apply to any channel: clicks and sessions are different metrics; redirects on the landing page can strip campaign parameters; users who leave before the tracking code runs never become a session; browser settings and opt-out add-ons block collection; a landing page without the tag passes nothing; a returning visitor can create more sessions than clicks. If your site loads analytics only after cookie consent, visitors who decline won't appear as sessions either.

Agree the attribution window. Counts depend on how long after a click or view an action is still credited. AppsFlyer defaults to 7 days for clicks and 1 day for impressions when the link sets no window of its own; Adjust defaults to 7 days for click-based and 24 hours for impression-based device matching. Set them before launch and use the same ones in every report.

Count post-view separately. On Twitch, APIQ accepts impression pixels, so conversions can be counted after a view as well as after a click.

Gap	Likely cause	What to do
Far fewer sessions than clicks	Redirect strips UTM tags, slow landing page, tag missing	Test every channel's link before launch
Installs differ between systems	Different attribution windows	Fix the same windows for both reports
Conversions without clicks	Post-view conversions	Report post-view and post-click separately
One channel lags in your data	Wrong or reused link	Check that the channel uses its own link

On our cases page, a 30-day fintech campaign on Twitch, with 9.86M verified views on 59 streamers at once, lists a reporting match rate of 99.94%. That is one campaign's figure, not a promise for yours; for any such number, from us or another seller, ask what exactly was compared with what.

Red flags in any seller's report

Only totals, with no breakdown by channel, creative or day.
Screenshots sent by the streamer after the stream instead of captured at the moment of display. Typical of flat-fee deals; see CPV, CPM or flat fee.
Impressions or “reach” presented as views, with no definition of what a view is.
No audience size recorded at the moment the ad ran.
Anti-fraud described in one sentence, with no willingness to share the criteria.
Your own tracking links or pixels are not allowed.
Channels in the report that you never approved.

Questions to ask before you buy

What exactly counts as a view, and is that the unit on my invoice?
Will I get a screenshot of every view with the streamer, time and audience size?
How do you detect a hidden or broken creative, and what happens to those views?
Which parameters does your anti-fraud check, and can I see the full list?
Can I use my own attribution, with a separate link for each channel?
Do you support post-view measurement, and with which attribution windows?

With APIQ, the media plan arrives within 24 hours, and the launch follows 1–2 days after the streamer and creative list is approved; many clients start with a pilot to get their own numbers. More on the brands page.