For brands
How to Measure Livestream Ad Impact
By Marcell Dechant · Chief Executive Officer
To measure livestream ad impact is easier than in television and harder than in search. Easier, because every placement is a technical event you can log. Harder, because several steps sit between "was displayed" and "had an effect", and no counter covers them. This piece sorts out which number proves which step — and where a number stops being a statement.
Level 1: Delivered
The bottom level is the most solid. A placement was delivered if the stream was genuinely live at that moment and had viewers. Exactly that can be recorded server-side, with channel, timestamp and viewer count.
The difference from an estimated reach in a media kit is substantial. An estimate says what a channel usually achieves. A delivery log says what happened that evening. So when comparing offers, the first question is not "how many impressions" but "what produces that number".
Level 2: Seen
Here it gets fuzzy, and that should not be glossed over. That a placement was on screen and the stream was running does not mean anyone looked. Every medium has this gap, television and outdoor included.
What distinguishes the livestream environment is the attention posture: viewers stay with one channel for a long time and chose it themselves. What does not distinguish it is the measurability of that attention — you would need eye tracking, and there is none here. Anyone selling you a viewability rate to two decimal places is selling a model, not a measurement.
Level 3: Responded
Response is hard to measure again, but rare. A click on a tracked link, a redeemed creator code, a visit through a short link: these are timestamped events that can be attributed to a campaign and a channel.
Two warnings. First, click rates in streams are inherently low — most viewers do not have a hand free, and many are watching on a TV. Second, a code shifts the measurement: it is often redeemed days later, which complicates attribution but in exchange evidences actual revenue rather than an intermediate step. How we relate clicks and cost to each other is in our CPM explainer.
Level 4: Worked
Awareness, favourability, purchase intent — no delivery statistic measures those. A survey before and after the campaign does. Brand lift studies of that kind make sense above a certain budget and are waste below it, because the sample can no longer separate the effect from noise.
The pragmatic substitute for smaller budgets is a test with contrast: a campaign in a clearly bounded period, a comparable period without one beside it, and then a look at branded search volume, direct traffic and revenue. That is not a clean causal proof, but it is more honest than an over-read click rate.
What to actually demand
Four questions are enough to compare offers:
- What produces an impression — an estimate or a delivery log?
- Do I get the breakdown per channel and per day, or only a total?
- Are clicks attributed unambiguously to a campaign, and how is double counting prevented?
- What is explicitly not in the report?
The fourth is the most revealing. A provider who can name the limits of their own numbers has understood them. Which metrics end up in our closing report is shown on the advertiser page.