How do you actually measure a streamer sponsorship?

Why every streamer sponsorship runs unmeasured today

Look at what is already on screen across live streaming: energy-drink logos, sportsbook boards behind the fighters, peripheral and crypto brands parked in the corner of a thousand channels. Millions of dollars in committed sponsorship spend, and almost none of it is instrumented. The logo goes up by hand, and the reporting afterward is a screen recording and a hopeful email. This is not a gap in a tooled market. It is an untooled category, and your marketing team is being asked to buy in it on faith.

It is not a bad meter. There is no meter.

Out-of-home advertising spends billions on impressions estimated from traffic surveys, and nobody can prove who actually looked at a billboard. A painted billboard cannot count anyone. The reason a live stream finally can is not cleverness, it is architecture: the billboard now has a server behind it. The ad on a stream is a browser source, and a browser source can count, continuously, for every creative it renders. That one fact is what turns an unmeasurable medium into a measured one.

How the viewer-second is built, kept honest

The unit is the viewer-second: one real viewer, present for one real second while your creative is on screen. It has two inputs, and Publsh never blends them silently. The live audience comes from the platform's own count of concurrent viewers, and Publsh labels that platform-reported. The seconds your creative rendered come from the ad itself as Publsh serves it, measured on both platforms; the audience they are multiplied by is platform-reported on both Kick and Twitch, and Publsh labels every figure by how it was known. Multiply the two and you have the delivered exposure you actually bought, not a follower count and not a guess.

Worked example: your creative renders for three hours in a session that averages 2,000 concurrent viewers. That is 3 hours by 3,600 seconds by 2,000 viewers, or 21.6 million viewer-seconds, which is 6,000 viewer-hours. (Illustrative math, not a promised audience.) You and the creator work from the same serving record of when the ad was on screen, because it came from delivery that was recorded, not a figure typed into a deck.

How does Publsh count a stream ad's delivery?

Here is the mechanism, not a metaphor. One overlay link is pasted into the broadcast software once. From then on, while your booked creative is on screen, that overlay reports a heartbeat. Publsh sums those heartbeats per creative, only inside your flight's booked window, and only while the channel is confirmed live. Heartbeats from outside a live window are discarded as fraud protection, so an ad that rendered to nobody never reaches your invoice. Because each creative is counted on its own, a rotating board of sponsors is measured independently, each on its own on-screen time, with no separate billing path.

When the platform's audience feed drops for a stretch while your creative is confirmed on screen, that portion is marked as an estimate rather than silently dropped or invented, so you always know which parts of a report are measured and which are estimated.

Not every placement is worth the same: share of screen

A corner logo and a full banner strip do not deliver the same thing, and Publsh prices the difference instead of ignoring it. Every placement holds a known share of the screen, its area as a fraction of the full 1920 by 1080 canvas. Publsh combines that share with viewer-seconds into weighted exposure: audience, multiplied by time on screen, multiplied by how much of the frame you actually held. It is the live-stream version of the media value that sponsorship has always been sold on, except computed from real delivery instead of estimated off a rate card.

You pay for what rendered, not for a promise

Billing runs off delivered exposure, never a committed total on paper. Publsh tracks each flight against what was committed and against where it should be by now, and flags a flight trending to underdeliver before it closes rather than after. If a flight comes in short, you pay for what actually rendered. There is no clawback fight and nothing to renegotiate afterward, because the number came from the serving log, not from a promise either side made at booking.

CPVH: the efficiency number, with a benchmark or it does not ship

Once delivery is measured, efficiency is one honest number: cost per 1,000 viewer-hours, CPVH. Spend divided by viewer-hours in thousands. If a flight cost 900 dollars and delivered 6,000 viewer-hours, that is 150 dollars per 1,000 viewer-hours. (Illustrative.) It derives straight from the exposure and is comparable across creators of any size and across both platforms.

A rate on its own is not information, so Publsh ships every efficiency number with a comparison or not at all. Not 4.02 dollars, but 4.02 dollars against the category median, direction marked. (Illustrative.) Those benchmarks are pooled per platform and anonymized, and they surface only once there is a real sample behind them; below that threshold a report reads insufficient data rather than a number that would mislead you. Early on you will see that written on the page, and that is the point.

The one number no unmeasured channel can show: measurement quality

There is one figure here that only means something if you genuinely measure: measurement quality, the share of a flight that was fully measured versus filled in by fallback. A report reading, for example, 98 percent measured is honesty rendered as a feature. (Illustrative figure.) Every other channel reports estimates as though they were facts. Publsh is the only party in this market positioned to show your marketing team the difference, on every flight, in one number.

What about CPM and impressions?

Buyers ask for CPM out of habit, so here is the honest translation. A persistent on-screen banner has no natural impression boundary, so any CPM for it is a published convention, not a physical fact. Publsh can express viewer-seconds as an effective CPM and show it to you for comparison, but it leads with CPVH and viewer-seconds because those are the measured truth. If you want the full breakdown, impressions vs viewer-seconds is the whole game: one is a modeled guess, the other is a measurement.

Attribution without a single cookie

Measuring exposure is half the picture; you also want what it drove. Publsh attributes response through promo codes and tracking links you already own. When a viewer redeems a code, that flows back as redemptions, cost per redemption, and return on ad spend, and Publsh labels it as your data reported by your own systems, never asserted as something Publsh measured. It never tracks a viewer, builds a profile, or sets a cookie, so the model does not flinch as privacy law tightens and third-party cookies disappear.

One limit, stated plainly: a burned-in overlay is not clickable, so there is no click-through rate here, from anyone, ever. You measure real exposure and attribute real response, instead of reporting a click metric this medium cannot honestly produce.

The proof is in what Publsh refuses to report

The fastest way to spot an unmeasured channel dressed up as a measured one is to ask what it will not report. Here is what Publsh will not, on purpose: no invented impression count as a headline number, no unique reach or frequency (that needs viewer identity Publsh does not collect), no demographics guessed from a platform API, no brand-lift figure it cannot substantiate. A smaller set of numbers that are real beats a full dashboard of numbers that are modeled. Publishing what Publsh will not claim is itself the proof that the rest is measured.

You do not need new budget to prove it out

This is the easiest yes in advertising. Your team is almost certainly already paying a creator to put your brand on their stream. Run that same deal through Publsh and you finally see what you bought instead of guessing: the same spend, brand-safe because you approve every creative, and a level of proof you have never had at any price. On-screen time is measured on both platforms from the serving side, and audience figures are platform-reported on both Kick and Twitch, so the proof extends across both platforms.