Pause ads: what they are, and what it takes to run one

Every streaming service now sells the moment you hit pause. Here is what the format actually is, what one looks like in a real player, what each platform's specs are, and which of the numbers everybody quotes hold up when you go looking for the study behind them.

Published 16 July 2026 · My TV Channel

What is a pause ad?

The IAB Tech Lab, which is currently trying to standardise the format, defines it as "a TV ad experience initiated when the viewer uses a remote to pause the content being watched." That definition is doing something specific. The viewer initiated it. Nobody interrupted anything.

That is the whole pitch, and it is a good one. A pre-roll takes time the viewer wanted to spend watching. A pause ad takes a screen the viewer was about to abandon anyway, because they got up to answer the door. The industry term for this is non-interruptive, which is marketing language for a real distinction.

One quick disambiguation, because these two things share a name and Google mixes them up. If you are looking for how to stop a campaign from spending in Meta Ads Manager or Google Ads, that is pausing an ad, and it is a toggle in your campaign dashboard. This page is about the pause ad as a creative format on television.

What a pause ad actually looks like

This is a real one, screenshotted from the My TV Channel player on iOS. Not a mockup.

Look at what the player adds around the creative, because that chrome is most of the format:

• An Ad label in the corner. Not decoration. A viewer has to be able to tell an advertisement from the show without thinking about it.
• A close button, which here means what it means everywhere else in this player: leave. It does not mean dismiss the ad.
• A Resume pill, which exists precisely because the close button means leave. Without it the viewer who wants their show back has no obvious way to get it, and an ad that traps you is a worse ad.
• A QR code baked into the creative. This is the part that makes pause ads commercially interesting on a TV, where there is nothing to click and the viewer is holding the one device that can scan.
• A Learn More button for the platforms where a tap is possible.

The creative itself is a still. No sound, no motion. That constraint is not a technical limit, it is a finding, and it comes from the company that invented the format.

Where pause ads came from

Hulu announced the pause ad in December 2018 and unveiled it properly on 31 January 2019, with a launch in the second quarter of that year. The first two brands to run it were Coca-Cola and Charmin.

The interesting part is in the research Hulu did first. Jeremy Helfand, then Hulu's head of ad platforms, put it plainly: "Viewers were strongly against it being a video format." So the pause ad shipped as a semi-transparent, non-video banner that appeared five seconds after you paused, and Hulu kept it off TV-MA episodes and live TV entirely. The company said at the time it wanted half of its ad revenue coming from non-disruptive formats within three years, back when it had 25 million subscribers.

Everyone else followed:

Six years from one company's experiment to a format the IAB is writing guidelines for. The second public comment period on the IAB Tech Lab's Ad Format Guidelines for Digital Video and CTV v2, which covers the pause ad, closes today, 16 July 2026.

Pause ad specs, platform by platform

Every platform publishes its own spec sheet and none of them reference each other, so here they are side by side. Everything below is from first-party documentation. Where a platform does not publish a number, the cell says so rather than guessing.

Two things jump out of that table. The first is that the delay is not standard and nobody agrees: YouTube makes you wait twice as long as Hulu did. The second is the turnaround column. Twelve business days at Disney, two weeks at Roku, for a single static image. That is not a technical constraint, it is a human review queue, and it tells you what these companies are actually worried about.

How pause ads work technically

The format sounds trivial. Viewer pauses, show picture. We built pause ad serving into the My TV Channel player, and the two hard parts are not the ones you would guess.

Knowing that a pause is a real pause

A video player reports itself as paused for all sorts of reasons that have nothing to do with a viewer pressing a button. Ours pauses itself to swap the stream when it recovers from a network stall, and on a live playlist it can sit in that state for several seconds. If you hang your ad trigger off that, you fade an advertisement over a viewer whose stream just broke, and then you count it as an impression. We had to suppress those explicitly.

The same trap catches you from the other direction. If you track your own intent instead of the player's actual state, you miss every pause the system initiated: the viewer who paused from Control Center, the one who unplugged their headphones, the phone call that came in. Those are real pauses and the viewer is really looking at a frozen frame.

Timing

Our player waits 600 milliseconds before it shows anything, and cancels if playback resumes first. Somebody who taps pause and immediately taps play was adjusting something, not taking a break, and they should never see an ad. Once the ad is up it must be visible for 800 milliseconds before we count an impression at all.

We also count one impression per creative per playback session, not per display. A viewer who taps pause and play ten times has seen one ad, and any other choice quietly inflates the only number the format produces.

Privacy

Our impression beacon carries no viewer identifier. None. That is a deliberate constraint rather than an oversight: it is what keeps the format out of tracking territory under Apple's App Tracking Transparency rules, and it is why any frequency capping has to happen on the device instead of the server. You give up per-user reporting. You get an ad format that does not need a consent prompt.

Are pause ads effective?

Probably. But almost every number you will read about pause ads has been laundered, and since this page exists to sell you something, the least we can do is show our working.

Numbers that hold up

NBCUniversal, December 2023. Peacock's own case study reports a 41% increase in ad memorability versus a standard mid-roll, and that 66% of pause events run longer than 60 seconds among adults 18 to 49. The methodology footnote is specific: long-form VOD, ten-foot devices only, pause events under 5 seconds excluded, fast-forward and rewind excluded, durations capped at 600 seconds. That is a real measurement of a real thing.

Wunderkind Ads with OpenGlass, measured by TVision, June 2026. Pause ads drew 69% more attention than a traditional 60-second CTV spot, across 13 verticals. Automotive was the standout at 34.2 seconds against 12.2.

Magna and DIRECTV, 2025. 91% of people pause streaming content at least sometimes. 54% of pauses last one to five minutes. Between 60% and 67% of viewers, depending on generation, would rather see a pause ad than a frozen screen. Worth knowing that DIRECTV sells pause ads and sponsored the write-up.

Index Exchange, June 2026. Roughly two thirds of surveyed streaming publishers are live with or testing pause ads, and 80% of agencies and DSPs expect to increase pause ad spend next year. The survey does not disclose how many people it asked.

Numbers to be careful with

The 8-person study. That NBCUniversal case study also contains the quotes about viewers finding pause ads acceptable and unobtrusive, and those come from a qualitative study the document itself sources as n=8, conducted over virtual one-on-one interviews in Q3 2021. Eight people. A good chunk of the industry's confidence that viewers like this format traces back to eight people in a video call five years ago. The measured 41% memorability number is solid. The "viewers love it" narrative built on the same PDF is eight people.

"Nearly twice the attention." The Wunderkind result got reported that way. The actual figure across all 13 verticals is 69%. The two-times number only holds in the strongest categories. Both are good results. Only one of them is the headline.

"51% of viewers take action after seeing a pause ad." This is attributed to the Video Advertising Bureau and repeated by nearly every page on this topic. We went looking for the study: the sample, the year, the definition of "action." We could not find it. If you can, we will link it here.

"91% aided brand recall, 98% message recall, 24-second average dwell, 100% viewability." From a TripleLift and MediaScience eye-tracking study, and these are the numbers that currently answer "are pause ads effective?" in Google's own results. No sample size published, no date, no methodology. TripleLift's own blog elsewhere says 92% rather than 91%, which tells you something about how carefully these get handled.

The 41% versus 43% muddle. NBCUniversal's PDF says 41% memorability. Various write-ups attribute 43% memorability to NBCUniversal, and a separate 43% to a FreeWheel and TripleLift study, and a 43% lift in site visitation to NBCUniversal as well. These are different claims from different studies being blended into one statistic. If you see 43% next to NBCUniversal's name, somebody copied it wrong.

None of this means pause ads do not work. The Peacock memorability result and the TVision attention measurement are both real, both methodologically legible, and both point the same direction. It means the format's evidence base is thinner than its marketing, which is true of most ad formats and is never admitted by anyone selling one.

What pause ads cost

Nobody knows yet, and that is the most honest answer available.

Index Exchange surveyed the market in June 2026 and found the gap: buyers expect to pay under a $15 CPM. Publishers value the inventory at $15 to $25 or higher. That is a bid/ask spread, not a price, and it is what a market looks like before it has settled.

The reason it has not settled is in the same survey. 70% of buyers named the lack of standardisation as the main obstacle. Every platform has its own dimensions, its own delay, its own definition of an impression, and its own two-week human review queue. You cannot run one creative across six services, so you cannot buy at scale, so the price stays argued about. This is exactly the problem the IAB Tech Lab's guidelines are meant to fix, and why the comment period closing today matters more than it sounds.

The case against pause ads

Every vendor page on this topic skips this section. It is the most interesting one.

Start with who pauses and why. The industry framing is that the viewer walked away, so the screen is free. But some people pause constantly and are still very much watching. From a Roku user, on being told the platform was considering more pause advertising: "I am hard of hearing and pause often to read captions." Pausing to re-read a line, to catch a name, to look at something on screen, these are not moments of absence. They are moments of unusually close attention. Selling them is a different transaction than the one being described in the decks.

Then there is the trust problem. In May 2025 a rumour spread that Roku was going to show pause ads over HDMI inputs, meaning over your games console or your Blu-ray player. Roku publicly denied it. But the rumour moved as fast as it did because it was completely plausible to people, and that reaction is a data point about where the format's goodwill actually sits. The Reddit thread on streamers pushing pause ads harder runs to nearly 400 comments and it is not warm.

And buyers are not uniformly sold either. The recall figures have been questioned by agency people on the record, which is unsurprising given how few of those figures come with a sample size attached.

Our own read: the format is genuinely better than a pre-roll for the viewer, and that is a low bar it clears easily. It stops being better the moment somebody decides that if a pause is worth one ad, a pause is worth an ad every time. The thing that makes a pause ad tolerable is that the viewer caused it. That property does not survive being maximised.

The rights problem nobody writes about

Here is a problem we ran into building this, and it applies to anyone thinking about generated ad creative.

Our ad generator will turn any product page URL into a finished ad. Paste a link, get a video and a pause still. It works well. It also does not know, and cannot know, whether the person pasting the link has any relationship with the brand on the other end of it.

Somebody can paste a URL for a brand they do not represent and get back a broadcast-ready ad for it, complete with logo, product photography and a call to action. Airing that on a consumer TV app on the say-so of an anonymous visitor is a trademark problem, and it is not one you can solve with better software.

So we did the boring thing. Every creative in our system defaults to not approved and to rights not confirmed. The serve endpoint requires both flags to be open before it will return an ad to a player, and the admin console refuses to set a creative live while the rights flag is off. Nothing airs because a job finished. Something airs because a person looked at it and took responsibility for that decision.

Which, we suspect, is what those twelve-business-day turnarounds at Disney and two-week queues at Roku actually are. Not a rendering pipeline. A person, checking.

Running one on your own channel

Every other page written about pause ads is written for advertisers buying them or platforms selling them. Almost nothing is written for the person who runs the channel. So, briefly.

A pause ad needs three things: a player that can tell a real pause from a stall, a still image, and a way to count that it was shown. That is a smaller technical problem than most of what running a channel involves. The hard parts are commercial. You need advertisers, and you need to be sure each one has the right to the brand in their creative.

On My TV Channel, pause ad serving is built into the player on iOS, iPadOS and macOS today. Free-to-air channels can carry advertising and the platform takes 0% commission, so ad revenue goes to whoever runs the channel. Paying subscribers never see an ad. Apple TV, Android, Roku and Fire TV are not wired up for pause ads yet.

FAQ

Sources

• Hulu, "Hulu Unveils New Pause Ad Experience", 31 January 2019. Launch date, Coca-Cola and Charmin, the 5-second delay.
• Digiday and TechCrunch, 31 January 2019. The Helfand quote, the non-disruptive revenue goal, the TV-MA exclusion.
• NBCUniversal, Pause Ad Innovation Case Study (PDF), December 2023. The 41% memorability figure, the 66% of pauses over 60 seconds, and the n=8 qualitative source line.
• Index Exchange, "Streaming TV pause ads are ready to scale if the industry can standardize", 18 June 2026. CPM expectations, the standardisation obstacle, publisher and buyer intent.
• TV Tech on Wunderkind Ads, OpenGlass and TVision, June 2026. The 69% attention figure and the vertical breakdown.
• eMarketer and Digiday on the Magna and DIRECTV study, July 2025. Pause behaviour and duration. Sponsored by DIRECTV.
• IAB Tech Lab CTV Ad Portfolio. The format definition and the guidelines under public comment.
• Platform specs: Google Ads Help (YouTube), Amazon Ads (Prime Video), Disney Advertising, Roku Advertising, Paramount.
• The Desk, May 2025. Roku's denial of the HDMI pause ad rumour.

If a figure here is wrong, or you can point us at that Video Advertising Bureau study, tell us and we will fix it.