# 32 Video Advertising Statistics for 2026: Every View Has a Different Denominator

> Current spend, cost, attention, and conversion data with the counting rules attached.

- Author: Rishikesh Ranjan · Published: Sep 18, 2026
- Type: Essay
- Tags: Metrics, Acquisition, User Behaviour
- Growth levers: Acquisition (primary), also Revenue
- ~1821 words

---

Paid video took a larger share of the advertising market in 2025. [IAB and PwC](https://www.iab.com/wp-content/uploads/2026/04/IAB_PwC_Internet_Ad_Revenue_Report_Full_Year_2025_April_2026.pdf) measured $78.0 billion in U.S. digital video advertising revenue and reported a $15.8 billion increase in one year. Our calculation from the report's displayed rounded totals puts video's share of the entire increase in U.S. internet ad revenue at about 44.2%. The market signal is clear. The performance signal is not, because an impression, a viewable impression, a platform view, a completion, and an incremental conversion count different events.

These 32 video advertising statistics keep the denominator, population, and measurement year attached. The 2026 label describes when this collection was assembled. Most observed campaign and revenue data comes from 2025 or 2024. For the wider category, use the [video marketing statistics](https://www.productgrowth.blog/p/video-marketing-statistics) pillar. For organic and paid vertical formats, see the [short-form video statistics](https://www.productgrowth.blog/p/short-form-video-statistics) analysis.

| Metric | Value |
| --- | --- |
| statistics | 32 |
| source records | 13 |
| of 2025 revenue growth came from video | 44.2% |

> **How to read the list:** Reported means the publisher supplied the number. Vendor-measured means the company observed it in its own inventory. Official definition describes a counting rule. Our calculation uses compatible inputs and shows the arithmetic. None of those labels turns a market average into your account's benchmark.

## The market grew faster than its measurement language

The strongest current market record is the full-year revenue series, not a 2025 forecast that was published before the year finished. IAB and PwC collect revenue reported by companies that sell internet advertising and supplement it with public corporate data. The report also says PwC does not audit those submissions. That makes it an industry benchmark with a declared method, not a census with assurance.

| No. | Statistic | What it tells you |
| --- | --- | --- |
| 1 | U.S. digital video advertising revenue reached $78.0 billion in 2025. | The figure includes CTV, social video, online video, and short-form video. |
| 2 | Digital video revenue grew 25.4% from 2024 to 2025. | This is measured publisher revenue, not a forecast or an advertiser ROI figure. |
| 3 | IAB/PwC reported that video added $15.8 billion in revenue in one year. | The report's displayed one-decimal totals, $62.1 billion in 2024 and $78.0 billion in 2025, imply $15.9 billion because of rounding. |
| 4 | Video represented 26.5% of U.S. internet advertising revenue in 2025. | Share describes revenue concentration by format, not time spent or impressions. |
| 5 | Video's revenue share was 24.0% in 2024. | The format gained 2.5 percentage points in one year. |
| 6 | Video's revenue share was 20.9% in 2021. | The share rose 5.6 points across four years, using one report series. |
| 7 | Our calculation: video supplied about 44.2% of the market's $36.0 billion revenue increase in 2025. | Using displayed rounded totals, $15.9B of implied video growth divided by $36.0B of total market growth equals 44.17%, rounded to 44.2%. |
| 8 | The ten largest companies held 84.1% of U.S. internet advertising revenue in 2025. | A market average is heavily shaped by scaled platforms and does not describe every auction. |
*Statistics 1 to 8. Source: IAB/PwC Internet Advertising Revenue Report, Full Year 2025.*

![Using the IAB/PwC report's displayed rounded totals, the 36.0 billion dollar increase in U.S. internet advertising revenue from 2024 to 2025 comprised 15.9 billion dollars of implied digital video growth and 20.1 billion dollars from other formats. The implied video share is about 44.2%.](https://www.productgrowth.blog/media/posts/video-advertising-statistics/01-incremental-revenue.webp)
*Displayed rounded IAB/PwC totals imply that video supplied about 44.2% of the 2025 increase. The report's narrative gives $15.8 billion as the unrounded video increase.*

That 44.2% is a composition calculation. It says where a large share of new measured revenue landed. It does not say video caused market growth, delivered 44.2% of new attention, or returned more profit than search. The distinction matters because video and social categories overlap, and higher revenue can come from more inventory, higher prices, or both.

Use the market series to answer a category question: is paid video attracting more advertiser money? The answer is yes. Do not use it to answer an account question such as whether your next dollar belongs in connected TV, a social feed, or search. That choice needs marginal cost, audience quality, and an outcome measured inside the same campaign. Market revenue can justify investigating the channel. It cannot justify a budget transfer by itself.

## Cost moved with inventory, format, and screen

Public cost benchmarks become unreliable when the objective, screen, and inventory mix disappear. [Tinuiti's Q2 2025 benchmark](https://tinuiti.com/research-insights/research/digital-ads-benchmark-report-q2-2025/) avoids one common trap by reporting same-client changes from programs that remained active with a consistent strategy. Its annual managed spend exceeds $4 billion, but its clients are still a selected agency population.

| No. | Statistic | Denominator or limit |
| --- | --- | --- |
| 9 | 68% of IAB's 368 surveyed U.S. TV/video buyers called CTV a must-buy in 2025. | Buyer opinion for a largest or most important brand, not observed return. |
| 10 | 62% called social video a must-buy. | The same IAB buyer survey; social video can overlap the wider digital-video revenue category. |
| 11 | 37% called online video a must-buy. | A channel-priority response, not a market-spend share. |
| 12 | Tinuiti clients increased YouTube video spend 9% and impressions 10% year over year while average CPM fell 1% in Q2 2025. | Same-client change, not an absolute CPM available to every advertiser. |
| 13 | TV screens took 55% of YouTube video spend, up from 37% a year earlier. | Tinuiti-managed YouTube programs, Q2 2024 to Q2 2025. |
| 14 | Phones took 31% of YouTube video spend, down from 46% a year earlier. | A device-mix shift inside the same agency dataset. |
| 15 | Shorts represented 18% of segmented YouTube video spend, second to skippable in-stream at 63%. | Roughly half of YouTube spend was unsegmented, so these are not shares of all spend. |
| 16 | Reels supplied 21% of Instagram ad impressions in Q2 2025, up from 13% a year earlier. | Tinuiti client impressions, not all Instagram advertising. |
*Statistics 9 to 16. Sources: IAB 2025 buyer research and Tinuiti Q2 2025 same-client benchmarks.*

![Tinuiti-managed YouTube video spend shifted by screen from Q2 2024 to Q2 2025: TV screens rose from 37% to 55%, while phones fell from 46% to 31%.](https://www.productgrowth.blog/media/posts/video-advertising-statistics/02-youtube-screen-shift.webp)
*TV screens gained 18 percentage points of Tinuiti-managed YouTube video spend while phones lost 15 points. Other screens and unsegmented activity are not shown.*

> **There is no universal video CPM:** A useful cost comparison holds objective, geography, placement, device, audience, period, optimization strategy, and attribution window steady. A CTV reach buy, a Shorts view campaign, and a retargeting conversion campaign may all report CPM, but the impressions do different jobs.

Start a cost review by separating price from productivity. CPM tells you what one thousand impressions cost, not whether those impressions were viewable, reached new people, or produced a qualified action. CPV adds a viewing event, but the platform still decides what qualifies as a view. Cost per completed view gets closer to attention while favoring shorter creative. Cost per qualified visit or incremental conversion gets closer to the business decision, although it needs more volume and cleaner measurement. Keep the earlier costs as diagnostics, then judge the campaign on the latest reliable event in its outcome chain.

## Delivery is not attention

A video can render, meet a viewability threshold, finish playing, and still fail to reach a person who notices it. The [MRC viewability standard](https://mediaratingcouncil.org/sites/default/files/Standards/081815%20Viewable%20Ad%20Impression%20Guideline_v2.0_Final.pdf) establishes an opportunity to see. The newer [IAB and MRC attention framework](https://www.iab.com/guidelines/attention/) treats attention as a complementary signal that still needs an outcome beside it.

| No. | Statistic | What it measures |
| --- | --- | --- |
| 17 | IAS measured global video viewability at 79.7%, versus 67.9% for display, in its 2026 report. | An 11.8 percentage-point gap in IAS-measured inventory, not proof of attention or sales. |
| 18 | For in-stream video, the MRC baseline requires at least 50% of the ad's pixels in an in-focus browser tab for two continuous seconds. | A disclosed legitimate-click proxy can also qualify; viewability is an opportunity to see, not a completed view. |
| 19 | Innovid reported an average CTV campaign frequency of 7.09. | Average exposures in Innovid's 2024 CTV campaign data, not a recommended target. |
| 20 | The same Innovid dataset reported average CTV household reach of 19.64%. | Reach and frequency together suggest how concentrated delivery was. |
| 21 | Innovid said interactive CTV ads earned 71 additional seconds on average over standard pre-roll. | A vendor-measured interaction result across formats such as galleries and QR overlays. |
| 22 | Only 50% of DoubleVerify-measured CTV impressions had full app transparency in 2024. | Half of measured delivery lacked a complete app-level placement signal. |
| 23 | DoubleVerify estimated $700,000 of waste per billion CTV impressions without safeguards. | A modeled cost from the vendor's media-quality definitions, not an audited market total. |
| 24 | Bots represented 65% of detected CTV fraud, with four million infected devices generating fake traffic daily. | DoubleVerify's detected-fraud mix and device estimate, not 65% of all CTV impressions. |
*Statistics 17 to 24. Sources: IAS 2026, MRC, Innovid 2025, and DoubleVerify 2025. Vendor populations differ and are not averaged.*

![A connected-TV delivery path with separate vendor diagnostics. Innovid reports 7.09 average frequency and 19.64% average household reach. DoubleVerify reports 50% full app transparency, 700,000 dollars modeled waste per billion impressions without safeguards, bots as 65% of detected CTV fraud, and four million infected devices generating fake traffic daily.](https://www.productgrowth.blog/media/posts/video-advertising-statistics/04-ctv-diagnostics.webp)
*CTV delivery can complete while reach, placement transparency, and human validity remain unresolved. The Innovid and DoubleVerify figures use different vendor populations and stay separate.*

The useful operating sequence is delivery before persuasion. First ask whether the impression was valid and where it ran. Then ask whether the intended household or person had an opportunity to see it. Only after that should completion, response, and conversion enter the review. A high completion rate cannot repair unknown placement or repeated delivery to a narrow audience.

This sequence also makes troubleshooting cheaper. If app transparency is missing, fix supply controls before asking the creative team for another cut. If frequency rises while household reach stalls, examine duplication and audience constraints before raising the bid. If delivery is valid and reach is healthy but viewers leave early, creative becomes the sharper hypothesis. Each stage rules out a different failure. Combining them into one score hides where the campaign actually broke.

## A view can mean playback, two seconds, ten seconds, or 30

![Four separate video measurement clocks: the MRC in-stream baseline uses 50% of pixels in an in-focus browser tab for 2 continuous seconds or a disclosed legitimate-click proxy; TikTok records playback plus separate 2-second and 6-second views; YouTube Shorts counts at 10 seconds, completion if shorter, or a call-to-action, Like, Comment, or Share click; YouTube skippable in-stream counts at 30 seconds, completion if shorter, or a qualifying ad-element click.](https://www.productgrowth.blog/media/posts/video-advertising-statistics/03-view-clocks.webp)
*Four events commonly shortened to view. The thresholds and qualifying clicks differ, so the resulting rates are not one comparable unit.*

The official definitions make the denominator problem concrete. [Google's current help page](https://support.google.com/google-ads/answer/2375431?hl=en) counts formats differently, while [TikTok's metric guide](https://ads.tiktok.com/resources/help/article/video-play?lang=en) exposes playback, two-second, six-second, and completion events as separate measures. A cross-platform dashboard should preserve those names instead of renaming every column Views.

| No. | Statistic | Counting rule |
| --- | --- | --- |
| 25 | A YouTube skippable in-stream TrueView view counts at 30 seconds, at completion for a shorter ad, or after a qualifying click on an ad element. | A qualifying ad-element click can count before the time threshold. |
| 26 | A YouTube Shorts ad view counts at 10 seconds, at completion for a shorter ad, or after a call-to-action, Like, Comment, or Share click. | The Shorts threshold and qualifying click paths differ from in-stream. |
| 27 | TikTok's basic video-view metric counts when a video starts playing; its two-second view is a separate metric. | Playback and two-second retention should not share one label. |
| 28 | A TikTok six-second view can count after six seconds, completion for a shorter video, or an engagement within the first six seconds. | Time and interaction can lead to the same reported event. |
*Statistics 25 to 28. Current official definitions accessed September 18, 2026.*

| No. | Statistic | Why it is scoped |
| --- | --- | --- |
| 29 | RhythmInfluence reported Instagram completion of 5.35% under CPM optimization and 13.59% under CPV optimization in 2025. | Paid creator campaigns in one vendor dataset; the public report does not disclose sample size. |
| 30 | It reported TikTok six-second completion of 15.24% under CPM optimization and 23.72% under CPV optimization. | The denominator is six-second views, not impressions, and optimization selects for viewing behavior. |
*Statistics 29 and 30. Source: RhythmInfluence 2026 benchmarks, covering its 2025 creator campaign data.*

Those completion figures are useful because they disprove the idea that a platform owns one benchmark. Changing the bidding objective changes the audience and delivery the system selects. The public RhythmInfluence PDF also omits its campaign and impression sample sizes, so the values belong in a directional planning conversation, not a universal scorecard.

The repair is simple but easy to skip: preserve the platform's event name in the data model. Store TikTok six-second views separately from basic video plays. Store YouTube Shorts TrueView views separately from skippable in-stream views. Then calculate rates from the event and denominator you intend to compare. A dashboard may still show a friendly label, but its definition should remain one click away. Otherwise a change in format mix can look like a change in creative performance even when viewer behavior stayed the same.

## Conversion evidence has three levels

A dashboard can report that a conversion happened after an ad. A controlled lift study asks whether the conversion would have happened without the ad. Profit adds the final question: did the incremental value exceed media, production, and operating cost? Public video advertising statistics often slide between those levels.

| No. | Statistic | What it can support |
| --- | --- | --- |
| 31 | 82% of Wyzowl's marketer respondents said video marketing delivered good ROI, but 67% quantified ROI with views and only 32% with bottom-line sales. | Self-reported belief and measurement practice. Wyzowl reports 266 total respondents across marketer and consumer paths but does not disclose the marketer subsample size. |
| 32 | TikTok says more than 1,400 advertisers have run Conversion Lift Studies; a 2024 TransUnion meta-analysis found 52% of TikTok-led incremental conversions were exclusive to TikTok reach. | A platform-reported randomized-study summary, not a transferable conversion rate; the public page omits the meta-analysis sample distribution. |
*Statistics 31 and 32. Sources: Wyzowl's late-2025 survey, with 266 total respondents across marketer and consumer paths and no disclosed marketer subsample size; TikTok's Conversion Lift Study summary.*

![Metrics Wyzowl marketer respondents use to quantify video ROI: views 67%, engagement 63%, leads or clicks 52%, customer engagement or retention 40%, brand awareness or PR 36%, and bottom-line sales 32%.](https://www.productgrowth.blog/media/posts/video-advertising-statistics/05-roi-measures.webp)
*Views were selected 35 percentage points more often than bottom-line sales as a way to quantify video ROI. Respondents could select different measures; the choices are not exclusive.*

Wyzowl's result does not show that views are useless. It shows that the word ROI is carrying more weight than the underlying measurement. Views can diagnose creative and distribution. They cannot reveal incremental profit without a downstream outcome, a counterfactual, and cost. TikTok's lift-study program moves closer to causality by randomizing test and control groups, but its favorable summary still does not give another advertiser a conversion-rate target.

Match the strength of the claim to the design. A platform attribution report can describe credited conversions under its stated window. A holdout can estimate incremental conversions for the tested population and period. Neither automatically supplies profit, because profit also depends on margin, fulfillment, creative production, and measurement cost. When a lift test is too expensive or underpowered, say that the evidence is directional and choose the next decision it can support. Honest uncertainty is more useful than a precise conversion rate borrowed from someone else's account.

## Build a benchmark without fooling yourself

The safest benchmark is usually your own matched history. Write the comparison header before looking at the result. If any field changes, explain the change instead of calling the movement an improvement or decline.

Build the first benchmark from a period with stable tracking and enough observations to make ordinary variation visible. Segment only where the segment changes a decision, such as screen, format, prospecting status, or geography. More cuts can make the report look rigorous while leaving each cell too thin to interpret. Keep a short change log beside the series for creative launches, bid changes, audience edits, consent changes, and attribution updates. When the line moves, that record gives the team hypotheses it can test instead of a story it has to invent.

| Field | Write down | Why it matters |
| --- | --- | --- |
| Event | Impression, viewable impression, platform view, completion, click, or conversion | Prevents one label from hiding different actions. |
| Denominator | Served impressions, measurable impressions, views, clicks, people, or households | Determines what the rate can honestly describe. |
| Population | Country, audience, customer status, and eligibility | Stops a broad average from replacing the audience you bought. |
| Placement | Platform, format, screen, device, and inventory source | Separates CTV, feed, Shorts, in-stream, and other delivery contexts. |
| Period | Campaign dates and complete observation window | Controls for seasonality and late conversions. |
| Optimization | Bid objective and event the platform was trained to find | Explains why CPM and CPV campaigns can produce different completion rates. |
| Attribution | Window, click or view rule, and incrementality method | Separates credited conversions from conversions caused by the ads. |
*A denominator header for any video advertising benchmark.*

1. Validate delivery. Confirm valid traffic, placement transparency, measurability, and intended geography.
2. Measure opportunity to see. Keep viewability and household or person reach separate from served impressions.
3. Measure attention. Use watch time, hold, or a disclosed attention method that matches the format.
4. Measure response. Track completion, destination clicks, and qualified on-site behavior as separate events.
5. Measure attributed outcomes. State the window and whether the credit came from a click or a view.
6. Test incrementality. When the decision is large enough, compare exposed and control groups or use another defensible counterfactual.
7. Close the cost loop. Compare incremental value with media, production, measurement, and operating cost.

> **Steal this:** Steal this: Put the denominator in every dashboard label. Write 10-second Shorts views per impression, not video engagement. Write attributed purchases within seven days of a view, not video ROI. The longer label prevents the more expensive argument later.

Video advertising deserves the budget attention implied by its market growth. It also deserves stricter language. The useful conclusion from these statistics is not that video always works or that one channel wins. It is that spend has moved faster than metric standardization. Teams that preserve the counting rule can learn across campaigns. Teams that collapse every early signal into views and every credited action into ROI will keep buying certainty that the data never supplied.

For the next review, pick one disputed number and trace it backward. Find the event that created it, the people or devices eligible to enter its denominator, the placement where it occurred, and the rule that connected it to an outcome. Then ask what decision would change if the number moved. If the team cannot answer those questions, the metric is not ready to steer spend. Repairing that definition is often a better experiment than chasing another outside benchmark.

## Video advertising statistics FAQ

#### How much was the U.S. digital video advertising market worth in 2025?

IAB and PwC measured $78.0 billion in 2025 revenue, up 25.4% from $62.1 billion in 2024. The category includes connected TV, social video, online video, and short-form video.

#### What is a good video ad completion rate?

There is no platform-wide number that survives a change in format, length, audience, bidding objective, and view definition. RhythmInfluence's scoped creator-campaign data shows the problem: completion differs materially between CPM and CPV optimization on the same platform. Build a matched internal benchmark first.

#### What counts as a video ad view?

It depends on the platform and format. Google counts a skippable in-stream TrueView view at 30 seconds, completion for a shorter ad, or a qualifying ad-element click. A Shorts view uses a 10-second threshold, completion if shorter, or a call-to-action, Like, Comment, or Share click. TikTok exposes playback, two-second, six-second, and completion events separately. For in-stream video, the MRC viewability baseline is 50% of pixels in an in-focus browser tab for two continuous seconds, with a disclosed legitimate-click proxy also allowed.

#### Do video ad views prove ROI?

No. Views can diagnose delivery and creative response. ROI needs business value and cost. Incremental ROI also needs a credible estimate of what would have happened without the ads, usually through an experiment or another defensible counterfactual.

**Next job: Audit one video campaign's denominator.** Take the metric your team calls a view or conversion and write its exact event, denominator, population, placement, period, optimization, and attribution window. Rename every ambiguous dashboard column before the next budget review.

---

All posts: https://www.productgrowth.blog/archive · Site: https://www.productgrowth.blog