Completion rate is the proportion of video or audio advertisement impressions that play through to the final frame, expressed as a percentage. It exists because a video advertisement, unlike a banner, has a duration. An impression records that a file was requested and began to render; completion rate claims something stronger, that the file ran to its end.

That claim carries commercial weight. It underpins a buying model, cost per completed view, in which advertisers pay only for impressions reaching the last frame, and it steers optimisation inside demand-side platforms, the systems that bid for advertisers. In connected television it is the most disputed number in the market, because the formats dominating streaming supply make near-total completion unavoidable.

How the number is produced

The measurement runs on the Video Ad Serving Template, or VAST, the XML specification maintained by IAB Tech Lab that tells a player which file to fetch and what to report back. Inside a VAST document, a TrackingEvents node holds measurement URLs bound to named events. The relevant ones are ordinal: start, firstQuartile, midpoint, thirdQuartile and complete. A separate progress event takes an offset attribute accepting a timestamp, a percentage, or the keywords start and end, placing a beacon anywhere in the timeline rather than at a fixed quarter.

The player fires each URL as playback crosses each point. Completion rate is complete beacons over a denominator, and the denominator is where products diverge. Google's Ad Manager API enumerates the same VAST events and adds two with no equivalent in the specification: SKIP_SHOWN, fired when a skip control appears, usually after five seconds, and ENGAGED_VIEW, fired after 30 seconds of playback or at the end if the advertisement is shorter.

Two products illustrate the ambiguity. The Google Ads API defines video_completion_p100_rate as the percentage of impressions where the viewer watched the entire video, and trueview_view_rate as completed TrueView views over impressions. Yahoo! JAPAN Ads took the other route: the video view through rate it added on December 17, 2020 measures playback to 100 per cent of total length against the number of videos initiated. Starts, not impressions. A campaign losing viewers to slow loading reports a materially better figure on a starts basis, and reporting interfaces rarely state which is in use.

The MRC and IAB video guidelines set the technical floor rather than the arithmetic: duration measured at second granularity at minimum, segmented by progress events such as completions, quartiles or deciles, and presented against viewability. Viewability is a separate test, at least 50 per cent of pixels on screen for two continuous seconds for video against one second for display.

Server-side ad insertion complicates the chain. Where advertisements are stitched into the content manifest on a server rather than called by the player, completion beacons often fire from the stitcher, describing what the server transmitted rather than what a device rendered. IAB Tech Lab's Open Measurement SDK addresses that gap by running code on the device to measure viewability, duration and playback completion, and by passing signals on device and session state including television power status, as IAB Europe set out in April 2026.

Where completion sits in the transaction

Completion rate became a currency, not just a diagnostic. Cost per completed view, or CPCV, bills per finished playback rather than per thousand impressions. Teads built its reputation on the model for outstream video and later extended cost per completed view and viewable CPM to publishers in August 2023, having earlier billed only impressions clearing a five-second viewability threshold. AerServ opened a programmatic mobile marketplace transacting on completed views in July 2017, aimed at rewarded video in games, where completion is guaranteed by design.

Platform reporting follows the quartile structure. Microsoft's Premium Streaming campaigns, launched in August 2025, report views at 25, 50 and 75 per cent alongside completed views and completion rates, and bid on cost per completed view.

Origin and evolution

The infrastructure predates the current controversy by two decades. In November 2004 the IAB's Measurement Task Force issued a global standard for counting online advertisement impressions. Video needed separate treatment, and the IAB Digital Video Committee and the MRC released digital video impression guidelines in December 2009, prompted by advertisements playing automatically on page load.

The delivery layer was standardised in 2008, when the IAB published VAST 1.0 alongside VPAID. VAST 2.0 added quartile tracking, the moment completion rate became a portable, cross-vendor number rather than a proprietary ad server statistic. VAST 3.0, in 2012, introduced skippable advertisements through the skipoffset attribute.

The economics shifted on December 1, 2010, when YouTube launched TrueView, letting viewers skip after five seconds and charging only for views that were not skipped. Skippability made completion a variable. VAST 4.0 followed in January 2016, with 4.1 in November 2018 adding Open Measurement support, and 4.2 in June 2019 adding SIMID. IAB Tech Lab published the VAST CTV Addendum in July 2024, extending registered ad identifiers and high-resolution files to older VAST versions for large-screen environments.

Why the number matters now

Connected television has concentrated the argument. Streaming advertising spending reached $33.35 billion in 2025, and survey data reported in January 2026 put the channel at 26 per cent of media budgets on average. Completion rate is what sellers there lead with, because the numbers are extraordinary by open-web standards.

FreeWheel research from September 2023, cited by the Video Advertising Bureau, put the average completion rate for long-form premium video at 94 per cent. Format explains most of the variance. Spotter data published in April 2026reported 15-second skippable advertisements on creator content reaching 70 per cent on connected television against 57 per cent elsewhere, and 60 against 39 per cent for 30-second skippable formats. Non-skippable 15-second units returned 96 against 90 per cent. Those figures are self-reported, and the spread between skippable and non-skippable is the point: the top of the range is a property of the unit, not the creative.

Limitations and disputes

The criticism is explicit and named. Vlad Chubakov, Associate Director of Programmatic at Delve Deeper, argued in a November 2025 post that completion rate and cost per completed view provide minimal actionable intelligence, observing that streaming completion rates almost always reach 99.99 per cent and that cost per completed view functions as a proxy for CPM. Some marketers use it to show they are meeting KPIs, he wrote, but it is meaningless for optimising performance or quality.

Austin Lake, a paid media director, pressed the same point in June 2026: 98 per cent completion on non-skippable inventory is the structural baseline of the medium, and optimising toward it steers budget to the cheapest supply that logs a finished play. Dr. Augustine Fou of FouAnalytics added the conflict-of-interest objection in the same thread, arguing advertisers should not trust performance reports supplied by the vendor selling the inventory.

Verification data gives the argument teeth. DoubleVerify found that more than one in three connected television impressions deliver to a screen that is switched off, an estimated $1 billion in annual waste, with the HDMI-CEC protocol capable of signalling power state but almost never used in measurement logic. Its 2026 Global Insights report, published on May 7, 2026, recorded a 140 per cent rise in streaming fraud schemes in the first quarter and found 34 per cent of monitored impressions appearing outside streaming content, on music, fitness and screensaver applications. A completion beacon fires in all of those environments.

Sourcing discipline is weak too. A Lifesight report of August 12, 2026 set a 96 per cent completion figure against 16.4 per cent for standard video on one page and 53 per cent for social media on another, a threefold shift in the comparison base inside one document. Jamloop research published on July 9, 2026 found 62 per cent of buyers doubt connected television performance claims, with more than 70 per cent saying they would spend more if measurement improved.

Distinguishing completion rate from adjacent metrics

View rate counts views rather than completions. A TrueView view has historically meant 30 seconds of playback or the full advertisement if shorter, so view rate runs high while completion runs low on longer creative. The definitions have shifted repeatedly, which is why the two are conflated.

View-through rate has no settled meaning. Yahoo! JAPAN Ads uses it for 100 per cent completion; much of the market uses it for conversions attributed after an impression with no click, which measures outcome rather than delivery.

Viewability rate measures opportunity to see and is orthogonal to completion. An advertisement can complete off-screen, and can be fully viewable while being skipped.

Engaged view is an attribution threshold rather than a delivery one. Engaged-view conversions credit a conversion when a viewer passes roughly 10 seconds without clicking.

Outside advertising the term names something else: in survey research it is a synonym for response rate, and in web analytics the share of started forms or checkouts that are finished.

Recent developments

The market is moving around the metric rather than fixing it. OpenX and TVision launched pre-bid connected television attention targeting on March 11, 2026, turning attention from a post-campaign diagnostic into a buying signal for streaming inventory. IAB Europe's April 2026 guide argued that CPM alone is not a reliable value proxy in the channel, identifying completion-rate stability, attention weighting and outcome proxies as more meaningful. Its March 2026 conversion-gap analysis named missing identifier standards and cross-device attribution as the blockers.

Verification itself is changing hands. Nielsen agreed to acquire DoubleVerify for $2.15 billion, closing targeted for the first quarter of 2027, placing the largest independent auditor of streaming delivery inside the largest television measurement company.

Timeline

  • November 2004: IAB issues a global standard for counting online advertisement impressions
  • 2008: IAB publishes VAST 1.0 alongside VPAID
  • December 2009: IAB Digital Video Committee and the MRC release digital video advertisement impression guidelines addressing auto-play
  • December 1, 2010: YouTube launches TrueView, charging only for skippable advertisements that are not skipped
  • 2012: VAST 3.0 introduces skippable advertisements through skipoffset, plus ad pods
  • January 2016: VAST 4.0 separates media files from verification and interactive code
  • July 18, 2017: AerServ announces a programmatic mobile video marketplace transacting on cost per completed view
  • November 2018: VAST 4.1 adds Open Measurement support and merges the audio standard
  • June 2019: VAST 4.2 adds SIMID for interactivity
  • December 17, 2020: Yahoo! JAPAN Ads adds a video view through rate metric measured against video starts
  • September 2023: FreeWheel research records a 94 per cent average completion rate for long-form premium video
  • July 2024: IAB Tech Lab publishes the VAST CTV Addendum
  • August 2025: Microsoft launches Premium Streaming campaigns using cost per completed view bidding
  • November 4, 2025: Vlad Chubakov publishes a post identifying completion rate and cost per completed view as misleading connected television KPIs
  • December 4, 2025: DoubleVerify states that over one-third of streaming impressions deliver in television-off environments
  • March 11, 2026: OpenX and TVision launch pre-bid connected television attention targeting
  • April 2026: IAB Europe publishes its programmatic connected television guide
  • May 7, 2026: DoubleVerify reports a 140 per cent rise in connected television fraud schemes in the first quarter of 2026
  • June 29, 2026: Austin Lake's argument that a 98 per cent completion rate is automatic rather than earned is reported
  • August 12, 2026: A Lifesight report circulates with two different comparison bases for the same completion figure

Summary

Who. Video players and ad servers fire the beacons; supply-side platforms and publishers report the resulting rate; demand-side platforms optimise against it; IAB Tech Lab and the MRC define the events and the measurement floor; verification vendors including DoubleVerify audit whether the impressions behind it were real.

What. Completion rate is the share of video or audio advertisement impressions that play to the final frame, counted from the VAST complete event and divided by either impressions or video starts depending on the platform.

When. The counting infrastructure dates to the IAB impression standard of November 2004 and the digital video guidelines of December 2009, with quartile tracking standardised in VAST 2.0 and skippability introduced by YouTube TrueView on December 1, 2010. The metric's credibility became a public dispute between November 2025 and mid-2026.

Where. It applies wherever a timed advertisement plays: instream video on the open web, in-app and rewarded video, digital audio, and above all connected television, where non-skippable formats push the figure toward its ceiling.

Why. Completion rate is the basis of a buying model and a common contractual guarantee, which makes its definition commercially consequential. Where the format cannot be skipped, the number describes the unit rather than the audience, and reads as a measure of delivery mechanics rather than of attention.