Begin-to-render is the point in the ad delivery process at which an impression becomes valid. Under the rule, an advertisement counts only once the creative file has reached the user's device and the browser or application has started drawing it. Everything earlier in the chain, including the ad server's decision, the bid response and the arrival of the tag, counts as nothing.

The rule exists because the older method counted too early. For most of the display market's history, an impression was recorded when an ad server responded to a request, long before anyone could establish whether the advertisement had arrived. Users navigated away, tags failed, creatives never resolved, and the count carried on regardless. Moving the trigger later closed part of that gap.

What the standard requires

The controlling text is the Desktop Display Impression Measurement Guidelines, version 7.1, issued in October 2017 by the Media Rating Council with the IAB Technology Laboratory. It defines an ad impression as the measurement of responses from an ad delivery system to a request from the user's browser, filtered for invalid traffic and recorded as late as possible in the delivery of the creative material. The advertisement must be loaded and at minimum begin to render.

Both terms are defined narrowly. Loaded means the logical creative file has been transmitted and received on the client side. Render refers to painting the creative file or adding it to any portion of the Document Object Model. Measurement should cover the logical components needed to display the advertisement, but not inessential elements such as other tracking assets.

One clause does most of the work: the guidelines state the rule does not require any portion of the advertisement to be visible on screen. Begin-to-render establishes that a creative started drawing somewhere in the page, not that a person could have seen it. Visibility is the separate business of viewable impression measurement.

The guidelines also concede imprecision. Measurers will differ on the exact moment rendering is recorded, and vendors that cannot measure fully rendered impressions are encouraged to study the gap between render initialisation and completion rather than assume it away.

Where the count happens

Counting must be client-initiated. Server-initiated methods, described in the guidelines as count on decision or count on insertion, are prohibited because they sit furthest from the user. Mobile guidelines had already banned them in July 2013.

Permissible implementations include HTTP requests generated by an IMG, IFRAME or SCRIPT SRC element. The response can be a beacon, commonly a 1x1 pixel image though the guidelines set no restriction on media type, a 302 redirect, or the ad content itself where the creative doubles as the tracking asset. The mechanical requirement is ordering: counting must follow content delivery, never precede it.

Two ancillary rules protect the count. Cache busting is mandatory, using Expiry and Cache-Control headers plus a unique string such as a random number or timestamp, so an advertisement served from cache still registers. Pre-fetch and pre-render requests are excluded, since content loaded in anticipation of a visit that never happens has not been delivered to anyone.

Mobile applications work differently again. Where the Open Measurement Interface Definition is present, Campaign Manager 360 and Display and Video 360 count an impression when at least one pixel of the advertisement has appeared on the device screen, a threshold stricter than begin-to-render and closer to viewability. Display and Video 360 moved to that rule for OMID inventory after Google Ad Manager replaced its proprietary Active View technique with the Open Measurement SDK for mobile app display.

Origin and evolution

Impression counting was first standardised in November 2004, when the IAB issued a global rule for counting online advertisement impressions. Video diverged early: the Broadband Video Commercial Measurement Guidelines of May 2006, revised in December 2009 and renamed for digital video, qualified an impression post-buffering, which was later judged equivalent to begin-to-render. Display spent another decade counting on server response.

Pressure came from publishers. Discrepancies widened through 2016 as some demand-side ad servers began reporting rendered rather than served counts, leaving sell-side and buy-side numbers describing different events. The Modernizing Measurement Task Force, formed by the IAB in 2015 to prioritise updates to MRC digital guidelines, took up the question. Proposals went to public comment until February 1, 2017, and IAB Australia circulated its own begin-to-render measurement model in September 2017.

The MRC, with the IAB Tech Lab and the Mobile Marketing Association, released the final desktop display, mobile web and mobile in-app guidelines on October 31, 2017. References to served impressions were struck from the documents. A one-year grace period followed before compliance became an accreditation criterion. The digital video guidelines went out for comment the same day, closing on November 30, 2017 and finalised the following year.

The standard has since propagated into newer formats. The IAB and MRC Retail Media Measurement Guidelines of January 2024 require begin-to-render measurement of digital creatives, and instruct networks to study cases where an impression is reported but a product image or description fails to render. Draft augmented reality guidelines published for comment in February 2024 apply the same minimum inside an AR session.

Why it matters

Begin-to-render sets the denominator for almost everything else. Viewable impressions are drawn from rendered impressions, and there can be no more of the former than the latter. Invalid traffic rates, measured rates and viewable rates all resolve against the same base, so a change to what counts as an impression moves every ratio built on top of it.

That sensitivity showed in Google's platforms. Changes to three MRC accredited metrics on July 14, 2025 were expected to roughly double reported Invalid Begin to Render Impressions and General Invalid Traffic Begin to Render Impressions, with Inactive Impressions rising an estimated 30%. The underlying fraud had not doubled; the detection had improved, and the dashboards read as deterioration.

Accreditation follows the same vocabulary. Amazon DSP holds MRC accreditation for display begin-to-render impressions and click-throughs across third-party exchanges on desktop, mobile web and mobile in-app, plus Fire TV, granted in May 2024. Integral Ad Science became the first independent verifier accredited for third-party measurement on Amazon DSP on November 13, 2025, covering impressions, viewable impressions and invalid traffic.

Limitations and disputes

The most persistent problem is that rendering is not seeing. An advertisement can be painted into a container positioned below the fold, in a background tab, or in a slot the user scrolls past in under a second, and still qualify. The guidelines say so explicitly, yet the metric is routinely discussed as though it establishes exposure.

Definitions also diverge between the standard and its largest implementer. The MRC text places the trigger at the start of painting or insertion into the DOM. Google's documentation describes begin-to-render as recording an impression only once the creative assets have finished downloading to the browser, which is a different and earlier event. Both are stricter than counting on server response, but they are not the same rule, and reconciliation across platforms inherits the difference.

Discrepancies remain structural rather than exceptional. Google's own guidance notes that campaign variances of up to 20% are common between Display and Video 360 and Campaign Manager 360, attributing them to counting at different points, latency, invalid traffic filters applied at different times and third-party tags. The MRC's answer, the tracked ad, permits vendors to report advertisements where measurement initiated but content never rendered, provided the figure is labelled and disclosed separately. Adoption of that disclosure is uneven.

The standard also does not reach the auction. Bid requests, wins and billed impressions in real-time bidding fall outside the guidelines, so buyers can pay for supply no accredited count ever recognises.

Adjacent terms

Count-on-download. The predecessor rule, recording an impression when the ad code payload began executing in the browser or application. It sits between served and rendered, and remains in use where platforms have not completed migration.

Viewable impression. The next gate. At least 50% of pixels in view for one continuous second for display, 30% for units of 242,500 pixels or larger, and 50% for two continuous seconds for video. Every viewable impression must tie back to a valid rendered one.

Tracked ad. The MRC's segregated metric for advertisements where the measurement assets loaded but the creative did not. It may be reported alongside impressions but must not be labelled an impression without qualification.

Start render. A web performance metric, unrelated to advertising counts, measuring when a browser paints the first pixel of a page. It is derived from video capture in tools such as WebPageTest and is closer to first paint than to any impression standard.

Recent developments

Google's migration has been slower than its first announcements suggested. Display and Video 360 began shifting display impression counting from on-download to begin-to-render in September 2025, and Campaign Manager 360 followed on November 6, 2025. Help Center documentation updated on August 21, 2026 now schedules the change to billed display impression counting across both platforms for February 2027, framing it as necessary to retain MRC accreditation and warning of a slight decrease in billed volume. The public record does not resolve whether the 2025 changes covered reported metrics only or were rescheduled.

Accreditation activity has continued around the metric. HUMAN Security's viewability product earned MRC accreditation in April 2026 for display and video, filtering invalid traffic before reporting viewability rather than after. The MRC and IAB released attention measurement guidelines in November 2025, extending qualification a step beyond visibility, while the council's out-of-home work has reopened the question of what an impression means where no browser renders anything.

Timeline

  • November 2004: The IAB issues a global standard for counting online advertisement impressions
  • May 2006: The IAB and MRC publish the Broadband Video Commercial Measurement Guidelines, qualifying video impressions post-buffering
  • December 2009: The video guidelines are revised and renamed the Digital Video Ad Impression Measurement Guidelines
  • July 2013: Mobile web and mobile application guidelines prohibit server-initiated impression measurement
  • 2015: The IAB forms the Modernizing Measurement Task Force to update MRC digital measurement guidelines
  • August 2016: Publisher discrepancies over rendered versus served counts surface at industry forums
  • February 1, 2017: Public comment closes on the proposed count-on-begin-to-render criteria
  • September 2017: IAB Australia circulates a begin-to-render ad measurement model for public comment
  • October 31, 2017: The MRC, IAB Tech Lab and MMA release final desktop display, mobile web and mobile in-app guidelines establishing count-on-begin-to-render
  • November 30, 2017: Public comment closes on the updated digital video impression guidelines
  • Late 2018: The one-year grace period ends and compliance becomes an MRC accreditation criterion
  • January 2024: The IAB and MRC Retail Media Measurement Guidelines require begin-to-render measurement of digital creatives
  • February 2024: Draft augmented reality advertising measurement guidelines apply the same minimum within AR sessions
  • May 2024: Amazon DSP receives MRC accreditation for display begin-to-render impressions and click-throughs
  • July 14, 2025: Google implements invalid traffic changes roughly doubling reported invalid begin-to-render impression counts in Display and Video 360
  • September 2025: Display and Video 360 begins shifting display impression counting from on-download to begin-to-render
  • November 6, 2025: Campaign Manager 360 follows with the same shift
  • November 13, 2025: Integral Ad Science earns MRC accreditation for third-party measurement on Amazon DSP
  • April 2026: HUMAN Security's viewability measurement earns MRC accreditation
  • August 21, 2026: Google documentation schedules the move of billed display impression counting to begin-to-render for February 2027

Summary

Who. The Media Rating Council wrote the rule with the IAB Technology Laboratory and the Mobile Marketing Association. Ad servers and measurement vendors implement it, accredited platforms including Amazon DSP, Campaign Manager 360 and Display and Video 360 report against it, and verification firms are audited on it.

What. Begin-to-render is the minimum qualification for a valid ad impression: the creative must be loaded on the client and must at least start painting or be inserted into the Document Object Model. Visibility on screen is not required.

When. Video reached the equivalent standard in 2006 and 2009. Display, mobile web and mobile in-app adopted it on October 31, 2017, with compliance becoming an accreditation criterion a year later. Google's billed display impressions are scheduled to move in February 2027.

Where. In the browser or application, at the point the creative begins drawing, measured by a client-initiated beacon rather than by the ad server's response.

Why. Counting on server response credited advertisements that never arrived. Begin-to-render moves the trigger to the last practical moment before visibility and gives viewable impression measurement a defensible base to divide into.