This document describes Pinterest’s general methodologies around ad measurement, filtration, and reporting for Pinterest ad impressions and clicks across desktop, mobile app (iOS and Android), and mobile web for both display and video ads. 

Unless otherwise stated, the methodologies in this article apply to all ad formats (static, carousel, etc.). Further, Pinterest employs General Invalid Traffic Detection (GIVT) as defined by the Media Rating Council’s (MRC’s) Invalid Traffic Detection and Filtration Guidelines to help identify, and subsequently filter, non-human traffic from reportable and billable impressions and click metrics. 

Pinterest is currently MRC accredited for display Pin impressions and Pin clicks. Metrics other than these have not been accredited by the MRC.

Overview

Pinterest is a visual discovery engine for finding ideas like recipes, home and style inspiration, and more. Our mission is to bring everyone the inspiration to create a life they love. For advertisers, Pinterest provides an advertising network to deliver display and video ads to registered, logged-in Pinterest account users on Pinterest owned and operated (O&O) properties (desktop, mobile app and mobile web).

Pinterest serves ads alongside non-promoted, organic content (referred to as Pins) on O&O properties and does not serve ads to external publisher websites. In this capacity, Pinterest operates as an ad server, decision engine and publisher (Pinterest does not allow third-party ad serving to Pinterest properties).

 
Pinterest Ads Manager is the primary interface for advertisers to create, edit and manage ad campaigns on Pinterest. Impressions and clicks reported through Ads Manager reflect the billable impressions and clicks delivered and measured through Pinterest’s ad delivery system.

Pinterest counts an impression if at least one pixel of the Pin container appears on screen for at least one continuous second. This measurement is applicable to both display and video Pins. Pinterest’s method of counting video impressions differs from MRC/IAB standards, which means that video playback is not required for the impression to be counted. 

Video impressions on feed that do not meet MRC/IAB playback requirements are estimated to be ~9%. This is primarily a result of Pinterest's unique two-column grid, where auto-play is intentionally limited for optimal Pinner experience. Video View measurement follows MRC/IAB standards of playback for at least 2 seconds with 50% of video in view.

When one pixel of a Pin container becomes visible in the user’s device we record an impression and set a start time. Once one pixel of the Pin container is no longer visible, the end time is set for the impression. The start and end times are used to determine if the impression has been on screen for at least one continuous second, and is thus counted. Polling for changes in the user state is performed for at least 60 frames per second and at a 16 millisecond interval.    

All impressions use an insertion staleness window. If an impression occurred outside the window, the impression is considered invalid and will be dropped from reportable metrics and billing.

To minimize undercounting due to caching, Pinterest uses HTTP header controls and unique identifiers (random integer strings) assigned to each ad request.

Pinterest video creatives are by default configured to auto-play and will begin playback once 50% of the video player’s pixels appear within the viewable portion of the user’s device; videos automatically pause once scrolled vertically out of view based on the same 50% threshold. 

When auto-play is enabled, max. width video ads re-play until the user scrolls vertically out of view or exits the screen. Auto-play and click-to-play videos are currently not separated in reporting. Auto-play can only be disabled on mobile and web.

Pinterest uses data collection processes initiated by the user’s platform (web browser, mobile apps, etc.) via communication sent to server side endpoints. These are logged as raw statistics and further processing (general editing rules, invalid traffic filtration, etc.) is applied to these logs to determine billable vs non-billable events. This aggregated data is stored within Pinterest’s data warehouse and is available for customers to view campaign performance metrics on Ads Manager and through select API endpoints. Data collection is performed continually while the device is connected and batched if the mobile device is offline.

  • Obstructions: If an obstruction is detected within the user’s viewport, we end any active impression measurement (including video content that may be playing at the time of the obstruction). 
  • Abandonment: If users abandon their screens, no additional impressions are measured as only one impression is counted once the impression meets Pinterest’s logging requirements.
  • Ad blocker (Desktop): Pinterest does not count ads that have been blocked by ad blocker technologies. After serving content (inclusive of ads), ad blockers could potentially block Pins which will prevent on-screen measurement for these Pins. Thus, these blocked ads will not be counted as these do not meet the one pixel for at least one continuous second requirement for impression measurement.

Pinterest uses the following for first-party measurement and reporting of clicks:

  • An “in-unit click” measurement methodology referred to as a “Pin click.” For example, a user-initiated click on an ad that leads to content on or off of Pinterest.
  • A “click” or “click-through” measurement methodology referred to as an “outbound click.” For example, a user-initiated click on an ad that causes a redirect to another web location.

To collect and record Pin clicks and outbounds clicks, Pinterest uses a “one-click-per-impression” methodology (for example, a single impression may have only one associated click) as opposed to a multiple-click-per-impression methodology. Pinterest only charges advertisers for Pin clicks and not for outbound clicks.

In desktop and mobile web environments, the user’s initial click or tap on the display ad asset triggers a new tab to open and the user is redirected to the advertiser’s landing page. In this instance, Pinterest will record a Pin click since a user has clicked or tapped on the display ad asset. An outbound click will also be recorded since the click or tap caused a redirect to another web location.

In mobile application environments, the user’s initial tap on the ad placement triggers the ad creative to expand. In this instance, Pinterest will record a Pin click since the user has tapped on the ad placement. After the ad is in the expanded state, metadata (if provided by the advertiser) will appear below the expanded ad. A user can swipe or click on the metadata to be redirected to the advertiser’s landing page. In this instance, Pinterest will record an outbound click since the tap caused a redirect to another web location.

Note: For ads linking to a mobile deep link or an app store link, the user’s initial tap on a display ad placement triggers a redirect to the advertiser’s mobile deep link. Once this redirect has resolved, Pinterest will record both a Pin click and an outbound click.

All clicks use an insertion staleness window. If a click occurred outside the window, the click is considered invalid and will be dropped from reportable metrics and billing.

Pinterest uses multiple techniques based on identifiers, activity, and patterns from data in the log files in an attempt to identify and filter (exclude) invalid activity. This activity includes, but is not limited to known and suspected non-human activity and suspected invalid human activity. Pinterest also uses data from the Interactive Advertising Bureau (IAB) and Trustworthy Accountability Group (TAG) to identify traffic generated from known sources of non-human activity.  

Further, Pinterest employs controls at the account level. These account level validation techniques are performed at initial account creation and on an ongoing basis as user’s (both business and non-business) interact with Pinterest’s platform. These techniques are centered around flagging suspicious user accounts for potential silencing or deactivating, thus preventing these user accounts from generating traffic within the Pinterest environment (as Pinterest operates in a logged-in environment). The models used for account level validations are continuously monitored and evaluated for effectiveness in identifying user accounts that have violated Pinterest’s policies.

Pinterest uses robot.txt files to disallow all access to ads measurement servers.

Pinterest uses activity-based techniques to aid in the identification and downstream filtration of invalid traffic. If a user has been identified to have violated these techniques, subsequent events from this user will still be logged, but will be marked as non-billable in campaign reporting. The paid version of the reported billable metrics filters out the invalid traffic identified from these techniques. As such, gross level campaign reporting includes both billable and non-billable events while net level campaign reporting includes only billable events. Paid level metrics are synonymous with MRC’s definition of a net metric. Invalid activity can be derived by subtracting the net metric from the gross metric.

Pinterest's decision rate is at 100%, which means that all traffic across paid impressions, paid pin clicks, and paid video views are filtered for invalid traffic per MRC GIVT guidelines. The decision rate is computed as recorded activity where sufficient information and signals used to make an GIVT determination were collected; divided by the total number of activity intended for measurement and reporting.

Pinterest identifies and excludes internal traffic originating from Pinterest’s employee accounts, internal testing tools or accounts, and internal IP addresses. Traffic generated from these accounts is excluded from any measurement or reporting of campaign activity. 

Offline activity is batched and resent once the device is back online. If the offline activity occurred outside the insertion staleness window, the activity is considered invalid and will be dropped from reportable metrics and billing. Pinterest currently does not distinguish between online and offline impressions for the purposes of logging and reporting.

Pinterest controls the entire ad supply chain (inventory, pricing, serving, logging, etc.) on all Pinterest O&O platforms. Based on this implementation, Pinterest does not outsource any material processes as it relates to ad measurement. Further, Pinterest does not acquire or purchase traffic externally.

To create a personal account, users agree to Pinterest’s terms of service and privacy policy. When creating a business account, advertisers agree to Pinterest’s business terms of service and privacy policy. Pinterest may terminate a personal or business account if the user or advertiser is found in violation of Pinterest’s policies.

Advertisers seeking to advertise on Pinterest’s Ads Manager platform can either upgrade to a business account from an existing personal Pinterest account and agree to Pinterest’s ad terms, or create a new Pinterest business account and agree to Pinterest’s ad terms.

After advertising measurement events are processed, event data is aggregated for first-party reporting at both gross and net level within the Ads Manager UI. Ads Manager is updated hourly and final processing of data occurs on a rolling 24-hour basis. Further, Ads Manager reports metrics on the basis of coordinated universal time (UTC). For defined device classification reporting read our article, Set up device targeting

Pinterest performs data quality and integrity controls over advertising measurement data reported in the Ads Manager UI. Traffic (ads logging and invalid traffic) is also monitored across different environments supporting Pinterest’s platform to help identify any potential issues in both measurement and the accuracy of reported data. Pinterest also ensures that as part of any release of client facing code, a series of tests to validate logging accuracy of the in-scope metrics, must pass.

If Pinterest detects issues that impact advertiser data or spend, we will follow an investigation process into the cause and impact of such issues. Decisions to provide credit or refunds are made on a case-by-case basis and are dependent on the circumstances of the particular issue. In the event a credit or refund is provided, advertisers are notified through our customer operations team.

End of Other articles Links
Still need help? Contact us
User feedback
Was this article helpful?

collection_fields

How can we make this article better?