Meta restricts attribution windows and data retention in Ads Insights API

Meta dropped a bombshell on October 13, 2025, that will fundamentally reshape how advertisers measure campaign performance through its Ads Insights API. The changes, set to take effect on January 12, 2026, eliminate two view-through attribution windows and impose strict historical data retention limits that will force developers and marketing agencies to completely rethink their measurement infrastructure.

Chris Cutlip published the announcement on Meta's Developer Blog on October 16, 2025, detailing deprecations that affect eight critical API endpoints used by thousands of marketing technology platforms worldwide. The modifications apply to all API versions simultaneously, leaving no escape route for developers hoping to maintain legacy functionality.

The attribution windows getting axed

Starting January 12, 2026, Meta will pull the plug on two view-through attribution windows that have been fundamental to how advertisers justify upper-funnel spending. The 7-day view-through window (action_attribution_windows=7d_view) and 28-day view-through window (action_attribution_windows=28d_view) will simply stop returning data.

View-through attribution has long been controversial in digital advertising circles. The methodology credits conversions to advertisements that users saw but never clicked, operating on the assumption that mere exposure influenced purchase decisions. Marketing professionals have increasingly questioned whether this approach overestimates advertising impact by attributing conversions that would have happened anyway.

The timing is particularly interesting given ongoing debates about how Meta's attribution methodologies work. Recent discoveries revealed that Meta counts likes, shares, and saves as "clicks" within attribution windows, meaning conversions from users who never left Meta's platforms get credited to advertising campaigns based on engagement actions alone. Now the platform is narrowing which attribution windows remain available altogether.

What survives? The 1-day click, 7-day click, 28-day click, 1-day engaged view, and 1-day view windows remain intact. These options will continue working alongside default values based on individual advertisers' attribution settings for API fields including actions, conversions, and results. But the loss of longer view-through windows removes measurement options that many awareness campaigns relied upon to demonstrate value.

Historical data gets memory-holed

Meta is implementing three distinct historical data retention limitations that create an expiration date for performance analysis. Each restriction targets specific breakdown types, fundamentally changing how far back advertisers can query campaign data.

The first limitation affects all breakdowns for unique-count fields, restricting them to 13 months of historical data. Fields like unique_actions and cost_per_unique_action_type will return absolutely nothing when advertisers query with start_dates beyond 13 months. This eliminates the ability to conduct year-over-year performance comparisons for unique user metrics beyond a single year.

Hourly breakdowns face the identical 13-month guillotine. The breakdowns parameter value hourly_stats_aggregated_by_advertiser_time_zone will only return data within that window. Advertisers who analyze campaign performance at granular time intervals for dayparting optimization will lose access to historical patterns that extend beyond 13 months. Understanding how campaigns performed at specific hours during previous years becomes impossible.

The most restrictive limitation hits frequency breakdowns at just six months. Queries using breakdowns=frequency_value with start_dates older than six months will return empty data sets. Frequency data helps advertisers understand how many times users saw advertisements before converting, a metric commonly used in reach and frequency campaign analysis. The six-month limit eliminates any ability to analyze long-term frequency effects on conversion behavior.

Meta specifies that total values for API fields remain unaffected by these changes. Advertisers can continue accessing up to 37 months of historical data for aggregate metrics, matching the data retention available in Meta Ads Manager. But the breakdown restrictions fragment the measurement landscape, creating a situation where some data persists while granular analysis disappears.

Marketing mix modeling forced into slow lane

Marketing mix modeling (MMM) breakdowns will be restricted to asynchronous jobs exclusively. The breakdowns=mmm parameter will no longer function in synchronous API requests at all. Advertisers using MMM breakdowns must implement entirely new asynchronous job workflows to retrieve this data.

This change directly affects advertisers who analyze incremental advertising impact through econometric modeling. MMM helps businesses understand how different marketing channels contribute to overall sales by examining correlations between advertising exposure and business outcomes across extended timeframes. The restriction to asynchronous jobs means these queries will take substantially longer to complete, though they can theoretically handle larger data sets than synchronous requests allow.

The technical implementation requires a multi-step process: submit a job request, poll for completion status, and retrieve results once processing finishes. This workflow differs fundamentally from synchronous requests that return data immediately, forcing developers to rewrite code that currently relies on real-time MMM data access.

Eight endpoints in the crosshairs

The restrictions apply to both GET and POST methods across ad account, campaign, ad set, and ad levels. GET requests to /{ad-account-id}, /{campaign-id}, /{ad-set-id}, and /{ad-id} will face the new limitations. POST requests to the same four endpoint structures will also be restricted.

These endpoints represent the primary methods developers and third-party tools use to retrieve advertising performance data from Meta's platform. Marketing technology platforms integrate with these endpoints to pull data into dashboards, analytics tools, and reporting systems. The modifications will ripple through the entire ecosystem of tools built on top of Meta's API infrastructure.

The 90-day period between the October 13 announcement and the January 12, 2026 implementation provides developers time to modify their systems. But for organizations with complex measurement infrastructures that span multiple departments and external vendors, three months may prove insufficient for comprehensive migration.

Messenger Inbox placement getting deprecated too

On October 9, 2025, Meta announced that the Messenger Inbox placement (messenger_home) will no longer be available for targeting in ads creation. This change applies to API version 23.0 and higher immediately, with all versions affected from November 11, 2025.

Three endpoints face impact: GET /{ad-account-id}/adsets, POST /{ad-account-id}/adsets, and GET /{ad-set-id}?fields=targeting. Advertisers currently using Messenger Inbox placement in active campaigns can continue running those campaigns but cannot create new campaigns or ad sets targeting this placement.

The Messenger Inbox deprecation follows an accelerated timeline compared to the attribution and data retention changes. Advertisers have less than two months to adjust targeting strategies before the November 11 deadline removes the placement option entirely.

Why this matters for marketing measurement

These modifications align with Meta's broader effort to standardize measurement methodologies across its advertising platform. Back in August 2024, the company deprecated over 100 unique metrics from the Ads Insights API, removing numerous action_type breakdowns from the unique_actions and cost_per_unique_action_type fields.

The pattern is clear: Meta is systematically narrowing the measurement options available to advertisers while pushing them toward standardized approaches that align with what's available in Meta Ads Manager. This creates consistency between manual reporting and programmatic data access, but it eliminates flexibility for advertisers who preferred different measurement methodologies.

Meta has consistently emphasized Attribution Windows and first conversion attribution as preferred measurement approaches. The company encourages advertisers to explore these methodologies through the action_attribution_window parameter, positioning them as superior alternatives that provide insights into how advertisements influence customer behavior over time.

But this framing obscures a more uncomfortable reality: these changes remove measurement options that many advertisers relied upon, particularly for awareness campaigns where direct attribution becomes murkier. The elimination of 7-day and 28-day view-through attribution windows forces advertisers to justify upper-funnel spending through different mechanisms entirely.

The technical nightmare for developers

Developers integrating with Meta's Marketing API face a compressed timeline to update their implementations before January 12, 2026. Applications requesting deprecated attribution windows will receive empty data sets rather than error messages, creating silent failures in reporting systems that developers may not immediately detect.

The historical data retention limits require developers to implement date range validation logic. Systems that automatically query data beyond the retention limits will encounter null responses, breaking historical analysis and year-over-year comparisons that span beyond the restricted timeframes. This affects not just new queries but any automated reporting that pulls historical data programmatically.

Third-party analytics platforms aggregating data from multiple advertising sources face additional complexity. These systems must account for different data retention policies across platforms while maintaining consistent reporting interfaces for their customers. When Meta data expires at 13 months while other platforms retain data longer, creating unified reports becomes significantly more challenging.

Asynchronous job implementation for MMM breakdowns requires substantial code modifications for advertisers currently using synchronous requests. The multi-step workflow introduces latency into analysis processes that previously operated in real-time. Developers must implement polling mechanisms, handle job queue timeouts, and manage result retrieval through entirely different code paths.

What advertisers actually lose

The attribution window deprecations will fundamentally alter how advertisers measure campaign performance, particularly for awareness-focused campaigns where view-through conversions played a significant role in demonstrating value. Advertisers who relied on 7-day and 28-day view-through data to justify upper-funnel advertising spend must adjust their measurement frameworks or face internal pressure to reduce awareness campaign budgets.

The ongoing debates about attribution methodologies take on new significance in this context. Marketing professionals have raised concerns that inflated Return on Ad Spend metrics mislead advertisers about campaign performance. Digital marketing specialist Bram Van der Hallen highlighted systematic issues with how ROAS is calculated and reported across Meta's advertising platform, arguing that "one marketer showing you a Meta Ads ROAS of 2 is probably doing a better job than that other marketer showing you a ROAS of 10."

View-through attribution represents one of the primary mechanisms that artificially inflate performance metrics. The measurement approach credits conversions to ads that users viewed but didn't click, often overestimating the actual impact of advertising campaigns. Meta's decision to eliminate longer view-through windows may actually improve measurement accuracy by forcing advertisers to focus on more direct attribution methodologies.

But that improvement comes at a cost for legitimate awareness campaigns where view-through effects are real but harder to measure. Brand advertising often works through exposure and recall rather than immediate clicks. The elimination of longer view-through windows makes quantifying these effects substantially more difficult.

The historical data limitations particularly affect agencies and consultants who analyze long-term campaign trends for clients. Year-over-year performance comparisons extending beyond 13 months become impossible for unique-count fields and hourly breakdowns. Seasonal analysis that looks back multiple years to identify patterns gets truncated.

Frequency analysis restrictions at six months eliminate seasonal comparison capabilities entirely. Understanding how frequency affects conversion behavior requires examining data across multiple quarters and seasonal cycles. The six-month limit chops that analysis period in half, removing the ability to compare summer campaign frequency patterns with previous summer data or holiday season comparisons year-over-year.

Marketing mix modeling practitioners face workflow changes with the asynchronous requirement. While the restriction theoretically enables processing larger data sets, it introduces latency into analysis processes. Advertisers accustomed to real-time data access must adjust operational procedures to accommodate job submission and polling rather than immediate results.

The privacy and regulatory backdrop

These changes occur against a backdrop of increasing privacy regulation and data protection requirements. The European Union's Transparency & Targeting of Political Ads regulation prompted Meta to prohibit social issues, electoral, and political ads in the European Union starting October 6, 2025, demonstrating how regulatory pressure shapes platform capabilities.

Apple's privacy frameworks have significantly influenced how platforms structure attribution methodologies. Meta's implementation of SKAdNetwork 4.0 for iOS advertising demonstrates the industry's ongoing adaptation to privacy-focused measurement systems that provide enhanced attribution windows and coarse conversion values while maintaining user privacy protections.

The reduction in available attribution windows may reflect Meta's response to these external pressures beyond just technical considerations. Longer attribution windows have faced criticism for potentially overattributing conversions to advertising, particularly when multiple touchpoints influence customer decisions over extended periods. Shorter windows reduce attribution noise and provide clearer pictures of recent interactions that directly influenced conversions.

But Meta's public rationale focuses on performance improvements and consistency with Meta Ads Manager rather than privacy compliance. The company states these changes will "reduce discrepancies with Meta Ads Manager" and "improve overall API performance." This framing positions the restrictions as technical optimizations rather than privacy-driven requirements, though the distinction may be largely semantic.

Alternative measurement approaches emerging

Meta continues developing alternative measurement capabilities as it restricts traditional metrics. The company introduced the Instagram follows metric in July 2025, enabling advertisers to track follower acquisition from advertising campaigns across all campaign types at campaign, ad set, and individual ad levels.

Creative breakdown functionality launched in July 2025 allows advertisers to analyze performance by specific creative elements, including AI-generated image variations. This transparency provides insights into how Meta's artificial intelligence tools affect advertising outcomes, addressing a gap in campaign optimization capabilities.

Incremental attribution, which Meta has been expanding globally, optimizes for conversions directly caused by advertisements rather than all conversions within attribution windows. Testing data showed advertisers using incremental attribution achieved a 46 percent increase in incremental conversions compared to standard campaigns. This methodology addresses criticism that traditional attribution overcredits advertising by including conversions that would have occurred regardless of ad exposure.

These alternative approaches suggest Meta is attempting to replace the measurement capabilities it's removing rather than simply eliminating them. But the transition creates gaps during the period when old methodologies disappear before new ones fully mature. Advertisers operating during this transition face uncertainty about which measurement approaches will ultimately prove most reliable.

How this compares across platforms

These restrictions place Meta's data retention policies roughly in line with industry standards, though specific implementations vary across advertising platforms. Google AdMob recently cut historical data retention periods, restricting Ads Activity report data to seven years and User Activity report data to just 90 days.

The historical data limitations for unique-count metrics and hourly breakdowns at 13 months, and frequency breakdowns at six months, establish clear boundaries for retrospective analysis. These limitations may prompt advertisers to implement their own data warehousing solutions for long-term trend analysis rather than relying on platform-provided historical data access.

The deprecation of specific attribution windows contrasts with Meta's previous approach of offering multiple measurement options. This shift toward a more constrained set of attribution methodologies reflects the platform's emphasis on standardization and consistency with its primary reporting interface. But it also removes granularity that sophisticated advertisers valued for understanding different aspects of campaign performance.

What advertisers should do now

Developers should immediately audit their current API implementations to identify any usage of the deprecated 7-day and 28-day view-through attribution windows. Code that explicitly requests these windows must be modified to use supported alternatives or removed entirely before January 12, 2026.

Organizations relying on historical data queries beyond the retention limits should export and archive relevant data before the deadline. Once the restrictions take effect, this historical information will no longer be accessible through the API. Companies conducting year-over-year analyses or long-term trend studies need to pull that data now while it still exists.

Testing implementations with asynchronous job workflows for MMM breakdowns should begin well before the deadline. The transition from synchronous to asynchronous requests requires different error handling and timeout management strategies that demand testing under realistic conditions before production deployment.

Third-party tool vendors must communicate with their customers about how these changes affect available features and reporting capabilities. Some functionality that depends on deprecated metrics or extended historical data may need to be discontinued or redesigned entirely. Vendors waiting until January to address these changes risk losing customer trust when features suddenly stop working.

Marketing teams should evaluate their current measurement frameworks to understand which metrics and analyses will be affected. Campaigns that currently justify performance through 7-day or 28-day view-through attribution need alternative measurement approaches developed before those windows disappear. Historical analyses that extend beyond 13 months for unique metrics or six months for frequency analysis need to be completed or archived before the data vanishes.

The message is clear: Meta is narrowing the measurement landscape, and advertisers need to adapt quickly or risk losing critical performance data forever.

Timeline

• February 3, 2024: Meta launches Engaged-View attribution for video campaigns measuring conversions within 24 hours of 10-second video views
• May 22, 2024: Meta releases Graph API v20.0 and Marketing API v20.0 with automated version upgrades
• August 11, 2024: Meta announces deprecation of over 100 unique metrics from Ads Insights API
• October 2, 2024: Meta releases Graph API v21.0 and Marketing API v21.0 with multiple endpoint modifications
• January 21, 2025: Meta simplifies Instagram and Marketing API integrations replacing legacy objects
• May 29, 2025: Meta launches unified API structure for Advantage+ campaigns eliminating separate workflows
• June 22, 2025: Meta activates SKAdNetwork 4.0 requiring manual configuration for extended attribution windows
• July 11, 2025: Meta unveils Creative breakdown for Flexible formats and AI-generated image ads
• July 16, 2025: Meta advertising experts warn inflated ROAS misleads marketers highlighting attribution methodology flaws
• August 17, 2025: Meta deprecates additional Page Insights API metrics from November 15
• August 26, 2025: Meta introduces Instagram follows metric for ads reporting
• October 2, 2025: Advertisers discover Meta counts likes and shares as clicks for conversions within attribution windows
• October 9, 2025: Meta announces Messenger Inbox placement deprecation effective November 11, 2025
• October 13, 2025: Meta announces Ads Insights API attribution window and historical data restrictions
• October 16, 2025: Chris Cutlip publishes detailed announcement on Meta Developer Blog
• November 11, 2025: Messenger Inbox placement deprecation applies to all API versions
• January 12, 2026: Attribution window deprecations and historical data retention limits take effect

Summary

Who: Meta announced changes affecting developers, marketing agencies, and advertisers using the Ads Insights API through Marketing API versions. Chris Cutlip, a representative from Meta's Developer team, published the announcement detailing the technical specifications and implementation timeline.

What: Meta will deprecate the 7-day view-through (7d_view) and 28-day view-through (28d_view) attribution windows, eliminating two of the longer view-through measurement options advertisers currently use. The company will limit historical data retention to 13 months for all breakdowns of unique-count fields (like unique_actions and cost_per_unique_action_type) and hourly breakdowns across all fields. Frequency breakdowns face even stricter limits at six months of historical data retention. Marketing mix modeling (MMM) breakdowns will be restricted to asynchronous jobs only, eliminating real-time synchronous access. The Messenger Inbox placement (messenger_home) will also be deprecated for ads targeting across all versions.

When: Meta announced the changes on October 13, 2025, with Chris Cutlip publishing the detailed documentation on October 16, 2025. The attribution window deprecations and historical data retention limits take effect on January 12, 2026, applying to all API versions simultaneously. The Messenger Inbox placement deprecation applies to API version 23.0 and higher immediately, with all versions affected from November 11, 2025. This provides approximately 90 days notice for the main changes and less than two months for the Messenger placement removal.

Where: The changes affect Meta's Ads Insights API globally, impacting eight endpoints across ad account, campaign, ad set, and ad levels (both GET and POST methods for /{ad-account-id}, /{campaign-id}, /{ad-set-id}, and /{ad-id}). The modifications apply to advertising campaigns running on Facebook, Instagram, Messenger, and WhatsApp platforms. Third-party marketing technology platforms, analytics tools, and custom dashboards that integrate with these endpoints face worldwide impact.

Why: Meta states these changes will "reduce discrepancies with Meta Ads Manager" by ensuring use_unified_attribution_setting and action_report_time parameters are disregarded, making API responses mimic Meta Ads Manager settings exactly. The company also claims the modifications will "improve overall API performance" by restricting reach data for queries applying breakdowns with start_dates more than 13 months old, encouraging developers to leverage asynchronous jobs for such data requests instead. The changes align measurement capabilities with privacy requirements and industry standards while standardizing attribution methodologies across Meta's advertising platform, though this standardization eliminates measurement flexibility that many advertisers previously relied upon for awareness campaigns and long-term trend analysis.