Amazon Ads Clarifies Unified Reporting Constraints Ahead of Legacy Tool Sunset

As the January 31, 2027, deadline for the complete transition to Amazon’s unified reporting system looms, Amazon Ads has released a critical update to its "Data availability in unified reporting" documentation. Updated on November 17, 2026, the guidance provides the most granular breakdown to date regarding historical data retention and report-pull capacities.

For the thousands of agencies and advertisers currently scrambling to migrate legacy reporting workflows, this documentation is more than a technical reference—it is a roadmap for avoiding data integrity errors. By delineating the differences between "historical data held" and "maximum report pull," Amazon has effectively signaled that historical reporting in the new environment requires a fundamental change in strategy for long-term data analysis.


Main Facts: Navigating the Two Ceilings

The core of the documentation is a shift in how Amazon categorizes data access. The platform now applies two distinct limits to every time granularity: the Historical Data Limit (the total amount of data stored) and the Maximum Report Pull (the amount of data a single API call or UI report request can retrieve).

The Granularity Gap

The discrepancy between these two ceilings is most significant for daily reporting. While the unified system retains 24 months of daily historical data, a single report request is capped at 120 days. This means that an advertiser seeking a full two-year lookback must execute at least seven separate report pulls and reconcile them manually.

Weekly reports fare slightly better, offering a 15-month pull capacity within the same 24-month historical window. Monthly reports reach back the furthest—six years (72 months)—but can only retrieve 25 months of data in one instance. Only by running annual or custom date-range summaries can an advertiser pull the full 72 months of history in a single request.

The Marketing Stream Exception

Amazon Marketing Stream, the high-velocity API feed, operates under a different set of rules. It currently holds 95 days of hourly data, a significant improvement over the 14-day limit found in the standard Reporting app and Reporting API. However, the documentation notably leaves the "Max report pull" for Marketing Stream blank, likely due to the push-based nature of the service, which delivers data in real-time rather than upon request.


Chronology: A Moving Target for Advertisers

The evolution of these reporting limits has been a source of frustration for power users. When the unified reporting system first debuted in beta at the unBoxed conference on November 11, 2025, the industry was given broad, singular figures: two weeks for hourly, 15 months for daily/weekly, and up to six years for monthly/yearly data.

When the system reached general availability on June 8, 2026, these figures were reiterated. However, the November 17, 2026, update has effectively decoupled these figures. The "15-month" figure, once a catch-all for daily and weekly data, now applies exclusively to weekly pulls. The new 120-day limit for daily reports represents a significant narrowing of the "single-click" report window that many advertisers had grown accustomed to in legacy tools.

With the legacy report pages scheduled to become read-only on December 17, 2026, and to shut down permanently on January 31, 2027, the window for rebuilding automated schedules is rapidly closing. Advertisers who fail to account for these pull limits will find their new, migrated schedules returning incomplete data sets or "null" values.


Supporting Data: Field-Specific Availability

Perhaps the most technical aspect of the update is the disclosure of "late-start" dimensions and metrics. Even if a report covers a date range within the 72-month limit, it will return empty values if the requested metric did not exist at that time.

Dimension and Metric Availability Dates

  • Reach: July 1, 2024
  • Total/ROAS (Reconciled): August 4, 2024
  • Audience Segment: February 9, 2025
  • Advertised/Converted Product: February 13, 2025
  • Benchmark Group: June 1, 2025
  • Off-Amazon Conversions: June 27, 2025
  • Frequency Group: January 16, 2026
  • Ad Variation Fields: October 1, 2026

The complexity here is compounded by overlapping features. For instance, reach metrics reported at the "Frequency group" level are limited to the Frequency group availability date (January 16, 2026), rather than the general Reach availability date (July 1, 2024). This creates a fragmented data landscape where "Reach" may appear as a number for 2024, but as a "null" for the same period if the "Frequency group" dimension is toggled on.


Official Responses and Documentation Nuances

Amazon’s documentation offers specific instructions on how the system handles these "dead zones" in data. Crucially, the system does not issue warnings or failure notifications when a date range exceeds a metric’s availability. Instead, it returns partial totals or null values.

The most controversial element of the update concerns Sponsored Products video metrics. Amazon explicitly stated: "Only 5-second views and 5-second view rate report data for Sponsored Products video. Other video metrics may display 0 as a placeholder, which isn’t the actual measurement."

This "placeholder zero" is a significant departure from standard reporting practices. In data analytics, a null typically signifies a missing field, whereas a zero is a quantitative measurement of nothingness. By using zero as a placeholder, Amazon risks corrupting aggregate metrics. If an advertiser sums video metrics across various ad formats, the placeholder zeros from Sponsored Products will be added to real figures from other sources, potentially leading to inaccurate performance calculations.


Implications: The High Cost of Migration

The implications for agencies are profound. The migration tool provided by Amazon only moves active saved or scheduled reports from the last 24 months. Because these legacy reports often bypass the new "pull" limitations, they cannot be simply "lifted and shifted."

1. The Reconciliation Crisis

For agencies that consolidate performance data across multiple accounts and ad products, the new reporting reality is a minefield. With four distinct causes for a null value—non-applicable ad products, pre-availability dates, restricted combinations (like DVA+ halo metrics), and permission settings—it is now impossible to discern why data is missing without cross-referencing the documentation for every single field.

2. The DVA+ and Halo Challenge

The documentation specifically highlights that for DVA+, halo conversions are unavailable at the audience segment level. Base metrics, which include halo sales, return nulls when combined with audience segments. This forces advertisers to abandon their traditional "base metrics" approach and rely on "promoted metrics" to evaluate DVA+ performance, effectively breaking the continuity of year-over-year reporting.

3. Reporting Strategy Re-Engineering

Advertisers must move away from the "one-size-fits-all" report. Because of the 120-day daily pull limit and the 25-month monthly pull limit, data engineering teams must now build robust, modular pipelines. Instead of pulling a single file covering 24 months, pipelines must be restructured to perform iterative pulls that respect the specific granularity thresholds, then stitch the results together in a data warehouse.

4. The "Placeholder Zero" Risk

The decision to utilize placeholder zeros for Sponsored Products video metrics is the most significant red flag for data scientists. Until Amazon updates the reporting infrastructure to either return nulls or actual measurements, any dashboard that aggregates video performance is at risk of reporting artificially deflated averages.

Final Analysis

The November 17 update provides the necessary clarity that the industry has been requesting since the inception of the unified reporting project. However, the reality it describes is one of increased complexity. The days of simple, long-range report generation are over.

As January 31, 2027, approaches, the burden of data integrity shifts entirely to the advertiser. By publishing these constraints, Amazon has provided the necessary rules of the game, but the responsibility for cleaning, stitching, and validating these fragmented data streams remains squarely with the end-user. Agencies that fail to re-engineer their reporting infrastructure to accommodate these new ceilings risk losing visibility into their historical performance just as the legacy tools vanish.