Generate a summary report with aggregate reporting
Summary reports are a powerful tool for examining your data from Attribution Reporting and Private Aggregation. This pathway will take you through the key concepts and tooling needed to enable you to build effective reports.
Go back
6 działania
1check_circle
Introduction to summary reports
Optional
Learn the key concepts to work with summary reports and design your data collection.
2check_circle
Contribution budget for summary reports
subject
Article
Optional
Learn about the role of the contribution budget for Attribution Reporting summary reports and how to allocate it to capture the data you need.
Learn how to use Noise Lab, a tool that helps grasp the effects of various noise parameters, and that enables quick exploration and assessment of various noise management strategies.
[null,null,[],[[["Scale up aggregatable values before aggregation and scale them down after receiving the summary report to maximize signal relative to noise within the contribution budget."],["Split your contribution budget strategically across different measurement goals based on their importance and expected maximum values."],["Coarser aggregation keys and longer aggregation time periods generally result in less relative noise compared to more granular keys and shorter time periods."],["Summing up summary values from multiple reports, such as for rollups, increases the accumulated noise."],["Higher epsilon values in the Attribution Reporting API result in lower noise but also reduce privacy protection."]]],["Summary reports receive added noise, a random data amount, during processing to protect user privacy and enable differentially private measurement. This noise, drawn from a Laplace distribution, is applied once per key and is the same regardless of the summary value. The impact of noise is less significant on higher summary values. To mitigate noise impact, configure API usage for confident data-driven decisions and manage noise to remain below a certain percentage of the collected metric.\n"]]
