Data prep report overview (preview)
[This article is prerelease documentation and is subject to change.]
The data prep report in Dynamics 365 Customer Insights - Data helps you understand the overall data quality, the readiness of your data to produce insights, and helps you improve your data to unlock more and better insights for whatever sales or marketing strategy you have in mind.
Important
- This is a preview feature.
- Preview features aren't meant for production use and may have restricted functionality. These features are available before an official release so that customers can get early access and provide feedback.
Prerequisites
The data prep report automatically runs if the following prerequisites are met:
- Ingestion is completed with customer and transaction data.
- Unification is completed.
- Activities and relationships are mapped.
- Enable Copilot features powered by Azure OpenAI setting turned On. Default is On.
- Environment is in a supported geography and uses a supported language.
- At least 100 customer records, preferably 1,000 records.
- At least one year of transaction history, preferably two to three years. Ideally two to three transactions per customer ID, preferably across multiple dates.
- Customer identifier: A unique identifier to match transactions to your customers.
- Less than 20% missing values in required fields.
Data prep report
After unification is completed, the system automatically generates a data prep report based on your ingested and unified data, and analyzes contextual information on your data. This information is updated anytime you run unification.
Access the Data prep report from the Home page, the Data sources page, or the Predictions page.
Tip
If you don’t see the data prep report, it probably hasn't been generated because you haven't met the prerequisites. Ensure that you have completed ingestion and unification, mapped activities and relationships, and an admin has the global consent setting turned On in the Settings page.
There are four primary sections within the data prep report.
AI-generated data quality summary: A concise summary generated by an Open AI model of the data quality grade, insights readiness, and issues and recommendations sections. The summary appears on the Home page banner and within the data prep report.
Overall data quality grade: The grade indicates overall health of your data. The grade is calculated as an aggregated percentage (value ranging from 0-100%) with a corresponding level (high, medium, or low data quality). It's derived from weighted average scores across a set of data quality rules within industry-standard data quality pillars. Pillars such as completeness, consistency, uniqueness, accuracy, timeliness, validity, and integrity. If you have a high grade and corresponding high level of data quality, the quality of your data is sufficient to generate most of the insights available in the product with high confidence in meaningful results.
Insights readiness: Insights readiness indicates whether or not you have met the requirements to generate a specific insight. It's determined by comparing the baseline data requirements of each insight with the issues present in your data. When any issue violates any data requirement for an insight, the insight is deemed not ready to use. If an insight is deemed ready to use, it's likely to generate meaningful results.
Data quality issues and recommendations: These issues and recommendations provide comprehensive guidance on the issues surfaced in your data, including severity, which insights are impacted, and what recommendations for remediation to act upon. Issues are derived from the rules within the same industry-standard data quality pillars as the data quality grade. Any violation of these rules results in an issue. The fewer issues present, especially critical severity issues, the more likely you're to have a high data quality grade and have all insights labeled as ready to use.
Tip
The default view provides the most critical issues present in your data. To see all issues, organized by severity, turn off Show critical issues. To alter the view to show issues organized by other options, select Group by and make a selection. Available selections include severity, data quality pillar, and impacted insights.
In most cases, the issues and recommendations surfaced in the data prep report must be addressed by performing fixes on the source data outside of Customer Insights - Data, using data clean up tools such as Power Query. The new and improved data must then be reingested, and unification must be completed again for the data quality to improve. The data prep report refresh is only triggered when unification is completed.
Contextual information on your data
In addition to the data prep report, you get contextual information related to insights, specifically prediction models. Use this information to understand which prediction models are best suited to your data before you go through the time and effort of configuration and running the model.
On the Predictions page under the Create tab, models labeled as Use this model are most suited to your data while models labeled as Not ready to use aren't. For any Not ready to use models, review the full data prep report and make the necessary fixes to your data per the guidance of the Issues and recommendations section.