Skip to main content

Reporting Agent - Performance Insights

Discover AI-powered performance insights for your next monthly meeting

P
Written by Product Team
Updated today

Performance Insights help you quickly understand what changed in your dashboards - without manually reviewing every chart.

Instead of scanning each report one by one, you get a concise summary of meaningful shifts, trends, and anomalies based on the data currently displayed.

Their purpose is to:

  • Highlight important performance changes

  • Connect signals across multiple reports

  • Save time in weekly or monthly reporting

  • Help you focus on what needs attention

If performance is stable and no strong signals are detected, you may see fewer insights. That’s expected, as not every period requires commentary.


Analyzing your app performance dashboards

Performance Insights always analyze the data currently visible in your dashboard.

This means they take into account:

  • Your selected date range

  • Applied filters

  • Selected apps, countries, channels, or dimensions

  • The specific dashboard you’re viewing

If you change your filters or adjust the reporting period, the insights will adapt to the new context. You can click Regenerate to refresh the insights based on your updated view.


How Performance Insights are generated

Performance Insights do not analyze raw time-series data directly. Instead, Reporting Agent first transforms report data into structured performance facts, such as:

  • Direction and magnitude of change

  • Comparison vs. the previous period

  • Stability vs. volatility

  • Unusual spikes or drops

  • Distribution or ranking patterns in aggregate charts

This approach allows the Agent to make relationships between reports and to treat large amounts of data over extended periods.

Analyzing incrementality experiments

Want to feel more confident measuring the incremental impact of ASO & Apple Ads? In the Incrementality section, Performance Insights work differently from dashboard-level insights:

  • The input is the experiment results and their statistical interpretation.

  • The model analyzes the results of your experiment (e.g., lift, confidence, statistical outputs).

  • As an output, you'll get an executive summary of how your Incrementality experiment performed, with an interpretation of both models (Extrapolation & Interpolation).


FAQ

How long are Performance Insights kept in memory?

Insights are stored as long as you make no modifications to your dashboard. Based on the date range applied to your dashboard (fixed or relative dates), the insights are stored differently:

  • For dashboards with relative date ranges ("last X days"), the insights are stored until the end of the current day (UTC midnight). Since the majority of the data sources are refreshed once a day, it's not necessary to have more frequent updates.

  • For dashboards with fixed date ranges, insights are kept for 30 days.

Can Performance Insights be shared in view mode?

No, insights are currently only available to logged-in AppTweak users. If you want to share insights, you just need to copy and paste the text.

Did this answer your question?