What does the confidence score mean when setting up a data import?

After the AI analyzes your data source, it reports a confidence score for the overall analysis and for each detected meter. This tells you how certain the AI is about its field mappings and meter identification.

Overall confidence score (shown at the top of results)

90–100%: High confidence. The AI found clear, unambiguous field mappings. Your data source is well-structured and the analysis is reliable. You can proceed with confidence.
70–89%: Medium confidence. Most mappings are clear but some fields were ambiguous. Review the detected meters and field mappings before confirming. Pay attention to any warnings.
Below 70%: Low confidence. The AI found possible meter data but could not map all fields reliably. This often happens with non-standard or poorly documented APIs. Review everything carefully and consider providing a more detailed specification.

Per-meter confidence

Each detected meter also has its own confidence score. A meter with 95% confidence means the AI is very sure about its external ID, name, type, unit, and data value type. A meter with 60% might have an uncertain unit or type — you should verify it matches your actual meter.

Detected meters include

External ID: The meter's identifier in the source system
Suggested name: A human-readable name the AI inferred
Meter type: Main meter, sub-meter, production, virtual, or check meter
Energy carrier: Electricity, heating, water, gas, etc.
Unit: kWh, m³, kW, etc.
Data value type: Cumulative (always-increasing counter) or gauge (point-in-time value)

Optimization tips

The AI may also suggest improvements — for example, "Consider using incremental time filtering to reduce API response size" or "This API supports cursor-based pagination for better performance."

Where to review: Analytics → Data Import → click on your import → Review AI Analysis section.