SmartResearch introduces a few terms that are helpful to understand before you start using it. This section explains them in plain language.
Systems of Record
A system of record is the primary system where your transactional data is created and maintained.
SmartResearch connects to your systems of record to read approved data. It does not replace them or become a new system of record.
Metrics and KPIs
A metric is a quantitative measure, such as:
- Total invoice amount this month.
- Number of overdue invoices.
- Average days from invoice creation to approval.
A key performance indicator (KPI) is a metric your organization has identified as particularly important to monitor.
In SmartResearch:
- Metrics and KPIs are defined and managed in Metrics Studio.
- Each metric has a clear definition, formula, and data source.
- When you ask a question, SmartResearch uses these definitions to calculate results.
This ensures that different users and teams see results based on the same logic.
Smart Skills
Smart Skills are configurations that help SmartResearch interpret your questions and map them to the right data and metrics.
You can think of a Smart Skill as a reusable capability that knows:
- What kind of business question it applies to (for example, invoice status, vendor spend, exceptions).
- Which data and metrics should be used to answer that question.
- Any rules or filters that should always apply (for example, excluding certain invoice types).
Metrics Studio
Metrics Studio is the part of the product where metrics and KPIs are defined, reviewed, and maintained.
In Metrics Studio, authorized users can:
- See the list of metrics available to SmartResearch.
- Review metric descriptions, formulas, and data sources.
- Confirm that metric definitions match how your organization wants to calculate them.
Confidence Score
When SmartResearch answers a question, it may show a confidence score. This is an indication of how confident the system is that:
- It understood what you asked, and
- It used the most appropriate data and metrics.
In general:
- A higher confidence score suggests the answer is more likely to reflect your intent and the configured definitions.
- A lower confidence score suggests you may want to:
- Rephrase the question to be more specific.
- Check which metric or filters are being used.
- Cross-check the result against a known report for important decisions.