Dataverse (Direct) setup guide
The Dataverse (Direct) build parses Copilot Studio conversation transcripts live in the Power BI model — no infrastructure to stand up. The template is pre-set to Dataverse, so you supply two inputs, sign in, and refresh.
1. Set the two required inputs
The template is pre-set to Dataverse — the Source Mode parameter defaults to Dataverse. You
set two required inputs:
| Input | What it is |
|---|---|
| Dataverse Url | Your Dataverse environment URL. Accepts several URLs separated by ; for multi-environment. |
| Org Data CSV | A CSV of user and department context for the report’s breakdowns. |
There is no “Agent 365 CSV” input on this build. (Transcript CSV Path exists in the model as an alternate/shared-model expression, but it is not a required startup input for the Direct path.)
IDSL-DIRECT-001
CaptureCapture the StudioLens - Dataverse template's Edit Parameters dialog using synthetic values. Show the Dataverse Url and Org Data CSV inputs; exclude real environment URLs, tenant names, file paths that reveal identities, credentials, secrets, and customer records.
Proposed altStudioLens Dataverse template parameters with synthetic Dataverse Url and Org Data CSV values.
2. How the model reads Dataverse
The model reads Dataverse through OData.Feed against the Dataverse Web API at
/api/data/v9.2/ (built from your Dataverse Url). It does not use the native Dataverse connector.
3. Authenticate
On the first refresh, choose Organizational account and sign in with an org login that can Read the Conversation Transcript table. Set the source privacy level to Organizational. This path uses no app registration and no client secret.
IDSL-DIRECT-002
CaptureCapture the Power BI data source sign-in showing Organizational account selected and the privacy level set to Organizational, using a synthetic account. Exclude real usernames, tenant names, environment URLs, credentials, and secrets.
Proposed altStudioLens Dataverse Organizational-account sign-in with Organizational privacy level.
4. Multiple environments
The Dataverse Url accepts several environment URLs separated by ;. The model pulls
conversationtranscripts from each environment — each fetched inside a try ... otherwise so an
unreachable environment is skipped, not fatal — tags each row with its environment, and unions
them.
In the Service, configure per-environment credentials under Settings – Data source
credentials, and keep privacy levels consistent (all Organizational). If privacy levels differ,
Power Query’s Formula.Firewall may block combining the sources.
5. Org Data CSV — path options
The Org Data CSV can live in several places; the connector and gateway needs differ:
| Org Data CSV location | Connector | Gateway |
|---|---|---|
| SharePoint file URL | Web.Contents |
None (cloud-to-cloud) |
| Local / synced path | File.Contents |
On-premises data gateway required |
| UNC path | File.Contents |
Gateway required |
6. What the parser produces
The Direct model’s M functions parse the raw transcripts into the transcript tables:
| M function | Output table |
|---|---|
RawTranscripts() |
Raw transcripts |
ParsedBase() |
Parsed activities |
Parse_Sessions() |
Agent Sessions |
Parse_Turns() |
Agent Turns |
Parse_Errors() |
Agent Errors |
Parse_SubAgents() |
Agent Sub-Agent Calls |
Parse_Performance() |
Agent Performance |
The Agent Catalogue self-derives from the parsed transcripts.
7. Verify the connection
The built-in Dataverse Diagnostic table returns live row counts of conversationtranscripts and
systemusers. If conversationtranscripts is 0 but systemusers is greater than 0, the
connection is fine — the environment simply has no transcripts in scope.
Note on credits
Message-credit / consumption analysis is not part of the Dataverse (Direct) build. The PPAC message-credit pages are available on the Fabric path only.