Connect YouTube Analytics to Power BI

The YouTube connector covers your own channel’s analytics. Fields are grouped like this:

• Per video, per day: views, engaged views, estimated minutes watched, average view duration and percentage, likes, dislikes, comments, shares, subscribers gained and lost
• Video details: title, URL, thumbnail, length, publish date, tags, category, captions, privacy status
• Channel: current subscriber count, number of public videos, total views
• Breakdowns: traffic source, content type (Shorts, long-form, live, stories) and audience age and gender as percentages

Data runs about three days behind, because YouTube Analytics reports with a delay. The connector doesn’t include thumbnail impressions or click-through rate. YouTube ad campaigns are reported through Google Ads, not this connector.

Example: where the Windsor.ai channel’s views came from

For Power BI on the web, a query URL like this one loads views and watch time by traffic source (replace YOUR_API_KEY with your key):

https://connectors.windsor.ai/youtube?api_key=YOUR_API_KEY&date_preset=last_28d&fields=traffic_source,views,estimated_minutes_watched

On our own channel, for 14 September to 9 October 2026 (pulled 12 October), it returned 34,190 views, the same total the YouTube template showed for that period:

Source values in the query output: YT_SEARCH, EXT_URL, NO_LINK_OTHER, YT_CHANNEL, SUBSCRIBER, SHORTS. “Other” covers playlists, other YouTube pages, suggested videos, ads and notifications. Minutes are estimated_minutes_watched.

Adding creator_content_type splits the same views by format: Shorts had 9,313 views and 9 new subscribers, long-form videos 24,877 views and 30 new subscribers. Search brings most of our views, but viewers from external sites watch about three times longer: 0.92 minutes per view, against 0.31 from YouTube search.

Example: YouTube views next to GA4 sessions

Add Google Analytics 4 as a second query in the same report to see what the channel sends to your website. For the same 14 September to 9 October 2026, GA4 (field session_source_medium) recorded 379 sessions on windsor.ai from youtube.com / referral, 239 of them engaged. Set against 34,190 channel views, that is about 1.1 website sessions per 100 views.

GA4 also listed 464 sessions from accounts.youtube.com / referral. That referrer typically comes from Google sign-in redirects rather than from videos, so we left it out of the comparison.

Example: which videos keep viewers watching

The free YouTube template has a per-video page built from video_title, views, likes, shares, subscribers_gained, average_view_percentage and estimated_minutes_watched (screenshot alongside). On our channel for the same period, the most-viewed video, “How to Connect Instagram to Claude: 60-Second Setup”, had 2,650 views and a 51.45% average view percentage. “Connect Instagram to Claude in 60 Seconds” had fewer views (880) but held viewers longest of the top videos, at 59.16%.

How these numbers were pulled: Matei Ioan Nicolae (SEO Director, Windsor.ai) ran the test on 12 October 2026 on the Windsor.ai YouTube channel and GA4 property, both connected to one Windsor.ai account. The YouTube figures come from the connector for 14 September to 9 October 2026 (the latest day available on 12 October). The template total, the traffic-source total and the content-type total all came to 34,190 views. GA4 sessions are filtered on session_source_medium for the same dates.