We experimented with YouTube from November 2022 to July 2023, but have paused creation and publishing for now. We may try again in the future.
Although videos were driving X00s of views each (some hit X000s), and we received some positive feedback, we didn't see an increase in signups, traffic, or mentions from the videos. For example, the video on why and how we use GitHub as our CMS got 3,000 views in 1 one week, but made no noticeable impact on signups.
We also were starting to run out of obvious tutorial and SEO blog content to turn into videos. Basically, we ran out of low-hanging fruit. New videos would have taken increasing amounts of time.
Learnings
- The less PostHog-related videos did better across all three types.
- Title and thumbnail matter more than video content.
- YouTube growth compounds heavily, but it requires multiple years and 100s of videos to reach the scale we'd need for a meaningful impact.
- Lighting is what makes the most difference in video quality.
- The PostHog demo is the most important and popular video we have. It should be updated at some point.
- OBS works well to manage the recording of videos (screen, audio, webcam).
- Only 2.5% of video traffic came from
posthog.com
. 60% came from YouTube features like search, channel pages, or recommendations.
Types of videos we made
- Tutorials like How to bootstrap feature flags
- SEO-ish content like The best GA4 alternatives for apps and websites
- "Essay" videos like The modern data stack sucks
YouTube comments
YouTube comments are posted to Slack using Make. It's a tool similar to Zapier, except Zapier doesn't support YouTube comments.
For access to Make, ask anyone in the Content & Docs team. They're all admins and so they can add you.