Discuss our relationship with Self Managed
Goal
- Surface our current state in Growth in relation to Self-Managed vs GitLab.com(or any SaaS instance).
- Challenge that state and see if there are plans to change that relationship in any way.
Historically
Growth has limited experimenting and a lot of our Growth concepts like onboarding to GitLab.com for various reasons. This may not be a comprehensive list:
- Hard to get tracking data like we do on GitLab.com with snowplow, which makes it hard to be data-driven for self-managed.
- Usage Ping data could be used with internal events, but then a lot of instances are air gapped so the observable data will be small and experiment cycles longer due to the frequency of that data(weekly at best).
- Reticence for growth type alerts/banners from self-managed instances that feel like marketing. We've seen this in the past with issues coming in where some of the Growth items targeted all GitLab instances
- We'd need to be more strategic in this space so they would come off more as 'helping' than marketing as some have felt them to be in the past.
- For the alerting/marketing type alerts, we'd need to consider a instance level setting to opt in. Of course it isn't likely anyone would opt in, but to consider opt out, we might need to bring in legal to the discussion.
- Self Managed instances are far slower to update, thus further reducing the ability to be data driven.
- We have certain concepts with our open core product(ce) where we need to be more strategic with and keep concepts of free in place. i.e. guiding
ce
(free) instances to paid licensing.
We do target all instances of GitLab for more general feature improvements that are not intertwined with subscriptions. Examples of those are:
- invite members and groups modal.
- invite registration improvements to the registration area.
- others?
What would targeting Self-Managed instances mean?
Currently a lot of our things like Onboarding and LearnGitlab are namespace in focus due to where the subscription is applied at that level. Wether that be for trials, paid subscriptions or LearnGitLab itself.
To translate these concepts, in any way, to Self-Managed, they would likely need to be instance focused due to where the License(subscription) for those instances are applied.
This may also mean re-imagining experimentation and our relationship with it in terms of self-managed instances. In the past we have indicated that we'll use GitLab.com to drive our data driven decisions that affect all instances. However, it seems like less of our experiments and features lately translate to all instances of GitLab due to our focus on conversion.
Question
Is it time to challenge this concept again and if so, how might we go about it?