camerondeleone
commented
2 years ago I am pretty regularly running into missing data issues using the Demo (Calls) project during demos. The main project obviously has good, usable data, but am not comfortable demoing from there (especially when many calls are recorded/distributed, etc).
It makes sense for live demos to be on a dataset that users can then try out themselves, recreating the analyses they are shown. Here is a list of the features I hit during a demo / where I expect to see what data, in case it's helpful for ongoing work on this — basically, here is my wishlist 🙏
Events
- Custom events (eg insight viewed/analyzed)
- Autocaptured events (specifically, pageviews and clicks, since these are concrete and easy for users to understand) attributed to both anonymous users and identified users. Need this for toolbar demo.
- event stats, ideally showing at least a few events' usage for the past 30 days ('how do i track my usage?' comes up often on calls)
- Perhaps most importantly importantly, events are attributed to groups ('Organization' 'Instance' and 'Project' are great). Currently demo data does not have group id, so it harder to upsell on this feature when some visualizations appear empty.
- At least enough events with feature flag data to show a breakdown on trend/funnel
Persons
- Anonymous users
- Identified users (preferably identified have multiple IDs so I can talk about how identification works)
Recordings
- currently use production data and navigate to my person object for this, but better not to. (They can be short/limited but should at least contain a few events each beyond simple pageviews)
Groups
- A few test companies, with fake metadata
@simfish85 if you have any suggestions or things you'd like to see, feel free to add.
simfish85
Twixes
Overview
A key component to any product-led growth or sales funnel is the ability to try the product before using it.
Right now, teams can get a feel for the product by ingesting their data and getting events in, but it's high effort and may not lead to the best, most accurate experience of the product due to their taxonomy.
Motivations
What about focusing on demo videos instead?
Goals
Goal 1: Develop an excellent taxonomy and a system that perpetually seeds data from it into a PostHog org
Summary:
Develop an excellent taxonomy. An excellent taxonomy... 1) Makes it easy to demonstrate the value of core analytics functionality (for example, funnel breakdowns) while feeling sufficiently random (the way a real product would) 1) Is comprised of events and paths that are commonly understood and applicable to industries
Build out a process that perpetually seeds a PostHog org with data given a taxonomy
Open question: Which industry should we base our first dataset on?
Goal 2: Expose the demo data project to the public
Summary:
After Goal 2: Demo can be sent directly in Slack and sales conversations at this point.
Goal 3: Build CTAs into demo environment, AB test demo signups
Summary:
After Goal 3: Wait for learnings from the AB test and optimizations into core experience before investing into further datasets.
Goal 4: Build out additional datasets
Summary:
@samwinslow I'd like to get your thoughts on this and chat about it later this week. Seeing how @jamesefhawkins' time looks I think it would make sense for us to prioritize some of this work before the work around expanding our pipeline. So I'm thinking this could be the next project after usage reporting (and before Next Steps - or @smallbrownbike takes Next Steps).