Super guided query running

[timgl]

Is your feature request related to a problem?

Running a query on PostHog is pretty complicated. You likely need prior understand of product analytics to make sense of it.

Describe the solution you'd like

A guided way of running queries that walks you through the questions you want to answer and sets up the queries to do that. James G said the goal should be for you to learn from the product.

This should look something like typeform.

Describe alternatives you've considered

Understand product analytics in depth.

Additional context

Thank you for your feature request – we love each and every one!

[paolodamico]

Linking the doc from our discussion today

[paolodamico]

Alright, wireframes coming out of the oven. Please refer to the doc above for full context on the options, goal of this feature, persona, etc.

• Option 1: Full guided wizard (with optional best practices templates add-on).
• Option 2: Subtle hand holding on normal flow.

Option 1

Wireframes: here.
Can be quite powerful for both personas. Quick keyboard navigation can greatly speed up insight creation for seasoned users. It can also increase the discoverability of less obvious features (e.g. math operations, breakdown, ...), as we provide clear examples of the end-results.

Some downsides: it'll introduce another query building interface to maintain (risk of disconnect with main insight builder), it could get very complex if we want to handle every possible case (the flipside being it wouldn't let you do all types of insights available).

Templates
While working on the templates feature for Option 1, I thought of a few additional considerations?

• Templates will probably be more useful if thought of as dashboards rather than individual graphs? A single graph is provides a very specific view, yet it requires almost the same effort to set up as a full dashboard (e.g. if I input my revenue events and properties, you can give me a ton of different views on revenue, not just a simple graph).
• Templates will require a very individual hand, researching and creating each one, and accounting for different edge cases. Basically, each template could be its own sub-product.
• We probably would have to work out a few very concrete examples and test them out to understand its true impact?
• We could also do more generic templates, e.g. for instance to discover functionality around more complex features (e.g. number of events per active user).
• We could consider extending the depth of functionality, and not only show graphs, but fully parsed conclusions from analysis (e.g. your retention is good 👍 or your retention is bad 👎 ), but this could make each template more complex.

Option 2

It's more subtle and may be worth doing anyway, we reduce the number of clicks and bring attention to the relevant items on the screen at each step of the way. We do introduce a home to the insights page to make the decision of what type of graph you need easier.

Downsides: Still requires some basic product analytics knowledge. Convention over configuration, might make some more complex features a bit harder to discover.

Wireframes for single series graph

Wireframes for multiple series graph
If a user clicks on add graph series, we basically guide them through the process of selecting another action/event. We also enable formulas, and general filters.

Additional

In both cases there are certain UX optimizations that we should include, some details:

• When doing a breakdown default to a pie chart if the number of props is less than n (~10 maybe).
• If you only have a single series, hide one of the filter types (i.e. you can filter only with general filters or with specific filters, but not both).
• Same thing if you only have a single series, we hide the formula input, as well as "Value" mode for bar charts.

[timgl]

This is great.

It feels like we could to the first wire from option 2 pretty easily. We can see if people like that way of presenting information and build on that (or whether they skip straight to the tabs in the top). That'd feel like a good MVP to me. We can then still go down route option 1 or option 2.

I'm not so clear how the rest of option 2 would work, I might need to see high fidelity mock ups but we can wait with that.

[posthog-bot]

This issue hasn't seen activity in two years! If you want to keep it open, post a comment or remove the stale label – otherwise this will be closed in two weeks.

[posthog-bot]

This issue was closed due to lack of activity. Feel free to reopen if it's still relevant.