-
Notifications
You must be signed in to change notification settings - Fork 1.3k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Replace "save funnel" with "save funnel to dashboard" #4979
Comments
Updating here. Problem we had with funnels is that the Add to dashboard button was accidentally dropped. #4986 fixes this. This fixes the problem in the short-term. The real problem we still have is that we have a confusing and inconsistent (different for funnels than other insights) experience when saving insights. Think it's worth putting more effort to figure out the right UX, and have the same UX across all insights. Keeping this issue to work on it. @clarkus I'm not sure if the latest version we're considering is the best way to go yet. Some concerns that come up:
|
I was also confused about this 😅 . I think the core way funnels diverge from other insights is this concept of saving them to the project while not saving them to a dashboard. If we drop the ability to save a funnel to the project, does this list go away? If this is the case, that's probably a breaking change that should be communicated to users. Otherwise there won't be a way back to an existing funnel if it isn't already on a dashboard. I am trying to better understand our model. If this is changed for funnels, my understanding is that the existence of an insight on a dashboard is the only way to save state. There is no concept of saving an insight outside of its placement on a dashboard. Is this accurate? |
There kind of is a third system. If you go Insights > History, you can save "reports" that way. However that's a third system on top of dashboards + funnels and one that I don't think is widely used. See also: #3408
Conservative proposal: Put the feature behind a feature flag, enable the flag for everyone who has been using it so far and see if it gets any usage after that. If it does, then talk to them and see whether there's a valid usecase. Eager proposal: Just nuke it and see if someone complains. |
Given we still have the history component, I'd be more in favor of the nuke option. That would clarify the "save to dashboard" label and be much more straightforward. That said, we should align this across all insights if possible. We have the same model for each, so aligning the button labels would be a good change to reinforce the model. |
We definitely have a bit of a messy experience here. Here's what I was thinking (mockups for this are on Figma). Feedback? @clarkus
Things outstanding:
Thoughts @clarkus ? Think it's worth spending time building full mockups for this, happy to do them or if you prefer to. |
@paolodamico I have some ideas for this and I am happy to spend some time on it today. I think the most obvious and valuable model is going to be saving insights as standalone objects. This gives the user a means of starting and stopping work on an insight as needed. Saved insights / reports could also be a centralized place to manage and update insights used on one or more dashboards. |
Here's a quick stab at stand-alone insights based on the existing components for managing dashboards. Note that I am redoing some templates to make them a bit more composable... if you see repeated data, ignore it for now and know that I am cleaning this up as I work on the core problem. This is the first step in establishing a clearer model for managing insights. My next steps are to clarify the saving workflow across types of insights to make it more consistent. Notable changes to the workflow:
More in progress work at https://www.figma.com/file/9yWtngNb1AIuf6KmXaEPJA/App-doodles?node-id=587%3A0 |
@macobo @paolodamico I spent a bit more time on this. I'm not quite happy with the workflow, but this would at least get us to a point where we can save insights independent of dashboards. My problem with the workflow is that building an insight doubles as an exploratory flow. You don't always start out with the intention of saving your work - you very well may cancel out once you've answered your questions. I don't want to diminish the ease of getting into that exploratory workflow. For this reason, I've broken away from the model we use to create and save dashboards. In the dashboard case, you create the object, then populate it. In the insight case, you populate and preview the object, and then optionally save and / or add it to a dashboard. You can see that I have two main actions on the create insight view - I am also slowly trying to consolidate and improve consistency across the insight creation views. If you see any major layout changes here, just know that they're an in-progress thing mean to fill the gap while this model is figured out. Thoughts? |
Lots of good stuff here! We’re conflating a few things here (which are all semi-related), so I’ll try to break down my thoughts by topic. Saved queries, queries on dashboard, and recent queries Defaulting to Your Insights Analysts/users who are running a bunch of ad-hoc queries @kpthatsme may have some thoughts here, but if we want to optimize for customers exploring data, we may want to make the query stage a little more visible. (I wonder if there’s a way to combine an empty query stage and some number of saved reports?) BUT, this is UI seems great for read-only analysts. “Want to run your own query? Great! There’s a button for that.” (But New insight might not be the most obvious way to label it - as a user invited to view reports, I’m not sure what that button means at this point.) UI Even if we were to implement smooth scroll that moves you down to results after you run the query, I wonder if we can tighten things up a bit so they aren’t so far below the viewport? P.S. I’m loving the shading between the steps and the filtering - makes it super clear what you can drag up/down! |
Yeah I think this is the case for all insights, but it's especially problematic for funnels given the unbounded nature of steps. The more steps you have the more you displace the visualization. I agree with your feedback. I am still working to try to tighten it up while still making it feel obvious. |
Ignoring layout changes for now, but the overall UI improvements here are amazing!
|
@paolodamico @corywatilo this is really the core of my issues with the design I shared above. I want it to be straightforward to trigger this exploratory flow but also want to give standalone insights a bit more value by making them more discoverable. The data exploration use case seems like something that should be available on every page of the product. Anytime you want to validate an idea or explore some data, you click on a prominent "new report" or "new insight" action and you're immediately in that view. What would be super cool is if it were modular so that you could trigger that view without leaving your current context. Example - I am looking at a person session, I see some common attributes that give me a general idea I want to test. I trigger this building view and it overlays the current screen. Given this is targeted towards an exploration case, it's optimized towards seeing data and results as soon as possible. Once I test my idea, I can save that work for later, or just close it with a prominent action and I'm right back into my person session. You could also do this via a new tab, but there's some power to having this on-demand building experience without hijacking the user's browser tabs 😉 .
I am working on this today actually. I don't like it either. I want to be able to prompt for those other insight details or at least provide some smart defaults so that it's less critical to provide that information when you're first saving. The general ideas was to chain the workflows, but I'm not sure they need to be separate.
I could see "save as" working. You could argue that providing a duplication option could offer the same functionality just from a different starting point. Maybe we should provide both.
Dashboards could just aggregate the tags from the insights they contain.
I'll work on this today as well. Totally with you on making things easier to find.
Yeah I think pinning impacts the list of items in the overflow menu under dashboards. The same could apply to insights, but I'm not sure it's super obvious. I'd question if we need the overflow menus if we provide this structured listing of all items. We also have a pretty capable search that can directly open objects... we could do more to make that the centerpiece for power users quickly jumping to a known context. Tagging items as favorites feels more approachable to me simply because it's more prevalent in other products. |
Makes sense! I like the idea of performing quick insights like you do for email composing in Gmail, sounds like it's worth testing. Aligned on the rest! Let me know if you need any help. |
This update is focused on the insight creation view, less so on how you get to it. This is intended to front-load the insight details (name, description, tags) in the header of the page with reasonable defaults. For example, we have a placeholder name and the description (if unmodified by the user) would automatically update to match the funnel step progression. A user can click to rename / edit the name or description. Tags can be added, as well. This would give us a one-click save action with no prompts. If the user elected to save and add to the dashboard, they only get a prompt to select a dashboard. There are a ton of other changes in here that I am working on in related issues. The goals I'm chasing for this iteration are:
Thoughts? I haven't put much time into the overlay idea yet, but it would likely follow this same format. There's a chance I can get the query building to work in a 320px column, in which case we could show the visualization area more prominently. I'm working on that next. |
After working through this a bit more, I think the base improvement to be made is collecting the insights details prior to saving. If we don't have to enforce unique insight names, we can provide a reasonable default that makes saving a one-click, immediate action. If the user opts to add to the dashboard, the save action happens first followed by a prompt to add to a specific dashboard. If we do need to enforce unique names, we might consider introducing a simple numerical ID into the insight name. This would be enough to make it identifiable and findable again. One thing I'm not clear on is the relationship between insights and panel names on dashboards. Can panel and insight names diverge? Are they always the same? Is there any other insight attribute used to ensure uniqueness when added to a panel on a dashboard? I am going to save the rest of this solution for the larger issue at #3408. |
Is your feature request related to a problem?
Currently, it's possible to "save" funnels in a way that's incompatible with the rest of trends.
Describe the solution you'd like
It should be possible to add funnels to dashboards.
Describe alternatives you've considered
Additional context
User was confused in papercups: https://posthogusers.slack.com/archives/G01JXEDAL22/p1625235789002100
cc @paolodamico
Thank you for your feature request – we love each and every one!
The text was updated successfully, but these errors were encountered: