GitLab Analyze:Product Analytics UX meetings, 23 Aug 2023

Previous Meeting

Next Meeting

⏯

youtube image

►

From YouTube: 2023-08-23 Product Analytics UX/PM Sync

Description

Today we talked through some issues that came up in implementation of the Dashboard Designer re-design and found a path forward.

A

Hey this is the product analytics pmux weekly sync for August 23rd 2023. We have one agenda item today um discussing a demo of the proposed pivot on the dashboard in BC um yeah. Do you want to kick us off and share your screen to talk through this.

B

Sure, let's see share my screen right can you'll see.

A

You're in the agenda looks good.

B

Cool, um let me start off with the uh issue. That is the context for this jet item. That is, the new custom dashboard flow, specifically the dashboard designer MVC. So this is the first iteration that we want to launch to customers. It is currently um being tested by some internal users that have this feature flag enabled, um and with this MVC we have a new design.

B

The design looks pretty slick, um but while implementing this like, we ran into some like practicality issues in terms of just how the grid works and call to action is being shifted around and also some of the edit experiences a bit of a like has like a question mark on it, for example, uh and I think the designs include um anything that's like tells us how this title should actually be edited uh or what the edit state looks like.

B

uh Rob and I played around with some ideas in terms of like having it um like live editable.

B

This HTML content there's like an attribute you can put on for that or switching it to an input box um and eventually, while talking through this, we just ideated and took a bias for action since we're lacking a ux designer or a dedicated ux designer for our team to try and come up with an approach that solves the issues uh that we were like talking about in the concert and um yeah I've gone into detail here uh on the comments.

B

I'm I'm sure everyone in this call has already seen it, um but basically, what we saw is that, like we can maybe do like a iteration between what we have now and our vision and use some of the existing pajamas components and use some of our existing forms in settings and like the new MRN issue, screen as a inspiration to come up with something practical that works for now, and it's like good enough for our users for fish release and basically, what we came up with was this design that I can demo here.

B

um So this is the create your dashboard screen. We took the dashboard title added from um this is how it looks when you create a new, Mr or issue which I can just quickly show here. As you can see, it has a similar look and feel to it. You have like your page title describing the action that you're taking and then you have your fields that you need to input um so that you can create issue um similar if I don't have the title.

B

I can still call this primary field, but it will say: hey you need to go fix this and we're looking to like potentially deliver this as as a similar virtual experience.

B

um So what I also have here is, as you can see, you get error. This could probably be on that form group instead of like jumping up, but the idea is there um we added an empty State here uh to just show like hey, so it's basically guide first-time users into there's two things you need to do like one is title.

B

The second is ADD visualization, and once you did that we have some Progressive disclosure in terms of only showing the filter, then so that we limit the amount of control that users initially see, because this is only really applicable for I feel like users that are a bit more advanced and kind of know like that they're going to want to filter or like preview how their data is going to look I.

B

Think, first time, users probably going to ignore that filter, but it's still useful to to show, and similarly our edit experience um and that's the other part that this design kind of um neglected a bit is that we currently have an edit experience. It's not just new dashboards. It's also an added experience that we need to support and that whole step flow doesn't fit well in with the edit experience.

B

um So what we have here is um pretty much more or less the same. We just don't have that empty state that might come back in some form when we add the ability to remove dashboard or panels from the dashboards, because you'll maybe end up with something that has no like yeah, like users, may want to start from scratch and remove all panels. That does bring up the interesting question about whether we should let users save an empty dashboard.

B

My immediate gut feeling is yes, like I, don't know how you what they're going to use it for, but we probably don't want to limit their action without good reason, yeah um and then uh I'll I'll stop first there for any questions, because, like from here, I can talk about like a potential second iteration, but first I want to stop here for any questions or feedback about, like specifically change what you think, whether we should like take advice for action and go ahead with us or.

A

I think we should um I think bias Braxton going ahead. A couple of little nitpicks, one yeah move the the air into the right below the title, so it's on par or the the same experience.

A

um The other question I have is about the filter. um So if we're rendering data and showing the filter it's helpful, but if I'm editing a dashboard and I change that filter. My first instinct is that that Now, the default filter on the dashboard and I, don't think that that's true.

B

I see, that is a good point like I think the filter is kind of necessary here, because it gives you that same what you see what you get experience in terms of like you can already play around with your visualizations that you've added and give you a good idea of what it's actually going to look like. At the same time, you raise a good point in that users may think that this is like a static element.

A

For that in the tool tip that just says something that that indicates in this I don't know if we could, if we're achievement,.

B

It's all types, something.

A

Like the filter is not a favorable thing, um that is.

B

A

There for the data exploration, part of it, um but I wouldn't get hung up on that. uh I I, like this I think this is looking good. It's consistent with the edit experience. As far as the list of things are there, the title: is there I'm saving the dashboard, yeah I, don't know about letting users save an empty dashboard, but I think that again, that shouldn't hold us up. um I appreciate the bias to action. I think that this is a good path forward for the for the NPC.

B

Okay, great yeah, as as long as you're happy with this, and you think that this is going to solve most of our customer requirements for the MVC, then I'm happy to also go ahead. Rob. Do you have any thoughts.

C

um We've got a separate issue to discuss the visualizations right hand: sidebar yes, based upon Kevin's, work previous work um and trying to iterate towards that as well, um which we've been beyond discussed when we sat down and explained about the omvc structure. um There's it's kind of just to wanted to ask. There's conversations going on in this issue with um sir Justin Mendel yeah. Probably, um is that in relation to something we're going to be working on based off these designs, or is that in as a secondary augmentation like a next iteration.

B

That is probably a next situation, so what Justin um and I have been discussing when James has also been involved in this discussion, is about the call to action to open up the available visualization side drawer. So we have these other parallel issues to this dashboard. Designer MVC, which is like I, said to Implement our new ux designs, which transport migrates the existing dislike temporary sidebar, fixed thing that we have here, which looks pretty ugly into a side roller component.

B

uh The issue with the side roller component is, you need a call to action to actually open it um and the good news is. We now have two um options which I think are viable, um but I've asked Justin for his feedback in terms of like which he prefers like we can either have this elongated button slash pseudo panel at the top.

B

This solves the thing of we have a panel that moves around and it also um breaks the what you see what you get like experience in terms of like having a panel, that's sticky, to the top left, um because this is small enough where I think like. If we remove it, if users aren't going to be like, oh, this isn't what I.

A

B

Know set up in the edit experience and similarly we could uh make it in line with the filter and add it to the right hand. Side here, I think both Works um I think it's also something we can iterate on, uh but so far I've pinged Justin for his feedback, depending on what he says. I think we can go either direction.

B

We could also do one and then implement the other, but this is for a follow-up, so I think by the sounds of it from what we just discussed like we are unblocked to carry on with uh this first iteration of design, which I'm like demoing here and all we need to do in terms of like feedback, is just updates. The um uh error experience in terms of highlighting the the field that users need to to input.

A

um On when the dashboard is empty, we have the caller action to add a visualization. Is there an indicator in that sidebar.

A

That says, add visualization, then it lists them.

C

I think in the next iteration, if we have got that call to action somewhere, then we can add a call to action to that empty State like we normally would so uh that way, users can just click the button and it will open that side. Drawer.

A

Yeah yeah now I'm just thinking the list. Is it intuitive for a user to click, one of those to add it or to click and drag it in.

C

A

C

This in the code, because it's.

A

C

It looks like plain text or a link. It doesn't look like something actionable in the same way like the current designers yarn, showing off um with the check multi select checkbox that looks actionable um like I click. It I do something, um so we did discuss this um and ways and things that we might be able to do to make it more obvious visually that you want to do something with this.

A

Since we're already talking about an iteration out of that view into something we know is more actionable I, don't think we need to spend a lot of Cycles trying to make this iteration of it more actionable, um so I I think we can move on and if we get user feedback, hey I didn't realize that that was a thing. I could drag over I click to create visualization and then I recreated, something that was already there. We know that this iteration should solve for that, and we can more quickly move to it.

A

So not not too worried about it. It'll be something for us to look at in feedback um or kind of the hypothesis would be. This version tell user that they should or how to add something um where's, the other one didn't.

B

Yeah totally agree cool.

A

B

A

Is looking good um great work, guys.

B

B

Well, thanks for everyone to like willingness and openness to just collaborate and like quickly iterates on this um yeah I. Think it's a great display of gitlab values in terms of iteration bias for Action short, toes everything so well, then see yeah.

C

B

C

I did want to say one little thing about the two designs she showed both are lovely and, in fact, I'm kind of leaning to the first one, just because it looks a bit more snazzy, um but the one thing I did want to say is relating to James's previous point about the um filter. Looking almost like a preset uh when you do add that button in between it makes it even look, it looks makes it look even more like it's a preset value from a you know.

C

This is right next to the title, which is a preset and it's looking in a similar vein visually in a similar vein. um So yeah, that's the only thing I'd say regarding that is. We may want to make it even as James said to clearer that this is for just the current iteration that you're looking at it's not being saved somehow don't know how.

B

Yeah I think the the end Vision will definitely solve or help with that user experience, because I think the end vision is to have no filters and that you add filters to this um kind of as like a tokenized search that you have a button here that says, add filter that in itself could actually also be confusing, because now it could be seen as like.

B

Oh you're, adding a static filter and, like you, won't be able to to edit again, um but definitely yeah some copy or something I'm, also fine with hiding it for the first iteration uh that we take to our customers. Maybe we can get some feedback from our internal users once we get. This live um like out there on like gitlab.com, um and we can just like propose that question and say like. Does this make sense to you.

C

I would be open to hiding it yeah.

B

A

C

Only going to have a media data anyway, because we have no way of importing and the only place that might be a problem is edit dashboard. If you haven't received data in a while in the last seven days so, which is unlikely at least for now,.

A

Yeah, that seems fair enough. Leave it or hide it. I'm good, either way um and we'll get some feedback from users thinking about at least for our internal product management, Persona and developer persona. The tool that they're used to is one of the tools they're used to could be cisens. It behaves the exact same way where you can't you can change the filter, that's on the top of our states and things go um and it saves it. But it's not in the edit experience it more. A live edit of I'm.

A

Changing the panels on the dashboard or I'm changing adding a new panel to a dashboard I can change the filter there and that persists, but that persists whether you're editing a dashboard or not. So it's not something that you State per se, so it kind of mirrors that experience, so it shouldn't be like it's not totally foreign or totally um unknown uh to what a user might be used to.

B

Cool so both by the sense of it, it seems like, let's, let's hide it for the first iteration. It's going to be a really small change to hide this, so I'll include that in the NVC implementation plan. Oh awesome, in that case, I think we're done with this is General item, so I'll stop sharing.

A

Yes, I think we've done the agenda then as well um any other ux orientation questions we can go through. um Overall, here's synchronously I.

C

Guess I um I think I feel like I want to like voice make sure we're with an agreement is obviously we are currently looking to fill the design position um within our team.

C

um So as far as I'm aware and how I've been viewed this or me and yarn have gone back to uh just covering the design position, we're not looking to create a whole new mpz designs or create new entire new pages. It's just about um taking what we've already got and just iterating towards um where we want to be in the smallest possible steps, rather than like doing big full-blown Vision stuff like we were doing originally.

A

Yes, um small uh baby steps, iteration uh the only thing where I might ask for you to to do some design. If we get that far without a designer is going to be that new usage quota page, it is modeled a hundred percent after the existing pipelines usage quota page um and that's what I hacked together a buyer brand based on. So there's not a lot of design work there. It's just reusing existing components. Unplugging in our data versus pipeline data.

C

Well, that sounds like we can just first iteration duplicate pretty much. What they've got just change the fields to match ours, and then we can based off usability analysis by seeing what users say about it in a feedback issue, we can adjust in small steps again.

A

Yeah I think that we'll have a designer by then anyway, but um that's the only new page running through issues in my head. That's the only new page I can think of that. We'll create they're going to need. You know all new design or new implementation, otherwise we're working on existing Pages iterating.

A

um This is the big one and we're going to wait for new designer to be in place before we start working on a new vision. First visualization designer.

C

A

A

Visualization designer it works.

C

I mean thank God, we're not the ones having to do the design now. Yeah.

A

There's giving you some research on on that we're gonna go. Do some user studies I? Think excellent.

A

I think that we are at the end of the agenda, then Gents.

A

Cool thanks for the time this morning, I appreciate it. I'll get this uploaded unfiltered.

C

Thank you and thank you for running it Jan as well. Going through that discussion, I appreciate it.

C

It's cool cheers, yeah.