GitLab CI/CD UX Team Design Reviews, 25 Jun 2020

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: Verify & Release UX Team Design Review | 25 June 2020

Description

00:23 - Veethika Mishra: DAG visualisation MVC - Public issue

15:16 - Rayana Verissimo: CI/CD Group Dashboard

A

A

Welcome everyone to verifying the release, design review weekly session and we are having a couple of interesting items on the agenda, so we check I'll just hand it over to you. If you would like to share that one sure I would start sharing my.

B

Screen readily.

B

Alright, so in the agenda, you would see two links in the future link. The two things that I would be talking about today is number one: the public facing issue for feedback for gathering feedback for the dag MVC that was very recently released and the other one is a discussion issue which was kind of a backstage thing for this first one, that's where we were trying to figure out what should go in the public feedback issue that would catalyze the community engagement and so on so yeah.

B

It was a very new approach for me because, because the first time that I had a chance to conduct research, you know in something that could be participated by community users like a public issue, so I had started off with consulting with Laurie around what the discussion guide should look like and what are the things that we should be considering, and it was a pretty lengthy discussion, but before it could come to an end, the feature was released and the link was activated for the issue.

B

So we had to hurry a bit and put information on the other one. In a haste.

B

A few things that I did here was I looked at some previous research solution, validation issues by other teams and I tried to observe, like what are the good practices, I mean or have people done in the past that has worked out for them, and one thing was putting the questions and threads like a separate comment, so that they can have separate trails for each one of them, and so eventually, when I would be transferring these feedback to, let's say a mural board for affinity. Mapping I, wouldn't have to put in a lot of effort.

B

I did that, but unfortunately it didn't work out so well. For me, I would get to that just to give a brief context about the data visualization just in Thea. So what dag does is when it comes to pipelines the jobs they are not always communicating. In a streamline fashion, there could be scenarios in which the communication is very interlinked and even though it could be visualized, it would be very difficult for users to make sense of the visualization and follow the connections and the triggers. So that's where that comes into picture.

B

It simplifies the interaction for users with the visualization so that they could get the exact information that they want from it. Now we began with a set of assumptions, but what happened through this is that I observed that there's another set of assumptions that we didn't even know we have. For example, if you look at the designs for dag everybody would see this horizontal visualization, but from some of the comments, it kind of surfaced that, in majority of the cases the length of the visualization is exceeding the width.

B

So there's been a conversation if it would make more sense if we turn this by 90 degrees, that's one, and then we had some questions about the colors which are being used like. If you look at this there's a lot of colors, because it's a very complex pipeline, it turned out, first of all that it's most of the colors which are being used here, which are from the charts colors from pyjamas.

B

They were probably not all of them had gone through the accessibility testing, even though there was a huge effort around it and the users who only end up seeing brown pipes crossing each other, which is like I feel, which is a huge problem for us. Then yeah like I, was saying: I tried putting all of these questions and different comments, but looks like they were very royally ignored throughout. Nobody commented on any one of them, but nonetheless we got a lot of inputs on this and very good comments.

B

That could would definitely allow us to improve this feature going ahead. So I was discussing about the next steps with Tao on this, and it seems like, even though there is no heart, stop that we would become any cating to our users for providing their feedback on this. But internally for ourselves we would still I mean let the feedback flow in for another two weeks or two and a half weeks, and then try to analyze the feedback and come up with some actionable points for ourselves, so that we can work on them.

B

For the upcoming milestones. That's the progress for now and yeah. There was a lot of learning involved in this entire exercise and that's all I have to share for this.

A

Just before we cannot go into any discussions, do you? What do you have any like questions to the team? Any questions you are like looking for answers and maybe they can help answering.

A

B

Would definitely like to know if somebody has a past experience with the.

B

Moderating the discussions that happen in these public issues, because that's one point that I am facing problems with.

C

I'd say: that's undoable almost like it's gonna be a mess regardless.

C

It's it's yeah, I. Think.

D

It's very difficult and.

C

I'm I mean we can't kind of be certain right that they come in at of the issue through the feedback link. So you could test, but I have really wondered like the sub discussions have been along. Like has been a long point of the case where people are not lining on like on at all answering the sub discussion, partly because some people, at least in get lab or entering through email, or something like that, where you don't even see some discussions, but otherwise people just like they just want to be.

C

You know easy easy comment, easy comments at the bottom of the other of the page. That's that's the only thing they will see.

A

You were referring to the unanswered comments right now, Dimitri on the issue or I.

C

Yeah I think this is kind of like a common case. Right like we have all had these issues where we have put down multiple comments for people to answer, and they just see the last one and answer that one and forget about all the rest. It's it's a theme like it's. It's it's hard. It's a hard problem to crack. I.

A

Think something that can help is always like point. Your question to somebody like. If you ask a question tag, people who you think and have answers if that, like technical question, that one of your engineers, if that's product related that p.m. if that's even you acts, you can tag the whole our team yeah. What like specific designers think tagging some people help. Maybe somebody else has other advice is how to not have these cases.

E

And just looking at at the thread and the issue, I would say: keep the source software updated, for example, with the teams or like some learnings of findings that you already have, because maybe if people are yeah just discussing the same thing over and over, you have too much feedback. You can point to them that yeah. These are the things that we already discussed in the Asian description, because the longer the conversation gets, the last people are going to refer to. What's in the middle right of the that, much much content and.

B

I think that again comes back to a moderation. Issues of one has to actively keep a check that people don't engage in a similar conversation again and again.

E

Read what we discussed about yeah, wanting that we didn't release management is that whenever Jackie sees that there is customer feedback somewhere, she schedules a call to with these people and then you'd get all the context in you know. 30 minutes 45 rather than going over and over in the tread and muscle, might be helpful. If you find these figures, that's yeah really involved with the issue.

F

Even if they don't have like a reply, too long, I took these check mark, they won't say like okay, that makes sense.

F

We're gonna have mine or discussion next week or whatever something that allows them to to see that their comment was seen by someone from the develop team because, like I, think that's the way to encourage them to keep keep leaving these comments if they feel that they are just like talking in a vacuum and no one's paying attention, then, like they're, not gonna comment again, we want them to be always bringing their thoughts and opinions, so I definitely try to under a lease like with an emoji but definitely I.

F

Try to you know just reply to them with something.

E

And also my experience, people are always gonna be in title of course, of an opinion, but it's up to you to say: okay, I have enough material. Now, let me regroup and then yeah get back to them with a different proposal or with the yeah part of a solution, because you're gonna be like keeping track of good friends. It's yeah, that's very hard like what immediately say.

B

There's one more small thing:.

B

I had yeah I transferred comments from the slack thread, so there was a panel discussion going on of the slack thread which Tao had to intervene and made them stop and how to redirect them to the issue. It was getting like very intense they kept on going on and on about certain things, so I. What so? What I moved from slack to here was everything that concerns you ex and research, but so I also have comments like the hover effects.

B

Omg like the emojis that came in so tower feels I think a little differently about this. She thinks that maybe we should only have actually this stuff, but, in my opinion, like these small little things, help us to understand like what worked out, and they would probably also help us in taking decisions in the future about like what are the things that made a positive impact on the users and how what are the things they reacted positively towards. So I would want to know the reaction of elements I.

E

Agree with putting everything there, especially that it's so like rough feedback right and you can link to a select, read it's gonna disappear in the next 90 days. I think it's. It's valid, really put an information unfiltered in there, especially in the comments, and if something, for example, is that the hover effects I wore to follow up, you open an issue about that or you can group all this this in in an epic or something I.

E

Do that quite a lot, especially I think when I was working with attach pipelines, there was a lot of feedback in slack and we treat that as customer insight customer input because we are using get lab right. We are not putting it so I see the value.

C

Great as well I think like, if you, if you change the words you will put in some personal bias, and it is about the reference to original comments, think.

B

So any other follow-up questions on this.

B

It's not the lichen structure.

B

Thank you. So much everything.

A

Thanks for sharing with you guys really interesting to see how this feature is going, I guess it's gonna be a big learning. For us, maybe yeah I'll be in here is to run a little retrospective around that, because I know it was a little bit of a complicated process and I know you kind of like arrived in the middle of that, so you know probably would be very helpful if you would be mounting all of kind of like that.

A

The issues, the good things, the bad things that you have seen and later, if you could be brave it, probably even among the team, or at least between me and you- that would be really good. That.

B

A

Very useful for me to see how can we avoid some of the things that we've done here and improve the next I'm, not saying it's all bad course, it's great great yeah that we are moving in the direction, but thanks a lot for sharing, so we still have some time in hand I've seen that you put another item for the future, not sure if you intend to share it today. If we have time, maybe it's not yet ready. But let me.

E

A

That as well, okay,.

E

It's okay, it's just because I'm still I like trying to think about exactly what I want to get feedback on. So maybe, if anyone yeah, if people have a topic where they really want to get feedback, you can take over if not I, think share my screen.

C

We couldn't discuss this morning right, Ayanna,.

E

Let me open up my my figma file first, this is regarding the CI cg group. Password director. Best words, name is depending Jackie and I are calling it CSD.

E

So treat it like that, so just for some context, let me find the issue here. The.

E

Dashboard just for context where it's my involved here: okay, yep.

E

Our managers directors people that need you know to manage multiple projects across different groups. They don't really have an overview right now, we'd get lab that allows them to see how their projects are doing right. You have all these different views. Where you can see environments where you can see releases, you can see, add appointments, but everything on a project level.

E

If we want to give users the ability to on a group level so, for example, with get level of level, see the last pass pipeline for all projects or to see the latest fight the last five deployments. The last five release is upcoming releases, etc. So that did these users these directors they can have, of course, a better overview of everything that's going on in their projects or in their groups, but also that when we look at other tools or competitors, this is something once we have this ability. Something's.

E

That's really gonna bring us closer to tools such as z-bo labs right, where you can do everything, and you know we manage the information around your releases, so that means say I had to hear this. Let me just double check: yes, director CSD dashboard. This was actually validated by Jackie I think a couple of months ago when she was interviewing customers around managing release at the group level also deployments, but also we have a research issue or a research inside open two years ago.

E

Let me add here that already say directors or you know: users need the ability to track project status at a group level, so the CIC dashboard right now we are looking at displaying five and I should be sharing my screen, but.

E

Yeah here yeah, so this is a bit much yes, so we are looking at show displaying five main metrics in this dashboard, which is like the number of releases or like a total count of releases in all projects that belong to a group. The percentage of projects that have releases the last five passed and failed pipelines, the duration of the last 30 pipelines, for example, in the specific group. So these are very low.

E

Fidelity, mock-up and Jackie actually put this together and was really helpful to help me visualize her perspective from this dashboard and if you're familiar I, think I can ID. I can also say that if you look at tools like CBO labs, they're, very powerful dashboarding features right. So when I look at this I feel like yeah. The problem is solved itself.

E

We have all these endpoints that we can just connect to whatever graphs and show on the UI, and then here we have a dashboard and that's actually what we're going to do for the MVC just bring. You know some graphs and yeah. Here's your static dashboard on a group level where you can see the health of specific projects, so users are not gonna, have the ability to customize this dashboard for now. But then, when I start looking at this, it becomes a bigger problem, because we know that dashboards with gitlab are a bit of I.

E

Don't know, say problem, but they're, inconsistent and I and neek is working on. A new post is working on political, like the analytics, pyjamas component initiative and that's awesome, I'm also involved there.

E

We have I, think monthly conversations around pyjamas and dashboards, etc, but I think the main challenge to me is that to define the vision for now the low-hanging fruit, because when I start looking at words at the group level- and this is what I have for the screen here- is that there's at least four ways to show and display data right now in the product and when I think about the MVC and what we need to deliver.

E

Maybe in like into three milestones, it's probably gonna be like a year-to-year, long effort, my current struggle- and that's maybe something that you, my designer friends, can help me is to look at when I look at the product right now and I see all these different dashboards.

E

What is the simplest easiest way to just just create this CSV group dashboard without having to redesign it in six months, knowing that the pajamas components are you know, being proved and that there is discussion around this dashboard design that the secured team is doing different things that yeah analytics components are being built?

E

Please feel free to stop me and ask anything because I know that it's a it's quite a lot. So what I did here and then this yeah this file has pretty much everything that Jackie and I have been working on, but just in a visual way. So we know, for example, that yeah in organizations like it lab. We need to give this reliable visualization of project statuses, that we can easily aggregate these metrics in the UI, but questions other than the UI aspect.

E

What is the idea workflow for users that are tracking metrics at the group level right? We don't really have anything, that's solid right now in the project. Where do they want to ask access this thing from it's really dis, leaving the instance level or at a group level how the users want to compare metrics in a single view, what metrics they want to prioritize and why? What additional interactions can we already foresee?

E

And how do they want to stay informed like do they want to take any actions on this on the interface when they want to follow up like create issues fix things? From this dashboard view, it was like I started inflating the problem, because I think that's gonna be interesting for the future and then before I go back to the UI. Just give you an overview of the jobs to be done, like super simplified, is to find stash port, compare the metrics and then, in the future, add metrics at projects.

E

Customized view share, metrics we're not even talking about these things right now, so we really focus on this one job to be done. Look sorry about this one which is C and compare these metrics and in a way to link it with other product areas. So let's go back to what I want your opinion from more or less because very much in my head right now, so I start with the analytics dashboard. This is like what implemented in the product. I know that this, for example, secure team.

E

There is a change in the in the secure dashboards, I think the operation dashboards that it's in master. It's gonna be released this week, but it's not in the product yet so with all these things, I told you when I look at these dashboards. I started looking at the thinking of the positives and negatives of these views. This UI I'm shared views compared to what we want to do for this CI CD. We give good dashboard so, for example, the analytics view when you land here there is a choice.

E

You can see it's a drop-down menu and then, when you click select page, you can select issues. Dashboards, merge, request, dashboard, whatever that updates the UI.

E

This is one. Then we have the value stream analytics dashboards that it's a bit more complex, but then I think developers are working on graphically well, but the performance of this dashboard is really yeah. It takes a lot of time to load all this content and I wonder that the more we add things to the group dashboard to see I see a group dashboard I am already worried about performance, but that's for the future, and then we have the security dashboard that is being updated.

E

I saw some designs from Andy that is similar to the previous one, but again it loads like a thousand items at a time, and it doesn't really keep that that's worth looking for and, of course, last but not least, the environments dashboard Dimitri. You know this. This hunted us for a couple of milestones. It's there, people use it, but they cannot really get any value from it. It's a copy of the operations dashboard, so the environment, dashboard.

E

You have these cards and then you can see yeah your wide projects and then you can see the the status of the last best environments and the last appointment, and you can see it's a card view.

E

So, when I look at all these things and I start thinking about what would be the CI CD group, dashboards, I feel cornered. That's how I feel right now, partner with the four or five different dashboard times and get lab, and what should I take as the source of truth for this new dashboard. So I'm not gonna, go into details here, but I just try to put the data that we want to display.

E

So, for example, as I mentioned, before total count of releases number of coming releases like this graph weights, comparison of historical versus upcoming releases, last five updated releases, less five pass and fail pipelines, etc on ey and then using this different dashboard views that we have implemented in the product. So this one was is like static. It loads everything on page.

E

Second, one has that drop-down that I think that doesn't really give users like much discoverability about the options, and if you want to compare things, you have to click load weights and same thing but yeah them.

E

This is still like a first version and I already read this over Nadia in and Jackie and there's a consensus that we want to have everything one for you and I'm gonna prototype this for the validate with users. So that's you know layer of navigations, you you land in a dashboard. If we can drop down and then you select what you want to see, it's a bit yeah, it's an overkill but broad questions. Based on this, this story, I told you is anyone working dashboard, that's the first thing!

E

If so yeah, let's talk about it and kind of similar to it, music, a question: have you worked with something like like a redesign that is this challenging in the past? And what direction did you take for that because yeah this is it. We are gonna start implementing this soon, and my gut feeling say is that we're just gonna have to re redesign the whole thing, and this is a new feature, so I feel like I'm, coming from a place where I want to start with. You know, following the best approach for dashboarding with Caitlyn.

E

That being said,.

E

B

Have a follow-up question for this, so I'm of course, I'm not very aware of this type of context, but from what you told I do get some idea, I'm, just my understanding for borders. People want to see what makes sense for them and that changes from role to role. So are you giving this freedom to the users here? Maybe reprioritize the tiles or this option.

E

No not for now so the the MVC. The main goal is to just display the data on the interface, so we already validated which data points to show. So that's a five less releases in a project, five less field and pass pipelines, total count of releases number of becoming releases and I think the comparison of call me versus actual released pythons. But I discussed this with Nadia last week. That I feel that, as a as a group like in in release management, we could be spending more time. Looking at yeah this.

E

How did you just call the flexibility that we want to give users, because that's gonna be the real power of the dashboard and not necessarily what data points to add in the UI, but to answer a question now we're not the giving any flexibility right now.

A

What do you people think about the consistency question as well like we have so many components places like if even if we would be knowing, which probably would want to build, think what components should be used for this? Making sure that we don't have will not have to change everything again and like six months of time, I.

F

Also have that struggle become muscle, building pores right now.

F

F

Mean perhaps he didn't get I honest question on well boy? Do you feel that you're gonna build something there like it's gonna, be like wildly ready sign eventually, like your concern, is just like a consistency like a flow like a minor details in the consistency that are gonna change.

F

E

It's both it's, it's not necessary when I mention components. Let me like the pajamas component, but the.

E

How do you call that the the overall layout of the dashboard, because, when I look at the dashboards right now, I get lab I'm, not sure the the most recent ones, the ones for security, but I? Look at them and I'm not sure if there was any design thinking behind those dashboards. They really just date on the UI and I feel that if we just replicate that, because we probably Jackie is investigating that, but we probably don't have any, we probably don't have much data on usage or adoption.

E

That's my where I'm coming from I'm afraid, just like the the environments dashboard that we're just creating that one thing just for the sakes of having a dashboard yeah.

B

E

It's not usable and that's and that's where we are not. We are going to spend time now we're scheduling some time to do that, but to look at the direction of the dashboards.

F

To take into account like consumption, information and usability and everything like I wouldn't worry too much about it, because if anything, what you're doing it's gonna become the standard like I actually was gonna. Tell you later. You can share those designs with me because they look pretty good. Well, I'll, probably emulate those styles and that layout that you have there so I mean my take is like you should.

F

Why do the best and that, like I, think everyone will follow so one example is that I think I will like replicate that type of layout or you're using there.

F

Maybe that's yeah. Maybe it's because we're in the same team and we have this ability. Someone else is gonna build a different dashboard and they're like they're gonna, be see. You're gonna be like Oh way. That looks way different right, yeah, I, think.

E

That's what we see like from the past the lack of alignment- it's why we have so many ways to visualize metrics across the product like group, live oysters, level, project level whatever, but it's definitely one of my goals is to make this proposal, but even before put in front of users align internally and see what direction other designers and other teams have for dashboarding and I linked here. Just FYI Erland here this other issue that it's for identifying CIA components for the dashboard.

E

So looking at the metrics, that's we want to display what get love, UI components that exist today we can reuse. So when I mention like this granular effort of building hope, we don't want to be components we just want to use whatever is there to display something UI and I? Think that would be yeah? That's the low-hanging fruit, but I still feel that yeah. That far is a very dangerous words in tables when you have to build them.

E

They're always gonna give you a headache and I in general, I'm saying other things, because I feel, like I, still have a lot of questions that I need to clarify about how to better align the direction of these dashboards with you know the existing efforts that are there being driven by other teams and I agree with you on missing, like yeah. If you design and people use, that's, what's gonna kind of drive, yeah.

F

Well, I was, even if you like, the design doesn't have these so I think we should start there.

F

Like General Thomas is missing and we can actually talk with Tory about it is like layouts like they don't have any kind of some layout.

D

F

Gonna build a dashboard, we were like presenting information like you know, like you, have food like things.

A

On you mentioned Tory and I'm curious, kinda, GGO, like just I, don't know like, may have made this problem known to foundation teams you know like because it sounds you're being faced, was like slightly bigger problem that you know like in one stage group of course, and it's like we have to for sure, make people aware of that, and maybe even go back with Jackie to the wall, drawing board and say hey. You know like don't to me so. First, the other issues, I love at least like have an alignment.

A

There I understand that that she probably has this but yeah.

E

We align yesterday on that and we're gonna do as a mentioned before. Let's just put this thing life and then, let's start an f4 to idh, and then you know, do the design, because we have the same challenge with the environments dashboards where we need. We have bugs and enhancements, but I I know that dashboard doesn't work like how are you gonna paginate, a dashboard that has cards within a group, and then you have to like.

E

Imagine that you have a project with I, don't know mm environments and then you can add a hundred projects to a dash right. So the information it's just, not scalable, so the more I work with dashboards and I would I feel that, let's just do whatever we have to do, but then across get lab and that's something yes, definitely to align with 2x foundations, but also with the designers that independently are working on words to maybe identify the common low hanging fruits. Are we working on layouts now and delivering guidelines for pyjamas and then building?

E

You know one, for example the CIC dashboard as the vision and then once it's implemented. Teams can reuse that to redesign our own dashboard because we're not for that we're all doing different things and yeah summarize we're gonna do just deliver just do it and I'm starting prototyping a version to put in front of users and then I work with with Laurie I got with the question thing is I always forget the name. It seems.

C

So contradictory to me, to some extent that we know and like we're, just gonna push it out there we're just gonna work with the existing components and we're gonna do minimal possible, but on the other hand, we don't want to rebuild our interfaces in six months to a year time. We want to have an alignment and consistency across stage groups and we want to have components which basically solve those two problems.

C

If we iterate on those components, however, none of the stage groups are willing to invest in creating those States groups and we all put it in the fulfillment team or there. You said the team. Sorry yeah.

B

That foundation.

C

Team so like like we're trying to see if there's an alternative around building components, but to be honest, I don't think there is I, wouldn't.

A

Call it alternative to building components. I would say that we are meeting a plan alignment discussion, you know and on one hand, I understand like again. We all I think it's always a compromise between business users and like and the company goals right yeah. Maybe, for example, if there is an urgency or urge and Jackie wants to put something out there and in the meantime, to start collecting feedback. I agree that it's not the perfect solution in every case, but it depends how much effort but yeah.

A

It's it's a tough question, but I would understand why we want to get something out quickly there and then just use the time that we need to make a plan and iterate it's a tough one. But but it definitely I think we're meeting a plan and alignment here, because.

E

When I go I'm, not sure if you folks ever attending the analytics call like for the pajamas component analytics, we are constantly like bringing this the new items or future things from other teams. So, for example, this analytics dashboard. So they are changing it and they are giving the user their flexibility I. Think it's actually in one of the back showcase lights to.

E

Customize, what dashboards or what data you can see from the it lab sly and I, keep wondering why are we not just Beauty on top of that like? Why? Are we not allowing users to just get put some data in there get lab in the CI file and then render the data this dashboard?

E

In the same view like analytics, and then you pull release insights, so we build on top of what the analytics team is doing right rather than building our own dashboard and I agree with an idea that sometimes there's a lot of independent in most of the time, most a lot of independent initiatives and then once everything's out there like I, had yesterday or today with a new button to assign to a user once it's out there, people see and they were like what is this. This could be better.

E

This should be different, so I do agree that we need more alignment.

E

But that takes time and this there is a research inside from two years ago saying that users need a group view dashboard, so I, understand and I agree that we should just give it to them and if they are using it or at least if there is some adoption, I think that's what's gonna feel you know the argument of: let's do a redesign or let's invest in more validation or more prototyping, and then maybe other teams can can also can also get inside feedback from user interface from usability etc.

E

But then it's a bit of a it's a struggle.

E

Very sad topic, I know.

D

Everything it's not I didn't want to talk about it. It's just no.

A

But I think it's a great case for also for future to learn so yeah. It sounds like the next actions for you, hi Anna is to work with Jackie to see what is that, and we see that you can release, and then you will be working with other teens in gitlab to find to align, to find well to build, maybe the strategy.

A

There would be very interesting to observe this case and see if we could help or if any feedback or something could be done from our side, because this is a great example of things that probably will be coming out from time to time with different features. So we have.

E

I have Madea and also Nadia Nadia she's, also working in dashboarding and another takes Nick and Andy. So these are people that I have to get in touch with then I know so far. You see I'll probably come back with this in the future. You show the results.

A

Awesome thanks a lot for bringing this up I'm really interesting discussion, so it was good timing as well. Okay, this is everything from the agenda points. Doesn't he but he has anything to add to ask otherwise you can go and get ready for the your conversations be an exciting session. See you all there in 30 minutes, thanks so much for presenting thanks for awful feedback.