GitLab UX Scorecard, 19 Aug 2021

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: UX Scorecard Creating an On-Call Schedule

Description

Issue: https://gitlab.com/gitlab-org/gitlab/-/issues/337237

Slides: https://docs.google.com/presentation/d/1jmSpryH9dRbTrGxsd8AeFZy6CLdZ8GZ3MFq_JFTQHDs/edit?usp=sharing

Journey map: https://app.mural.co/t/gitlab2474/m/gitlab2474/1628838224425/7d69f84651be4ef52001d9cea688fc7052f591bf?sender=kcomoli8464

UX Heuristics: https://about.gitlab.com/handbook/engineering/ux/heuristics/

A

Designer and growth- and this is a ux scorecard on creating a non-goal schedule and escalation policies.

A

um So if you're not familiar with the us call card process, it's basically a heuristic process when we start by identifying the job to be done for the user and what's their desired outcome, map out a user's journey. So what is the current process to get to the desired outcome?

A

And we end up by grading the experience based on ux heuristics and a grading model that we have so who are the personas and jobs to be done for creating an on-call schedule so personalize our devops engineers, software developer and system administrator, and we have really two jobs done here, um so one for the on-call schedule and one for escalation policies, but they really work together so for on-call schedule.

A

The first one is: I want to set up on-call schedule to handle my alerts, so I can be confident that if another is firing, the appropriate people will be paged.

A

The second one for this collection policy is, I want to set up this collection policy for my uncle schedule, so I can be confident that even if the first person page to review isn't available, someone else will be notified.

A

um So the flows I just do like broke down this core card into two key flows. um So the main one is creating an on-call schedule and escalation policy as a maintainer in a paid group or project there is in florida.

A

Is that because on-call schedules and investigation policies are both a paid feature and part of the premium plan, and also to be able to use those features, you need to be the maintainer of your group or project, uh but I also choose to look at it from the angle of uh someone that doesn't have the maintenance permission and the reason for that is because, when we see stage adoption and feature adoption in gitlab, it's mainly driven by invited users and these users may not have the proper level of permission uh to access the feature.

A

Therefore, you kind of raise the question of like how can they uh perform the task, um but the main aspect here is from the maintainer perspective. So let's check this out, so I have here one of my test project that is on the trial, so I can access all features.

A

So the first thing I would do is go through the navigation and most likely monitor.

A

This is where I would expect to find it, and this is where I am finding it, so I can see on-call schedule and escalation policies, so I'm going to start by with on-call scheduled, and here I land on a nice empty state.

A

That's that's created on-call schedule in gitlab, rogue elders directly to specific members of your team at a schedule. um I really like this entire step because it's straightforward, it's really getting to the point and the call to action is really engaging.

A

The only thing that I would have expected from a feature perspective is maybe to have the ability to import a schedule from a third party, uh because I imagined uh maybe my team is using uh gitlab as a tool, but maybe you already have like on our or on-call schedule, set up somewhere else, uh most likely in google calendar or maybe in an excel sheet, and I would have liked to have the option to just import all that and not recreate everything from scratch.

A

That would have definitely like pushed me to adopt this even more, um but for the sake of this core card, I'm just going to create it from scratch.

A

So I'm clicking on add schedule and um asked to name the schedule, give it a description and pick a time zone. I, like the time zone, touch. I think it's really nice to have um so yeah. So for the name. I honestly have no specific idea on how to call this, and I kind of wish that um I had some kind of suggestion, um so maybe, as a helper, we kind of like here we have time zone. We would say said that no okay, I got in there.

A

I set the default time zone for the schedule for a participant like I wish. I would have something for a name like a suggestion uh about how to name my schedule um and yeah and clicking away directly fired and alerts. um So that's that's. Okay. uh I mean it's.

A

It's kind of pushing you to make sure that you filled everything before submitting, um but I don't know I mean in some perspective you might want to let people fill everything submit and then tell them when they have errors, um but anyway yeah I'm gonna call the schedule. I don't know uh tag on calls no description um and then the time zone is gonna, be yeah, amsterdam, time zone, um so I'm gonna add the schedule.

A

Okay. So here I have my schedule created and I have a nice tip alert here: uh try adding a rotation, so it's basically pushing me towards the next action. um So my skill has been successfully created. Now I need to add people to schedule, um but then also saying uh I need to enable notification for the schedule by creating an escalation policy.

A

um Okay, so that makes sense. um Okay, so let's let's try it out um yeah. So I see the add rotation out of rotation here. I just I don't see a primary cta in this page like we have at the schedule, but maybe add a rotation should be a primary. I I don't know it's just it's not directly clear. What is the next step, so, thankfully, the tip alert is here to tell me the process.

A

um Okay, I'm going to add a rotation, so I have to add a name say: givens rotation, and this one uh or no, maybe maybe this one- the rotation length is going to be one day. Please no rotation with shifts there are less than four hours or currently knows very end. With your view,.

A

Okay, oh yeah, okay, we'd, like you um say it starts on, say tomorrow at nine. uh Yes, I'm going to enable an end date, I'm going to say the same day.

A

Rotation and daytime must come after start time, yeah yeah. Alright, that makes sense. I just I just wish that this would have somewhat pushed the hour. I said here from like one more so that I don't get the other the the alert here um I mean it's nice to have it obviously, but I just wish that it would kind of take into account that it's not specifically me that triggered it in some way so to have it.

A

For example, this start at 10., um but yeah, I'm gonna say that I'm on call until 12 uh restricted time intervals, so yeah. So that's that's why I'm a bit confused between what's the difference between starting on having the end date and then the time interval? I just wish that we had a little bit of like texts around here to explain the the difference um between all these.

A

So anyway, I'm going to try it for this station on google be yeah so from 9 to 12..

A

That's a bit redundant, I don't know if I'm doing it well, actually, I'm going to add this rotation all right. So I can see my rotation here.

A

I can see who's doing the rotation starts and ends. You can see the time and time zone from day view.

A

Yeah next day, yeah, I'm here cool and then I succeed yeah I get the success message. um It's great.

A

The only thing is that it's removing the edit, the tip, alert. Sorry um so now I kind of remember that if I want this person to be notified, I have to go to create an escalation policy, but because dealer told me- and if I were to do this in two different days- I don't know if I would have remembered that this is where I needed to go. So I wish that the alert that was here previously would still be there or would be replaced by some kind of text so that it doesn't get lost.

A

Basically, um but okay, I'm gonna create my escalation policy, okay and then yeah so same thing. I think team two state is great here. What I just feel a bit disturbing in a way is that I have to go to another page uh to do this because it seems like uncle schedule and collection policies are tied together and you cannot use one without the other. So I don't know why it's a separate page.

A

I would have maybe preferred to have it directly in the schedule page or directly, even in this schedule, but anyway, I'm gonna go through this page.

A

Creating education policy in gate lab, set up description position for his page and when and even the first user's page, don't respond. This is exactly what I'm looking for.

A

The only thing that I wish I would specify is like how these person get notified um for now. I feel like this is not clear. Like I don't know if they receive an email, I don't know if I can add the possibility for them to receive a text as well um so yeah. I wish that would be specified here, but other than that. It's again it's like the first time to say it's, it's pretty straightforward and engaging so yeah. I guess I have to name my escalation policy again. So same thing.

A

I wish I would have some kind of helper around here to give me a hint, because this can be a bit blocking for me to just come up with a name, uh so I don't know how I'm gonna call it actually, um okay, so tests escalation, no just escalation, one, maybe uh no description and escalation rules. um If other is not acknowledging zero minutes, then yeah. I really like this. uh This is really great.

A

This is exactly what I would expect from this type of interface, like having some kind of conditional statement, is really making it clear about as to why. What is the escalation policy so yeah? If the other is not acknowledging.

A

15 minutes um no, this is five minutes.

A

Five minutes, uh then email on call user schedule select the schedule they want calls yeah um yeah, that's that's what I want and then I can add an additional rule as a fallback, um and I think I'm gonna do it, because the job to be done was about uh paging check. Yeah. The chapter of dawn was about painting someone else. um If the first person is not being notified, so creating the fallback, so I'm gonna do it. I'm gonna add an additional rule.

A

um So if the alert is not, let's say resolved in, say 16 minutes um yeah, so yeah again like the deal is firing a bit fast, um it's better to have another than no alert, but I would just prefer that it come after I'm trying to submit um so then I'm gonna email, okay. So this is clear. I know yeah email as well like okay. Now now I'm gonna be sure about that.

A

uh I know that people are gonna, be emailed and I'm gonna add email, another person yeah. That sounds good cool. So it looks like I created my escalation.

A

The thing is, I didn't, have any some kind of success message or reassurance. um I see I'm seeing that this is in the ui. I'm seeing it written that, if dealer is not acknowledged, this person will be picked.

A

If the dealer is not resolved, another person would be paying. I I see that, but I have no real way to verify this and the ui is not going to be assuring me even alert, saying, hey everything went well um rest assured that these people will be paying because right now it feels like the only way for me to verify if this works um is to basically have a problem during an on-call and see if the alerts are triggering are firing.

A

Sorry, so I just wish that there was some kind of way to test um the escalation policies when they're created- or at least some kind of reassuring message to tell you that everything is uh okay and ready to go um so yeah. That's that's the experience for a maintainer in a paid group so using the different heuristics.

A

This experience created as a b minus, which is good, so the realistic is made in almost every way, with no more than minor exceptions. The task is streamlining pleasant and highly usable.

A

It's pretty straightforward honestly.

A

So what I thought was really good in this experience was the empty states, the conditional statements for the escalation policies and the typically, thankfully I had it, because if I was if I was not guided by this alert, I don't know if I would have been able to do it this fast, but in terms of potential improvements uh in terms of the visibility of system status, I think we could apply the pecan rule and reassure people when they create an escalation policies, because this can be a bit stress inducing if I'm not sure that those uh scholarship policies are gonna work.

A

um Then, in terms of setup super we could maybe do a bit better in helping user naming their scheduling policy and specify the already type or the notification type beforehand, because I would have expected to see this in the empty state and not in the conditional statement of the escalation policy. But it's it's still uh mentioned, but a bit late in my opinion, but that's that again, that's not a big deal um and then the alert, the tip on the page.

A

That left was a bit surprising because, like I said, if I do this in during two different days or maybe like one, if I want to set my installation policy one or two weeks after I created my schedule, will I remember that I need to switch page to do it. I don't know.

A

Then we could also improve flexibility and efficiency of use so, like I said, suggest third-party integration to streamline a bit the setup phase and lower down the adoption barrier and try to solve a bit. This context version situation uh to have a scholarship policy in the schedule. Maybe so I don't have to like move uh over different pages and lose context of the schedule uh all right, so this is for this degree for the main job to be done um now from the non-maintainer perspective, I'm gonna do it to have it.

A

I have it here, um so this is the same. Actually, I might need to delete things or not. Maybe it's going to work. This is the exact same group. The only difference is that um switch on a different user. That is a developer of this group and I have the same empty state and I'm going to click on, add a schedule, um so I have the same the exact same model.

A

So I'm going to say I don't know schedule 2, two same thing for the time zone.

A

Add the schedule yeah and then I have an error uh and there is graphql error. The reasons that you're attending to says: there's no exit- or you don't have permission to perform this action. So it's okay, it's kind of clear. So it seems like I don't have the permission to do it. um So at least that's good to know, but I just wish that I would have known that beforehand before starting to add the schedule, um because that I wouldn't have engaged with.

A

If I knew I would have tried to find a way to get that permission first, um and at this point I I don't have like. I don't know too much how to do it besides, trying to check the members information and see who's the maintainer, so this person is the owner, so he would be able to give me the maintainer permission, because right now, I'm just a developer, so we probably reach out to him.

A

I don't know either with slack email or if we're working together in the same office just trying to reach out to this person. um But maybe we should have a way to do that from the empty states. To kind of request access to a specific feature, but as far as like my journey goes as a non-maintainer, this is where it stops, because I cannot do anything. I just need to request permission.

A

um So yeah I I excluded this um flow from the grading experience but, like I said, I think it was important to just check it out, because if I were invited to this project without the proper permission, I would kind of expect to have a way to be able to ask for it um so yeah. So the good thing is like at least I have another telling you what's going on. I can understand that I don't have the good level of permission, um but what could be improved is probably clarifying it earlier.

A

Another thing is: maybe lowering down the permission level. I don't know if that's possible, but I I'm not sure why it is a maintainer level uh permission especially like if we think again back to the personas being software developer. For example, I don't know if all of them are maintainer of the project or if the one that should set the on-call schedule on maintainers, um so maybe the permission should be lower down um and then yeah.

A

Another thing that we could do is proactively help people reach out to the owner and maintain her to get the right level of permission.

A

um So, to conclude, say that overall, the experiences intuitive and straightforward, but it has some context- switching it also could be faster with third-party integration.

A

The setup support is pretty good. For example, this alert that I had the grey alert was really doing a good job at explaining me the different steps um and the form also had a couple of helpers.

A

I just wish that there were more, especially for naming um it can be a bit blocking when you don't know how to name things so having some suggestion would be definitely helpful uh to move uh forward kind of faster and it would be also easier and then we're also missing the peaking rule so kind of celebrating, but also reassuring the user when they finished to set up escalation policies.

A

I think this would be really helpful to really nail the confident part of the job to be done, because it can be a bit.

A

Friction inducing, I would say, like I might be a bit worried as to uh is this going to work or not. um At this point I feel like I should be assertive that this is going to work moving forward, so I mentioned a way to test the escalation policy, but that's a some kind of a big feature to add. So maybe a simple step could be to just have a successful message when the user completes the escalation policy, so that's it for this scorecard. Thank you for watching.