COVID-19 lockdown impact on remote working at Clarity AI
NNIt’s been a while since we first heard about COVID-19. After 3 months of severe lockdown measures in Spain, we are living in what has been termed the “new normality”. Does it mean we are back to the office? No, it doesn’t! Following Spanish government recommendations, we continue working fully remotely.
The road to COVID-19 lockdown
During February (maybe before), Coronavirus started to spread across Spain fast and silently. After a few weeks, the Spanish government declared a state of alarm on 15th March. For 98 days, the whole country was locked down, with serious mobility restrictions, probably the harshest in the world.
On March 15th, many Spanish companies were forced to implement remote work policies overnight, with no time to prepare at all.
Remote working before COVID-19
Remote working has been part of the DNA of Clarity AI since its foundation with everyone being able to work from home. Even though the company had a fancy office in Madrid’s downtown, working from home was a common practice within every team from developers to designers, data scientists, researchers, business developers, etc…
While most of the people used to spend, at least, one or more days in the office per week, I have been working fully remotely for the last 4 years, including the last year and a half since I joined Clarity AI. But I wasn’t the exception, other team-mates have also been working 100% remotely. In addition, others were travelling frequently, working from wherever they were.
Were we ready to work fully remote?
Obviously no, we were not. No one was, but I would say we were in a good position, much better than most companies, including those in tech. Some big tech companies have been struggling during this lockdown, and they continue doing so, simply because their employees didn’t have the possibility to work with a laptop or they didn’t have a proper VPN to work from places other than the office. They never considered remote work as an option and now they’re paying the price.
I was very happy with the way we used to work before COVID-19. We had all the necessary tools to manage our work online as well as the equipment needed to work remotely. Of course, we were not perfect, having some areas to improve like communication or flexible working hours.
Additional problems with working remotely during a lockdown
In a normal situation, remote workers have to struggle with problems like being able to unplug after work, loneliness, communication issues or staying motivated and focused. But as a consequence of the lockdown, some extra problems appeared:
- Network saturation: Internet traffic increased dramatically after the lockdown. It made it more difficult to connect to online calls and meetings.
- Working with families: as schools were closed, team mates had to deal with their children being home with them. It forced parents to adapt their daily working hours to the daily needs of their kids. Also, background noise during meetings was frequent.
- Not having a proper place to work: yes, we might have had a proper desk, chair and equipment but when you live with others you may have to share the room, or even the desk you’re working on. Having a call with somebody else talking in the background is not the best scenario.
- Being focused wasn’t easy: during lockdown, there were a thousand more things that made your level of focus decrease. For example: having a family member infected, being worried about the situation, being eager to know more about the virus or just being stressed because you can’t see your family or disconnect by having a beer with your friends.
- No more on-site communication: direct communication in the office is, by default, more effective. Even though we run most of our meetings online, communication in-person in the office has always been a fast way to solve doubts or making quick decisions.
- No more face-to-face team building events: we do several events every month to spread our knowledge with the team, to create a nice working environment and, why not, have some fun together. For example, every Thursday, our co-working office building would have a party with free drinks and music for everyone.
There are always going to be things that are hard deal with like background noise, distractions or network issues, but we can definitely face the challenge head on and improve our working practices in areas like communication, flexible working hours and team building.
Improving our remote work practices
Having flexible working hours is one thing, but having to work and take care of a family with kids at home who have to do their schoolwork online is another. Improving our flexibility was key for our team, and we did so by taking some actions:
- Conversations on shared documents. Trying to foster async communication, we started to work more frequently with online shared documents. For example, we started to write Technical Design Documents, which has been a very helpful practice for us.
- Use of the availability calendar to block the hours you expect to be not available. When setting a meeting, we tried to be flexible on the times and adapt to others’ availability.
- Don’t expect immediate responses on Slack. Your team mate might not be available at that moment.
As you can see, the actions we took were closely related with communication. Even though, remote working was a common practice at Clarity AI before the lockdown, communication was an area where we saw a need to improve. As a lot of people were working at the office everyday, “corridor conversations” were a common practice, but with the lockdown, those conversations moved to Slack, shared documents or other public channels. That effect made us more transparent as a team. Now, we are all more aware of the decisions taken and the reasons behind each of them.
One of the things that makes working at Clarity AI a nice experience is the team atmosphere. We continue doing activities to maintain that good atmosphere like the biweekly Brown Bag Meetings. But also, we started new initiatives to keep us engaged and bring us together:
- Virtual coffee machine: a 24/7 available online call where you can enter while having a coffee and “meet” others like you’d do at the office.
- Virtual beers on Thursdays: online call to drink a beer with other team-mates as it was a common practice at the office.
In my opinion, the virtual world is not as good as the real world, but… welcome to the “new normality”.
Data comparison before and after the lockdown
With the COVID-19 crisis, data has become part of our lives, with tabloids and TV news displaying fancy charts and countries reporting data every day. Have you heard about “the curve”?
At Clarity AI, we love data. We use data to provide ESG scores but also, within our product & tech team, we use data to monitor our performance as a team.
In order to make a fair comparison, I’m going to compare our data from December to July. That’s approximately 3.5 months before and after lockdown started, 7 months of data in total.
We are going to compare data from two systems. First, we are going to try to see how our communication has changed by checking stats from Slack, our internal messaging tool. Second, we are going to compare some agile metrics extracted from JIRA by our custom agile metrics system. We are currently working with 3 JIRA issue types which have been named A, B and C for this comparison. (Check this article if you want to know more about our metrics system.)
Data accuracy warning
If we have learnt something about data during the COVID-19 pandemic it’s that you need to be very careful when analysing data and also you definitely have to trust your source of data to have good conclusions.
Therefore, we need to take into consideration that we are a living team. During this period the team has been constantly changing: people have left the company, people have joined the company, some went on vacations, some worked (voluntarily) on bank holidays… In the case of Slack messaging data, it includes some noise caused by our bots publishing automatic messages on Slack. Also, even though our metrics system extracts information automatically from JIRA, there might be some errors in the data due to wrong JIRA issue labelling or wrong issue movements across our Kanban board.
Slack messaging data
As “corridor conversations” are not possible anymore, most of those conversations have moved into Slack. That means that we have almost doubled our daily posted messages after lockdown.
Some more details about our daily posted messages after lockdown:
- Direct messages: +188%
- Private channel messages: +186%
- Public channel messages: +206%
- Total daily posted messages: +190%
I think it’s remarkable that public channel messages have increased with a higher ratio than other types of communication channels.
Agile metrics: Throughput
We could expect that the lockdown negatively affected our performance but, in contrast, our throughput rate has been increasing month after month.
If we look at our story points throughput data, we can also say that we maintained or slightly increased the number of story points we complete every month.
Therefore, it looks like we dealt quite well with the lockdown and our performance didn’t decrease.
Agile metrics: cycle times
Cycle time is one of the most important metrics in the Agile ecosystem. We can use it to provide estimations as well as to find issues in our system. So let’s check the cycle times evolution during the last 7 months for our main 3 issue types.
Looking at the charts above, I would say we kept a similar trend before and after lockdown with the exception of April when issue types B and C took longer than the 85th percentile. In the case of issues type B, our lead time during April was 7.2 days while our 85th percentile stands at 5 days. Our lead time for issues type C during April was 12 days while 85th percentile is 8 days. Well, that’s an increase. Fortunately, it happened only during one month and our lead times went back to “normal”.
Therefore, can we say this was an effect of the lockdown? I would say no, but let’s check our blocked issues data and see if we can find some other to explain this increment.
Agile metrics: blocked issues
For sure one of the biggest concerns for a company working remotely is increasing the issue blocking ratio as a consequence of communication difficulties. Let’s check how many issues we blocked every month and compare before and after COVID-19 lockdown.
As you can see, data is very variable in these charts. Anyway, I can highlight the case of April for issues type B. Remember that our lead time increased during April for these issues? Well, this could be the reason.
In any case, we are interested not only in the number of issues blocked every month but also in how long they were blocked. Find below charts representing how many days on average our issues are blocked by board column:
Again, we can see data is not stable at all. In any case, if we look closely at issue types B and C charts, we can see that the average blocking duration for April was high in both cases. In conclusion, lead times increases for issue types B and C during April were caused most probably because of blocking issues. Now the question is: are those blocking increments a consequence of the lockdown? No, they weren’t. We monitor our data in detail with every team every month. As part of that process, we analyse tickets taking longer than the 85th percentile (outliers) as well as every issue blocked during last month. Our conclusion was that root causes for blocked issues were related to our business and other things not related with the lockdown.
In closing
COVID-19 is here to stay for quite awhile longer. As you might have heard, we need to learn and adapt to live with this pandemic that came into our lives like an earthquake. Thankfully, we live in the XXI century where we have the tools to adapt our work to a new remote way. A remote way of working that for most people was imposed from one day to the other and that, probably, is here to stay with us forever. Do you continue thinking that remote work is not possible?
