The world’s leading publication for data science, AI, and ML professionals.

🥱 Water cooler talk: How much do you really work?

Analysis and visualisation of 850 days of work, learn, and sleep data. Which location makes me more productive?

Photo by Priscilla Du Preez on Unsplash.
Photo by Priscilla Du Preez on Unsplash.

How long do you work each day? By work, I don’t mean talking by the water cooler, taking 15 minutes to make a cup of coffee, or replying to all your Whatsapp messages on (or just in) the toilet. Also not the casual popping-over to Twitter, or writing "just one" personal email with 100 details about your last attempt to make kombucha.

I mean the real work time.

One way to measure it, is to track and categorise how you spend time on your laptop. It would allow you to see clearly what you actually do, if there are any interesting trends, or correlations between work time and mood or sleep.

Guess what. One self-tracking-obsessed, aspiring Data analyst has done it! 🎉 And you’re about to learn what she discovered.


The data in this post comes from my daily and weekly self-tracking forms which I’ve been filling in – you guessed it – daily and weekly, for over two years now. These forms track my mood, location, medications I’m taking, habits and I use them to log several journal-style thoughts on various prompts. Sleep data comes from Oura Ring, which I’ve also been wearing for over two years.

A shorter version of this post also exists as a Tableau-story.

Most hard working city

I used to travel a lot, so the first thing that was interesting to investigate in this data set were the cities in which I worked most.

The ranking is as follows:

  1. Lisbon, Portugal 🇵🇹 (31.25 hrs)
  2. Caen, France 🇫🇷 (19.44 hrs)
  3. Bali, Indonesia 🇮🇩 (17.86 hrs)

But, look at the values. Lisbon wins over Caen by a lot.

What happened there? you might wonder.

It’s pretty simple. I was without my partner and, as a rather unsociable person, had nothing much to do.

The second thing you might wonder is, "gee the other averages are just 17–9 hrs a week?, you don’t seem to work a lot!"

But, remember the first paragraph of this text?⬆️ Well, I believe that many so-called 40h/week office jobs are, at most, 25-hour gigs. So, my work time might not be so different from yours… I challenge you to start tracking it – get ready to be surprised.

Don’t get me wrong, it’s not a comment on anyone’s laziness, but rather on the inaccuracies of our perceptions. It can seem you’re "working" 40 hours, if you spend that much time in the office.

It’s also a comment on, what I consider, faulty work standards.

If you can get the same amount of things done in 10 hours as someone else does in 30, should you be required to sit in the office for the remaining 20 hours?

I don’t think so. And, you should probably both be paid the same, based on the results not on the time spent.

Short work time? Second part of the story.

Noticed that I often post about data? Well, I’m re-training to become a data analyst. It involves a lot of learning, which is what I spend a lot of my time on.

Below is a chart showing the combined values of work and learn time, by city.

Well, the "winners" are the same as in the previous case: Lisbon, Caen, and Bali. But, the numbers are much higher now. (Still no water-cooler talk included, and the timer is switched to tracking "family time" when I write to my mum about any kombucha attempts.)

☝️ These numbers should serve as an update of the glorious image of a digital nomad – working a few hours a week, by a pool in Bali. Nope.

The real image looks rather like this: use the low costs of life as a chance to invest as much time as possible into learning, while trying to find an air-conditioned room and a decent chair to sit on in, but not paying a fortune for it. You don’t have a fortune, you only work 15h a week, remember?

I found a good chair, but someone was faster to get it. Hustler Villa, Ubud, Bali (Indonesia).
I found a good chair, but someone was faster to get it. Hustler Villa, Ubud, Bali (Indonesia).

Is there a work city and a learn city?

Looking at the work and learn time separately we can figure out in which city more time was spent on which activity.

Looks like Lisbon totally butchered the competition as it comes to work time, and in the difference between the time spent on work and on learning.

On the other side of the chart lies Warsaw, where I spent much more time on learning than on work. And, boy do I remember that time, it was the period of some heavy SQL queries, and trying to apply for two jobs with very demanding selection processes.

Needless to add, their SQL SELECT didn’t return me. Which is why you can still enjoy my posts. Otherwise I’d be making money and a big positive impact somewhere else now. 😛

Time trends?

Looking at the trend lines, it appears that overall I’m spending less time on work and more on learning.

How can this be possible? Now it’s me who wonders.

One explanation can be that with time my experience grows, and so does my ability to charge more as a freelancer. In other words, I earn more per hour, and hence can work less for the same financial result. (A still low and miserable one, if I may take a moment to self-pity myself.)

How about grouping these numbers by quarter?

When grouped by quarter, the last two years tell a tale of early-year money-making motivation, which turns into a focus on the long-term gains of self-improvement as the year progresses. 💰 🥳🌹

Ok, this is a gross over-interpretation.

The only thing it shows is that in the last two years I spent more time on work than on learning in the first half a year. I’m sure you – one of the two people who read this – are at the edge of your seat regarding what the results will be for Q1 and Q2 of 2021.

Higher sleep score, not more work 🤷‍♀️

Lastly, I was curious to see whether sleep had any impact on my work time. My sleep data is tracked with an Oura ring which, every night, calculates a sleep score based on the quality of my night. The higher the score, the better my sleep.

Here is a scatter plot with a trend line measuring correlations.

Before plotting this chart I’d have assumed that a higher quality of sleep, would have increased my motivation and focus and thus lead me to spend more time on work.

But, in a surprising turn of events, it appears to be the reverse!

The higher the sleep score the less time I spent on work.

Why? Perhaps better sleep gives me time to reflect on my priorities, realise that work is a waste of life, and that I should rather go and appreciate every fallen twig I cast my gaze upon. Or, with better quality sleep I get more focused and do all that’s needed at work in less time. Or, a 1000 other explanations.

Importantly, other charts showed that the decreased time spent on work, didn’t coincide with a significant increase of time spent on learning. There was a small increase, but nothing statistically significant.

Next steps

It’d be valuable to measure not just the length, but the quality of work and learning. It would help quantify whether a decrease in work time means lower or greater Productivity, and be more precise about the impact of sleep on these metrics.

If you’re interested in data stories check out these ones:


Related Articles