In my recent blog post about the one-time-setup and recurring-costs of an office, I mostly focused on financial costs, human distraction costs, and the cost of increased barriers to hiring. This post talks about another important scenario: when your physical office limits potential company revenue.
Pivigo.com is a company in London, England, which connects companies that need help with data science problems with Ph.D data science graduates who are leaving academia looking for real-world problems to solve. This 2.5 year old company was founded by Dr Kim Nilsson, (ex-astronomer and MBA!), and as of today employs 4 people.
For Pivigo to be viable, Kim needed:
- a pipeline of companies looking for help with their real-world Data Science problems. No shortage there.
- a pipeline of Ph.D graduates looking for their “first non-academic” project. No shortage there.
- a carefully curated staff of people who understand both Academic and Commercial worlds are essential to help keep things on track, and make sure the event is a success for everyone. Kim has been quietly, diligently working on growing a world-class team at Pivigo for years. Tricky, but Pivigo’s hiring has been going great – although they are always interested to meet outstanding people!
- a physical place where everyone could meet and work together.
Physical space turned out to be the biggest barrier to Pivigo’s growth and was also the root cause of some organizational problems:
1) Venue: The venue Pivigo had guaranteed access to could only be used once a year, so they could only do one “event” each year. Alternate venues they could find were unworkable because of financial costs, or the commute logistics in London. Given they could only have one course per year, it was in Pivigo’s interest to have these classes be as large as possible. However, because of the importance of creating a strong network bonding between the participants, physical size of venue, and limits on skilled human staffing, the biggest they could do was ~80 people in this once-a-year event. These limits on the once-a-year event puts a financial cap on the company’s potential revenue.
2) Staffing: These big once-a-year events were super-disruptive event to all the staff at Pivigo. Between the courses, there was administrative work to do – planning materials, interviewing candidates and companies, arranging venue and hotel logistics, etc. However, the “peak load time” during the course clearly outscaled the “low load time” in between courses. Hiring for the “peak load times” of the courses meant that there would be a lot of expensive “low load / idle time” between each peak. The situation is very similar to building capacity in fixed cost physical data centers compared to AWS variable-by-demand costs. To add to the complexity, finding and hiring people with these very specialised skills took a long time, so it was simply not practical to “hire by the hour/day” a la gig-economy. Smoothing out the peaks-and-troughs of human workload was essential for Pivigo’s growth and sustainability. If they could hold courses more frequently, they could hold smaller, more frequent, courses and reduce the “peak load” spike. Also, changing to a faster cadence of smaller spikes would make Pivigo operationally more sustainable and scalable.
3) Revenue: Relying on one big event each year gives a big spike of revenue, which the company then slowly spends out over the year – until the next big event. Each and every event has to be successful, in order for the company to survive the next year. This makes each event a high-risk event for the company. This financial unpredictability limits company long term planning and hiring. Changing to smaller, more frequent, courses make Pivigo’s financial revenue stream healthier, safer and more predictable.
4) Pipeline of applicants: Interested candidates and companies had a once-a-year chance to apply. If they missed the deadline or were turned away because the class was already full, they had to wait an entire year for the next course. Obviously, many did not wait – waiting a year is simply too long. Holding these courses more frequently make it more likely that candidates – and companies – might wait until the next course. Finding a way to increase the cadence of these courses would improve the pipeline for Pivigo.
If Pivigo could find a way to hold these courses more frequently, instead of just once-a-year, then they could accelerate growth of their company. To do this, they had to fix the bottleneck caused by the physical location.
Three weeks ago, Pivigo completed their first ever fully-distributed “virtual” course. It used no physical venue. And it was a resounding success. Just like the typical “in-person” events, teams formed and bonded, good work was done, and complex problems were solved. Pivigo staff, course participants and project sponsors were all happy. Just like usual.
This maps shows everyone’s physical location.
To make this first-ever fully-distributed “virtual” S2DS event successful, we focused on some ideas outlined in my previous presentations here, here and also in my book “Distributed Teams”. Some things I specifically thought were worth highlighting:
1) Keep tools simple Helping people focus on the job-at-hand required removing unnecessary and complex tools. The simpler the tools better. We used zoom, slack and email. After all, people were here to work together on a real-world data science problem, not to learn how to use complex tools.
2) Very crisply organized human processes. None of these people were seasoned “remoties”, so this was all new to all of them. They first met as part of this course. They had to learn how to work together as a team, professionally and as social humans, at the same time as they worked on their project which had to be completed by a fixed deadline.
3) As this was Pivigo’s first time doing this, Kim made a smart decision to explicitly limit size, so there were only 15 people. This gave Kim, Jason and the rest of the staff extra time and space to carefully check in with each of the remote participants and gave everyone best chance of success. Future events will experiment with cohort sizes.
4) Each participant said that they only applied because they could attend “remotely” – even though *none* of them had prior experience working remotely like this. Pivigo were able to interview, and recruit participants who would normally not even apply for the London-based event. The most common reason I heard for not being able to travel to London was disruption to parents with new children – successful applicants worked from their homes on real-world problems, while still being able to take care of their family. The cost of travel to/from England, and the cost of living in London were also mentioned. The need and demand was clearly there. As was their willingness to try something they’d never done before.
5) I note the diversity impact of this new approach. This cohort had a ratio of 26% female / 74% male, while prior in-person S2DS classes typically had a ratio of 35% female / 65% male. This is only one data point, so we’ll watch this with the next S2DS event, and see if there is a trend.
The Virtual S2DS programme was a success. The project outcomes were of similar quality to the campus based events, the participants felt they got a great experience that will help their careers going forward, and, most importantly, the group bonded more strongly than was expected. In a post-event survey, the participants said they would reach out to each other in the future if they had a question or a problem that the network could help with. Interestingly, several of them also expressed an interest in continuing remote working, something they had not considered before.
For Kim and the Pivigo team, this newly-learned ability to hold fully distributed events is game-changing stuff. Physical space is no longer a limiting factor. Now, they can hold more frequent, smaller courses – smoothing down the peaks and troughs of “load”, while also improving the pipelines by making their schedule more timely for applicants. Pivigo are investigating if they could even arrange to run some of these courses concurrently, which would be even more exciting – stay tuned.
Congratulations to Kim and the rest of Pivigo staff. And a big thank you to Adrienne, Aldo, Christine, Prakash, Nina, Lauren, Gordon, Lee, Christien, Rogelio, Sergio, Tiziana, Felipe, Fabio and Mark for quietly helping prove that this approach worked just fine.
John & Kim.
ps: Pivigo are now accepting applications for their next “virtual” event and their next inperson event. If you are an M.Sc./Ph.D. graduate, with a good internet connection, and looking for your first real-world project, apply here: http://www.s2ds.org/. Companies looking for help with data science problems can get in touch with Kim and the rest of the Pivigo team at firstname.lastname@example.org.