Reopening the analytics restaurant

Here are three things we’ve learned about food during lockdown:

1. Cooking for yourself and your family can be fun.
2. Doing this every day for four months: not fun.
3. Dining out, which used to be a pleasant way to spend a few hours with loved ones, has become a sprint through a radioactive minefield toward a destination that may have disappeared by the time you arrive.

And here are three things we’ve learned from the Great Unplanned Remote Work Experiment:

1. Analytics work, as a species of knowledge work, can be done live or remotely.
2. Except for extreme introverts and those who have been working in a faraway village all along, remote work kind of sucks.
3. If you disagree with #2, consider the possibility that the downsides of unplanned all-remote work are accumulating within your team as a hidden debt, for which you will pay the price later.

As offices and schools begin to offer their game-theoretic announcements of whether and how they will return to live interaction this fall, Chief Analytics Officers (CAOs) are among the leaders who must play the game.

Picture an analytics team that has historically sat together in one office, with no more than a few team members based in faraway villages. The team’s weekly activities include building and maintaining models, running ad hoc analyses, doing exploratory analytics, giving requirements to IT, reviewing the slate of ongoing analytics projects, making presentations, and so on. Out of necessity, all of these activities have continued without interruption during the remote-working era, just on screens.

Let’s add a few more assumptions. Company leadership has set a general expectation that employees will begin to return to the office in the fall at, say, two days per week to reduce crowding. As the CAO, you have some discretion to set expectations for the “live days” of the analytics team but little influence on the calendars of colleagues you regularly interact with in the lines of business, IT, finance, and HR.

Further assume that, per your hard-won experience, all-live meetings are the most effective, followed by mostly live (at least 2/3 of participants), followed by all-remote, with the worst choice being mixed meetings (at least 2 people together in a room yelling into a speakerphone, all others remote).

Given these conditions, what instructions do you give the analytics team about how to set up their two live days each week?

One option is to allow each colleague to individually decide where to be each day, and to set up all interactions to be agnostic to location. This approach will maximize the number of meetings that have a mix of live (40%) and remote (60%) participants—which we just agreed was the worst possible outcome.

Another option is to hope for science to give us rapid testing and effective vaccines and treatments, enabling you to change the main assumption and have 80-100% of colleagues in the office at a time. This would bring you back to the pre-March 2020 condition, plus sweaty masks, minus handshakes. (It’s actually not that simple; there will be episodes of reversion to remote work as clusters of team members develop worrisome symptoms and drop out for days at a time.)

A more realistic assumption is that, for quite some time, the median analytics colleague will stabilize at two days per week in the office. As the CAO, what will you do to use their time well?

Your first instinct may be to sort the weekly calendar by priority and fill it up. In your two days together each week you can hold your staff meeting, an all-hands, a project status meeting, team meetings on all the most important projects, a weekly social event.

But this is the wrong approach. Instead, the activities to do in person are those that have convex returns, that is, where the potential upside from in-person collaboration is very high or where live meetings reduce the risk of ruin.

This would include brainstorming sessions with a rapid exchange of ideas and people talking over one another; exploratory, iterative analytics using dashboards; problem-solving on the trickiest projects; interviewing candidates for open positions that have 10x potential; and holding lots of one-on-one discussions with your most talented team members, if for no other reason than to reduce the risk of their unexpected departure.

These activities are convex enough to set aside one day a week to do them live, and to cajole the team into being present in the office for that day. But what do you do with the second live day?

The second live day should be set aside for cross-functional and spontaneous interactions. Child care and other logistics permitting, this day should float as much as possible from week to week for a given colleague. The idea is to achieve a well-coordinated presence of the analytics team on one day each week, and a random cross-section for three to four days.

In The Secret of Our Success, Joseph Henrich cites the example of the caribou-hunting Naskapi tribe in Canada. Through unconscious cultural evolution, the hunters have developed a system of divination that randomizes the direction in which they hunt from day to day (in this case, using the patterns of cracks in burnt antlers), which leads to more success than a system in which the hunters always return to the spot of their most recent find, since the caribou have evolved to avoid that spot.

In our context, the floating days of the analytics team would involve (safely masked) colleagues bumping into business partners and people in other functions, discovery of what projects these other people are working on, and, for the daring, impromptu after-work social activities among colleagues with loose ties.

Having spent months apart from the team, it’s tempting for a CAO to seize the opportunity to manage the team to spend as much live-time together as possible. But this situation benefits from a mixed level of control, leaving some of the in-person time for serendipity—and of course leaving three remote days for routine meetings and for the team to get real work done.

You should test this idea with your team ahead of time to determine if you have a critical mass of people who want to come in at the two-days-per-week level, and under what conditions. You should also be prepared to make continuous tweaks to the model, including the very real possibility of reverting to all-remote work for weeks at a time—which is not the worst thing, as long as you get to recharge the relationships eventually.

And if you are fortunate enough to have your favorite restaurant survive the crisis and reopen for business, you can take the team out for a socially distanced celebratory meal to thank them for their perseverance.

This article first appeared on the Braff & Co. blog at braff.co. Subscribe for automatic updates.