Why care about the care factor in an analytical team?
Lessons from getting to the starting line
Exhilarating as it has been to start up Data & AI at Reece, it has also been, and continues to be deeply consuming. Building a team is, at the best of times, hard work; and certainly being at the ground floor of an enterprise-wide transformation presents, in equal parts, limitless opportunity… and countless ways to stuff it up. Add to this unprecedented levels of talent scarcity, the work not only constantly feels unfinished, it can at times feel like it’s not started.
But I had time this weekend for pause and reflection. Lifting my thoughts to the 10,000-foot view of our practice, I was reminded of Bill Gates’ oft-cited quote “Most people overestimate what they can do in one year and underestimate what they can do in ten years.” — examining the team albeit only at the starting line today, I can hardly believe the people we’ve found, and how different we look from a mere few months ago.
Working backwards… but from where?
The core tenets from Working Backwards have always made sense to me — identify the problem, weigh up the value of potential solutions, chart a delivery path, stick to it. The approach seems to generalise well — professional, personal, macro and micro things. Want to live in a house? Then explore the options, evaluate them objectively, decide, make a plan and stick to it. Want to learn a programming language? Then investigate the alternative pathways, weigh them up methodically, choose, make a plan and see it through. Want to get to that dinner on time? Then look up the travel time on Google maps, and be sure to leave home with that amount of time to spare.
Naturally, when getting started at Reece, the first questions I contemplated were: what kind of organisation is Future Reece, and what’s our strategy towards that? What exactly does a Data & AI practice in service of this strategy look like… and at the heart of this question, what kinds of people do we build the practice around? Suddenly at this level of questioning, the Working Backwards method didn’t look quite as straightforward.
People don’t fit in tidy little boxes
I realised that it isn’t enough to write down, say, optimal career pathways and ideal profiles for our data engineering function — we’d also need to figure out (with a strong dose of humility and pragmatism) how we’d make reecetech the type of place these sorts of people would want to be. What makes things hard is that “these sorts of people”, by virtue of being “people” — is a shifting target.
To say people are complex is like stating the obvious, but perhaps the past year’s structural shift in the talent landscape is drawing greater attention to these complexities. Out of general curiosity, I ask every candidate what they’re looking for in their next role — and it’s become apparent it isn’t headline motivators like interesting work, respect and culture that are hard to satisfy. The issue is, what these motivators mean to someone evolves with the person, therefore what someone values in their role is an ever-morphing force to be reckoned with. People learn and grow, often as a product of their values, life stage and their environment. And in most workplaces, environments are themselves a function of even more people — making the challenge seem intractable, but also eminently fascinating.
Analytical environments are a maelstrom
I tend to think of analytical teams as the connective tissue of an organisation. Generally speaking, data consumers are rarely the same teams as data producers; these groups tend to sit in different parts of an organisation. Teams in Finance, for instance, consume data derived from transaction sources maintained by the POS engineering teams. Analytical teams fill the needs between data-producing and -consuming teams, building the platforms that turn source data into enriched, consumable forms. In this design, analytical teams tend to deliver across more people, from more diverse domains and cross-sections, than many other technical groups.
Such an environment presents unique challenges for an analytical team:
- Exposure to second and higher-order dependencies: For example, an engineering pod constructing a new web application is concerned with feature viability and user patterns. The downstream analytical team is equally interested in these, but in addition wears the impact of a new application on current reporting and analytics, as well as the demand created for new reporting and inference.
- A massive balancing act: Analytical work spans platform production-support, ad-hoc analysis, data product development and transformation initiatives. Each piece requires a different throughput, yet isn’t readily separable from the other. The same transactional information typically underpins: board reporting, operational decisioning, ad-hoc experiments and machine learning for new data products. All must be optimised across the team’s capacity.
- Things spiral, fast: This heavily connected nature of an analytical environment means the actions of individuals, however small, can be wildly multiplicative. Suppose a lapse in data lifecycle management leads to a core dataset becoming corrupted. Now three reports (one read by an exec of course), a few operations dashboards, and an ML pipeline are all on fire at the same time.
Against this vortex, the anchoring properties upon which to shape a thriving team seem elusive. But there are a couple of attributes, from experience, that have predicted stably for success in a data team — let’s call them Care Factors for success.
To succeed, first we must care
“Empathy is one of our greatest tools of business that is most underused”
— Daniel Lubetzky
The first Care Factor, unsurprisingly, is empathy — the ability to understand a problem outside of oneself, and grasp another’s unmet, unarticulated needs. Perhaps less obvious, in my experience, is that the state of organisational empathy is clearly revealed in the small stuff. Every expression presents an opportunity to build up, or tear someone down. Consider these scenarios
- “How are you placed next week to explore problem X together?” ––actually says “I respect your craft, and time, and I’d like to collaborate with you”
- “I know you’re busy but can you import this table ASAP? It should be fast and I’ve already promised the customer the report” — actually says “I have no concern for your craft, time or autonomy, or the needs of the others you’re helping”
While an analytical team is often a focal point for inferential activity, for an organisation to successfully derive value from data, empathy needs to be widespread: because everything is multiplicative when it comes to analytical data use. And it’s only when people are pulling in the same direction on empathy does the value of data-use scale.
The second (related) Care Factor is that one must care for a greater purpose than oneself. Because data spans a myriad domains and is hugely propagative, a me-first view of any one thing almost certainly will come at the expense of the organisation. An engineer, for instance, whose identity is wrapped up in being the “Know-It-All” guy, is likely also allergic to documentation, putting both the team’s effectiveness and institutional knowledge at risk. What we do and don’t do in one domain affects another domain; not considering the wake of one’s actions is a recipe for creating tech debt faster than any platform will remedy. Equally, if everyone takes concerted action to make things better across business, we have ourselves a Data Flywheel.
On the unreasonable effectiveness of caring teams
In organisations that are cultivating large-scale change, everyone matters in holding each other accountable to the standards of care, while keeping in mind the audacious goals that are larger than the individual. Our team’s most thrilling problems are only 50% cutting edge tech — the other 50% is building a movement around both tech and cross-team consideration, which is what Reece needs to accomplish change beyond what any single team, or group of friends can achieve.
Working backwards from the goal of building a great team is hard to get right. But as a team we’re optimistic that with care, we’ll get where we want to go. Bringing consideration into every operating facet (e.g. career ladders, performance principles, disciplines, and social rituals) has meant that caring is woven into our DNA. And the early signs are encouraging: an 84% reduction in issue resolution time, or the move to having 100% of delivery documented — looking past the numbers, one of my proudest experiences was seeing the team interact on a Hump Day social over a competitive game of City Guesser.
On a personal note, while ambitious change isn’t finished with a single team or a group of friends, I’ve learnt that it does matter to start with caring, in a team, like you would a group of friends. And the most unexpected find from these past months has been just how great this data posse’s capacity is to care.
