Data & the Science of Leadership

Data & the Science of Leadership

By | 2018-04-17T18:33:32+00:00 July 23rd, 2014|Blog|

I started developing software in 1994 and I’ve had the fortunate opportunity to work on some of the most ubiquitously-used applications on the planet. When you check out at the grocery store, log issues in JIRA, search for a new job on the web, play McDonald’s Monopoly, or when you manage your IT host and application alerts… there’s a good chance you’re using software that my team helped make for you. Over the years, I’ve learned one thing above all others.

Squeaky-wheel meetings are for suckers

A long time ago, about midway through my career, I was working as a Development Lead at a high-volume web development agency. At any point in time we were juggling 200 active projects with nearly all team members cross-allocated onto multiple projects each. When I first took over the team, we had a daily meeting called the Missed Deliverables Meeting. Each morning a sampling of leaders and project managers would assemble and discuss all of the work which was late to clients. This meeting was plain and simple… a squeakiest-wheel competition. There was always some shouting, often some crying, and occasionally the throwing of sharp objects. At the end, in fight-to-the-death fashion, the team collectively ‘agreed’…and I’ll use that term loosely… what work we would tackle for the day’s docket.

Note: Nowhere in that meeting would we talk about today’s regularly scheduled work. We’d discuss that work after it was late.

As a guy who loves Excel and passionately loathes squeaky-wheel competitions, I turned this problem into an algorithm. It was a frankenstein’d mashup of Active Directory, Bugzilla, MSProject, Sharepoint, and a database of my own creation which for a surprisingly long time after my departure was still called the DCQ (Dan Chuparkoff Query). Each day, the work for the day was triaged and prioritized algorithmically based on distance to launch date, the amount of remaining work, and a variety of other factors. This solution was a hack, I admit – but a glorious hack! And it solved the problem, eliminating the Missed Deliverables Meeting and virtually all of our late work altogether.

More importantly though, it taught me a crucial lesson in leadership. Squeeky-wheel meetings are for suckers. Data is the best leader.

You don’t have to like it. It just has to get more clicks.

When a company like Amazon makes a slight change to it’s website, it can quickly result in a daily increase in sales in the millions of dollars. Although I have yet to see the inside of Amazon’s Operations, I suspect that deciding how to optimize the Amazon purchase funnel is very much not a directive passed down from the two-pizza-team-leader, Mr. Bezos himself or even from his leaders below. These decisions are made with data.

When it comes to web development and shopping cart optimization, we now live in an age of widespread A/B tests – even, often A/B/C/D/E/F/G tests. This is Growth Hacking – the process of growing visitors, evaluators, and customers using small iterative improvements. Here within the boundaries of the web funnel, committee based decision-making is dead. Strong, decisive, experienced leaders and designers everywhere are being replaced by Excel-wielding, hipster, data-scientists. They leverage experiment-based tools like Optimizely. They’ve hijacked the web development decision-making process and they’re not giving it back. They know that data makes better decisions than people and they have the pivot tables to prove it. So why isn’t there something called Leadership Hacking?

Algorithms > Gut + debate + whine

Beyond the web funnel, leaders and the teams along-side of them are still making decisions the old way – fueled by gut, and debate, and whine. IT teams everywhere triage their constant stream of alerts manually one at a time without the aide of strong prioritization and clustering algorithms at their side. Agile Development and DevOps teams everywhere, discuss priority in standup meetings without a clear understanding of the high-level context containing the near-term tasks.

The color of the TRY button on your website has 10 charts and graphs showing you exactly what works best. But everything else in your organization is left up to the gut of a leader, debate between a small number of collaborators, or the whining of a passionate team member pleading for change. It’s time for that to end. We have the power… and the data… and the algorithms… to make this better. It’s time to embrace the science of leadership.

Data-Driven Decision Making for the Win

Why should strong data-backed decision-making be limited to your web funnel? It shouldn’t. The same disciplined demand for stats and experimentation should be applied to every decision leaders make. When you’re leading a team, leverage the wealth of data beneath you to make the right decisions at the right time. If you’re not looking at data when you make it… then it’s not a decision… it’s a guess. When you’re evaluating new tools. Don’t just open them up and click on things aimlessly. Install two tools side-by-side and A/B test them with real users divided into separated test groups. Measure the difference.

You haven’t A/B tested meetings vs. email for project collaboration. You don’t have a graph comparing wiki-usage vs. email-usage over time for information documentation. You don’t have an algorithm helping you to figure out which tickets to work on next. And you should. Lead with strong data behind you. A/B test everything. Prioritize using algorithms and you’ll find that your team becomes exponentially better very very quickly.

BigPanda was founded to help Modern Ops teams overcome exactly this challenge. Bring an algorithm into your triage and incident-management process. Scouring through streams of endless alerts and stack-ranking hundreds of tickets in your head isn’t just slow and tedious… it’s less effective.

Still haven’t tried BigPanda?

Automatically cluster Ops alerts from all of your monitoring systems into incidents.

Get a Free BigPanda Account

About the Author:

Dan was Director of Product Growth at BigPanda. Connect with Dan on LinkedIn.