Avoid the Reorg from Hell with Six Key Principles

Nearly 2 decades ago, I sat in the Taco Bell drive-through with a senior Engineer from Microsoft Hotmail. I was a lowly Engineering intern and a motley combination of cluelessness, curiosity and confidence, a walking Dunning Kruger effect. As I peppered him with endless questions about the Hotmail organization and its execution (thinking, even at that early age, that I knew better than my many layers of bosses), he eventually explained that a certain part of Hotmail had gotten “reorganized”. And that he no longer knew how well that team was doing or even if it was doing anything useful.

My youthful brain stopped in its tracks. What sorts of pointy-haired bosses were running this joint?! Why were they moving humans around like toy soldiers? Why, of all places, was this happening at Hotmail while the young geek gods of Google achieved the heretofore impossible feat by releasing an email client with ONE GIGABYTE of storage? Why couldn’t we just focus on technology and teamwork?

As a point of karmic justice, I am now that pointy-haired boss. I currently lead three different organizations at Square, a total organizational size of ~700 complex, interesting, motivated, messy humans. Most of these folks wound up stuck with me through, you guessed it, multiple different reorgs over the last 4 years. There are surely folks in our team who are asking the same questions I was.

I understand. Org Design — how leaders lay out their org chart — is a huge black box for ICs and early career managers and many re-orgs are poorly communicated. This understandably feeds perceptions that the bosses are incompetent or distant enough to make things worse by meddling when instead they could add value by just staying out of the way.
Funny but (probably) true
On the flip side, I also now understand and believe that if you never reorganize, you can’t ship at scale. Further, Conway’s Law is real, and how the organization is designed will directly impact what will ship and how it will ship.

So, we have an opportunity. We have the opportunity to talk openly and transparently about Org Design to help managers execute and communicate them better. To help stakeholders understand them better. More selfishly, I have an opportunity for Older me in my, ahem, advanced years, to redeem myself with Younger me.

Org Design is complex enough that it pays to start with clear definitions of concepts — these concepts apply whether you are creating a net-new team or re-organizing an existing team.

Pantera Negra 2: Wakanda para Sempre
Black Panther: Wakanda Forever
Adão Negro
One Piece Film: Red
Uma Noite no Museu: O Retorno de Kahmunrah
Babilônia
Emancipation- uma história de liberdade
Avatar: O Caminho da Água

Concepts and Definitions: Here are a few simple concepts tied that are important to know.

Spans of Control: This is just a fancy way of knowing how many direct reports a given manager in the management chain has. A general rule of thumb is that most managers should have 6–8 direct reports at steady state.
Layers: This is simply the number of managers between the top of the organization (or a sub-organization) and an individual contributor reporting to a first line manager.
Charters: This is a short document that refers to a team’s raison d’etre. It explains at a high level why this team exists, what its goals (and non-goals) are and the types of initiatives it will undertake to achieve those goals.
Seams: Borrowed from the glamorous world of tailoring, seams is a squishy concept that refers to the connections between two teams as well as the logical separation of responsibilities between them. For example, if two different product teams work on the same product, how are responsibilities divided up to ensure mutual exclusiveness while also being collectively exhaustive? Seams are critical because they ensure that teams are able to collaborate successfully with each other while still having clear, logical swim lanes that everyone can understand.
OK! With these definitions, we are now ready to articulate some key principles. Applying these with rigor will help managers avoid the eternal infamy and villainy that accompanies poorly executed and terribly communicated re-orgs.

0/ Clarity of Goals: Any org design exercise should start with a simple question — “what are we solving for?”. In my experience, there may be a multitude of reasons to update the org design but the first sign of trouble comes if managers engage in org design without absolute clarity of what their goals are, ideally in priority order. Are you looking to clean up muddled or overlapping charters across multiple teams? Are you looking to fund a new charter by way of defunding something else, resulting in team changes? Do you have 1 or more managers with too many directs? Do you have managers with too few directs? Are there too many layers given the size of the organization and is the goal to reduce layering? Are there too few layers?

After years of helping managers think through org design, the first common failure pattern is missing or unclear goals. Always have clear goals, have them in writing and have them agreed upon by every stakeholder involved.

1/ Consider Design goals separately from Human goals: Organizations of knowledge workers are, in the end, a deeply human endeavor. Any team will be filled with humans of diverse levels of experience, seniority, skill and career ambition. Some will be eager to convert from IC to manager. Some managers will want more scope. A successful line manager will be eyeing opportunities to become a manager of managers. Occasionally, a manager will have too much scope and be interested in reducing it to increase their focus and level of IC support. Frequently, senior ICs will be understandably picky about the seniority of their own manager so they can learn from them and be counted on for strong support.

It is common for managers to be worried about the retention implications tied to their decisions. If I decide X, will Persons Y and Z quit? It is also common for managers to be caught up in the emotional impact of their decisions — after all, design goals are nice, but they know these people. Frequently they are friends with them. Any well-intentioned manager would, and should, worry about the human impact of their choices.

While the attention paid to the needs and reactions of humans is understandable, it also leads to a common failure pattern — when managers try to muddle design goals with human goals at the same time, they can wind up with an org design that doesn’t actually serve many of the goals they started out or, worse, solves for a few humans while actually making existing problems worse and leaving other humans also worse off.

The counterintuitive but correct way to approach this is to at first engage in org design without any current team member names in any of the boxes of the org chart. Put another way, managers must ask and answer the following question — “If we were starting from a clean slate and building a brand new organization from scratch, how might we design it?” This idealized approach, while likely infeasible, forces managers to focus on system design first and foremost. This system design should successfully achieve the goals laid out in Step 0.

Once the idealized org chart is laid out, managers can now iterate by taking human needs into account and produce a more real-world org chart that is an evolution from idealized state and also takes into account as many human considerations as possible.

It’s a simple mental hack but it works.

2/ Pick Explainable, MECE Charters and Seams: Some re-orgs are tiny — an IC converting to manager and taking on a small team of 3 direct reports is an example. Yet others are quite large, involving multiple teams changing charters or getting created out of thin air, several different manager changes and so on. And many more are somewhere in between.

In most of these changes, charters and seams between teams really matter. In most tech companies at any scale, many different teams will work on different parts of the same product. If we fail to think carefully about seams, it is all too easy to wind up with teams that duplicate work since charters and seams were left undefined. It is possible to wind up with teams that engage in turf wars over a particular area of the product. It is easy to find situations where large, obvious initiatives that someone should be focused on get ignored altogether because no particular team was ever put in charge of them upfront.

Seams can help. Seams are both the logical separation between team charters and the connective tissue for team collaboration. For example, for a mobile point of sale app, “Seller” and “Buyer” may be a reasonable seam — one team focused on features that sellers/cashiers are likely to use whereas another team focused on features that buyers are likely to interact with while transacting, with both teams working together on features that impact both. For a workforce management product, “Scheduling”, “Time Tracking” and “Employee Onboarding” may be the right set of charters and seams. In other cases, seams designed around key organizational objectives or top company priorities may be the right way to go.

In any case — managers must be able to explain both the charter and the seams in their org design in plain English. If we are unable to draft clear, public written communication to explain this to our teams, it’s time to revisit the drawing board.

In addition, managers must ensure that seams and charters are designed such that the MECE principle is achieved — as in, ideally there is mutual exclusivity between teams (thereby resulting in clear swim lanes and reducing duplication) as well as collective exhaustiveness (thereby ensuring that no major area or initiative goes simply unaccounted for).

A common way to test that we’ve done this well is to pressure-test any proposal with known and funded initiatives from the next 12 months. At Square, we frequently ask managers to pull the list of funded projects and hand-assign them to the group of post-reorg teams and then review them one by one. If most projects can be assigned and the assignments feel logically grouped and easily explainable, we’ve gotten it right. But if a meaningful chunk of work feels orphaned or arbitrarily stashed all over the place, we know we have a “junk drawer” problem in our design and may need to go back to the drawing board.

3/ Design for (some) durability: Much of the negativity attached to reorgs is anchored to employees feeling like teams are simply shunted around every few months at the whims of distant senior leaders who are too eager to wield a disruptive management tool as they stumble from pillar to post.

I regret to admit that this perception is grounded in some reality. Since reorgs are inherently disruptive and may involve ICs (and managers) losing their prior managers to get new ones or losing their prior charter in favor of a new one, org designers should try hard to adhere to the Do No Harm principle. Additionally, org designers must be able to convince themselves that any reorg must be able to stand at least for 1 full year, thereby giving the team enough runway to achieve the goals set out earlier.

Once again, this can be pressure-tested by listing 12+ months worth of projects to each post reorg team. If a given team can easily run out of high-impact work in just a few months, we may have to revisit the overall design.

4/ Mind the (Cross-Functional) Ratios: Most software products are built by team members from different disciplines/functions that come together to achieve the team charter — Engineers, Designers, Data Scientists, Product Managers and Product Marketers. These disciplines may have further specialization within them — iOS Engineers, Android Engineers, Server Engineers, Frontend Web Engineers, for example. Any org design must consider the ratio between and across disciplines. For example, a high-level rule of thumb is to have 1 PM for every 8–10 Engineers. This ratio can be flexed up or down depending upon the specific nature of the team and product, of course, but it provides a starting point in most org design conversations. Further, teams can and should develop similar ratios indexing PMM to PM, Design to PM or Engineering and Data Science to PM or PMM or both.

Why are ratios helpful? Because any well-functioning software team is an orchestra combining work across disciplines. A team with too many Engineers and too few Designers will end up with Engineers either blocked on Design or with Engineers attempting to design on the fly — both bad outcomes. A team with too many PMs and Designers and too few Engineers, on the other hand, will be even worse off — able to produce high-quality, well designed ideas that have no hope of actually shipping.

Once you’ve got your org design, check every ratio between and across disciplines. Got it wrong? Back to revisiting the original design.

5/ Communicate in Order, in Detail and across Channels: The most reliable way to get every prior step just right but still end up with frustrated, confused teammates is to simply communicate poorly. An excellent comms plan can ease anxieties, create benefit of the doubt and set up ICs as well as managers for initial success. An excellent comms plan can even fix problems in Steps 0–4.

And yet, Big Tech reorgs are unfortunately rife with bad comms plans.

A good comms plan must bring stakeholders into the know in the right order. It must communicate in detail, anticipating concerns, questions and points of frustration. Finally, it must work across multiple communication channels. Let’s consider each of these.

First off — any reorg will usually feature a very small group that is responsible for the actual design. A somewhat larger group may be consulted. A senior leader may be the final approver. And a much larger group would be usually informed. It is usually critical to communicate with each of these groups in order, bringing them in the know (ITK) at the right time. A bad comms plan usually features people being brought ITK without forethought or sharing half-baked designs with large groups. The telltale sign? An active rumor mill that is in full bloom weeks ahead of any formal communication, with all the gossip, anxiety and frustration that accompanies this game of telephone.

Given this, we teach every manager to build and maintain a detailed ITK list, with concentric circles of stakeholders, brought in step by step with appropriate context and caveats.

Secondly — despite what our neat-looking org charts show with boxes and arrows, reorgs involve humans frequently being put into semi-awkward situations. Some ICs may be getting a new manager. Some managers may be changing or expanding scope. New teams or charters might be getting created on the fly. Any group of empowered knowledge workers impacted by a reorg can and will ask a series of questions when the reorg is announced. Why are we doing this? What are our goals? Why did we pick this particular path? What could go wrong? How will we measure success? Why should I trust that this is good for my career? Why should I buy into this new team charter? More importantly, these questions will be asked not just in public — but also in private, of other managers and stakeholders that were ITK earlier.

Given this, every reorg comms plan must feature a series of detailed write-ups — emails explaining the change to a broad audience, FAQs for any stakeholder fielding questions in 1:1s or office hours, quick talking points for any stakeholder walking through in small group settings.

Think this level of detail is bureaucratic overkill? Maybe — but in my hard-lived experience it is superior to the alternative, with teammates getting so frustrated by perceived stonewalling or sloppy responses that they exercise their right to not participate — by exiting your team.

Finally — reorgs must communicate with every available channel to leave no stone unturned. Group emails to various distros? Check. Slide deck walkthrough with visual org chart? Check. Open AMA to field questions in-person and on video? Of course. Office Hours after the fact? Natch. 1:1s and small group syncs to encourage shy folks to open up? Absolutely.

Flood the zone with every channel you have access to and every tool in the toolkit. You may not get everyone to agree in the end, but you’ll infinitely raise the odds of everyone getting to disagree and commit.

Once you’ve done all of this, in adherence to the best practices above, your reorg is now in effect. That’s all, folks!

Except, not. The real hard work actually begins the day after. Even if skeptical, many employees may give you the benefit of the doubt. It is now your mission, should you choose to accept it, to reward their provisional trust and turn it into a durable respect by making the reorg successful and achieving your stated goals. And maybe, just maybe, helping all your interns for the world they will live in two decades down the road.

Comments are closed