As companies rapidly grow, they need an organizational model that allows them to innovate and make decisions quickly.
As organizations scale, there are a host of novel challenges that make success far from assured. Ways of doing things that worked well at one size and scale will no longer function efficiently at the next level. The most insidious of these challenges are those relating to the speed of decision making. Maintaining a fast cadence on decision making becomes more difficult as companies find success and grow in size. Successful companies tackle this challenge head on and intentionally design organizations that can maintain speed and innovation at scale. Unsuccessful companies do not and then belatedly assign blame to imprecise factors like “having lost our way.”
It’s almost impossible for an organization to organically adapt during a period of rapid growth that is commonly associated with tech companies who achieve success. It’s also very unlikely that a company will find itself with an organization that can thrive at different levels by happenstance. All of that to say, a company that doesn’t pay attention to organizational design and how it impacts their business will not be long for this world.
At Hopper, we have had a front-row view to the challenges of scaling and how that necessitates intentional organizational design. At the beginning of 2018 we had 100 employees. By the end of 2019, we had over 400. This rapid growth put the organization under immense strain as the speed and clarity of our decision making declined.
Going into 2018, Hopper was small enough to employ what was effectively a top-heavy matrix design but still maintain a fast cadence on decision making. This centralized-style of management, requiring near consensus among several department leaders to make decisions, is effective at smaller companies because any delay in speed is counterbalanced by the simplicity of gathering all of the decision makers in the same room, typically with the CEO acting as arbiter.
As our organization grew, this way of operating was no longer working and we saw our cadence of product development decline precipitously. So in 2019, Hopper decided to change its organizational model and move from a horizontal matrix-based organization to a vertical structure based on single-threaded ownership. The objective was to increase the speed of decision making by pushing down ownership and autonomy to leaders of smaller, cross-functional teams. Each of these teams would be working backwards from a customer objective or business objective, as opposed to a task, and would have all of the talent and capabilities needed to deliver a product to our customers from start to finish. Instead of functional department heads, the teams are led by “Leaders” who are empowered to make decisions and manage trade-offs without escalation for approval.
To make this culture of autonomy work at scale, we paired it with a well-defined set of core operating values, tenets and a number of guiding principles that clearly outline what we want to achieve as a company, how we want to achieve it and how we expect our leaders to act. The minimal oversight is further facilitated by the creation of a few “forcing functions” and budget control to create a clear framework for our leaders to navigate in search of a clear goal: innovation for customer impact.
This type of leadership model is not new and has been effectively employed at many other organizations in the past and is especially common inside marketplaces - Amazon as the prime example - as the challenges of running a business with many distinct revenue lines often necessitates this type of organizational structure.
For Hopper, the results were spectacular. Almost as soon as we made the switch, our business soared to new heights, our revenues increased nearly 8x and the cadence of our product innovation improved dramatically.
A core part of our new organizational model is moving away from a small number of large, function-based teams and replacing them with a greater number of small, cross-functional teams. It’s central to the premise that decision rights are more broadly distributed throughout the organization than they otherwise would have been in the matrix-system. This requires us to identify a larger number of leaders who are capable of directing these cross-functional team units. To set up these leaders for success, it’s imperative to also structure the teams to have a well defined and manageable scope.
To that end, Hopper has organized itself into smaller and stable units of 8-12 persons centered around coherent business and customer objectives — a so-called Roman Tent Team — known internally as verticals. Each vertical has a clear objective and focus area and one designated leader with established decision rights. These verticals, for the most part, do not need to coordinate among each other to drive impact for the customer. It’s important to note here that the verticals are structured to work backwards from actual business or customer objectives and are not organized based on the type of work. Meaning, something like Customer Service is a valid vertical but something like front-end mobile development is not.
Business Fintech Unit | illustrative example:
One way to think of the verticals is to imagine their staffing as an investment into the leader. The typical vertical costs approximately $100K USD per month to operate. To operate a vertical for 12 months, then, is the equivalent of a $1.2M seed-stage investment from Hopper into the leader’s startup.
The leader maintains this contract with the business through the following of the organization’s guiding principles and tenets and by adhering to the company’s values. The business provides guidance and oversight on an ongoing basis through the utilization of the forcing functions like roadmaps and team staffing reviews.
Ultimately the goal is for the leader to create a positive return on investment, as measured by revenue generated over time in relation to the cost of staffing the vertical. If that can be achieved, there are many opportunities for a leader and vertical to grow in scope. Other times, if that cannot be achieved after a reasonable period of time, an alternative path must be pursued.
One problem with the matrix structure that Hopper was using is that we lost clear sight of who held decision rights on any given project. As that was unclear to the top of the organization, it was even less clear to the average employee inside the company.
The truth is that when it’s unclear to a group who holds the decision rights it’s always because no one in that group actually has decision rights. This creates two unfortunate phenomena inside an organization:
Decision-making by committee: The death knell of all innovation, this tends to occur when members of the group aren’t sure who is responsible for making the decision and revert to safety in numbers. The primary downside of the committee is that it diffuses responsibility and accountability for the decisions being made. It also leads to the outcome that charts the path of least resistance and because of that is not particularly well suited for innovative work like that at a tech startup. The real outcome is endless hours wasted in meetings and brainstorming.
Frustration at arbitrariness of decision making: In the innovation industry, committees don’t typically actually have decision-making authority; they just think they do. What usually happens inside the organization instead is that there is a hierarchical centralized system and information is slow to find its way up the chain of command. Eventually, though, the information finds its way to the actual decision maker and a decision is made that may or may not overrule the previous one made by the committee. Because the person holding decision rights was not involved in the committee and because the outcome of the decision may take weeks or months to find its way up and down the organization, this creates the perception of arbitrary decision making by the executive team and leaves the individual contributors who participated on the committee feeling frustrated.
An executive team itself, though holding more power, is not immune to the two problems above. The matrix system of multiple department heads acts as a committee of sorts in that it also diffuses accountability and responsibility for decisions. Department heads frequently find themselves spending their time attempting to achieve cross-department buy-in from their peers. Pleasing other internal groups or avoiding conflict takes precedence over doing what is right for the business and the customer.
That said, this plays out at a faster cadence than elsewhere in the organization as the executive team is closer in reporting line to the CEO, who is the one holding actual authority and serving as the final arbiter on any conflict. But, of course, the executives who have sought to form a consensus may also find themselves frustrated at the seemingly arbitrary nature through which their disagreements are resolved by the CEO.
Needless to say, this doesn’t scale well as the speed that information moves through organizations declines as they grow and it is not possible for the CEO or a senior executive to weigh in on every meaningful decision - nor should that be a goal. So innovation suffers as the company reverts to being equal parts slow, arbitrary and consensus driven.
In our model, Hopper seeks to overcome these challenges by assigning decision rights to a leader and making that crystal clear from the onset and before any work commences.
The leader is the decision maker of the vertical and is the person who is ultimately held accountable for the success or failure of the initiative. Information does not need to flow throughout the organization as the decision maker A) doesn’t need to coordinate with other verticals, for the most part and B) because of the smaller team sizes is never more than a level or two removed from the work being done. This proximity has a direct impact: decisions are made quickly.
The leaders are accountable to the business only through the results that they deliver. Command control of the business over the independent verticals is administered through forcing functions like roadmap reviews, weekly revenue reviews and ultimately through budget controls.
No organizational design is perfect and there are drawbacks to adopting the single-threaded ownership model. An organization will become markedly less coordinated as decision making is distributed and the presence of multiple autonomous teams will lead to the duplication of work and periodic miscommunication.
So while consensus-driven and centralized decision making is slow, there is no denying that it is very coordinated. This can even be the ideal organizational design to pursue if it’s more meaningful for the group to prevent error than it is to move fast. Think of banking, healthcare, government regulatory bodies, education and other organizations that are expected to behave with prudence. Likewise, if a company is already highly profitable it may want to avoid mistakes that could cause it to lose market share more than it wants to innovate to capture incremental profit. There is a lower cost to delay when one makes millions or billions of dollars in the interim period.However, when a company is trying to capture market share in the innovation sector it is more important to be fast than perfectly efficient and what Hopper has lost in coordination by pursuing a single-threaded ownership model it has more than made up for in speed.
So it’s expected that there will be less coordination and even some duplication of work in this organizational design. That’s tolerated because the cost of inaction is so high. Every month without profits is a multi-million dollar cost to the company that outweighs any real cost derived from inefficient coordination. So instead we move as fast as possible to grow the revenue to offset the costs of operating the company.
That said, chaos is also not a productive environment for innovation. To be effective, we must be reasonably assured that the various parts of the company are working at purposes that are aligned with the objectives of the company.
So we seek to replace the loss of coordination from consensus-building with a culture of autonomous alignment. One in which leaders are empowered to make decisions without close coordination but at the same time are expected to develop a detailed understanding of company objectives, organizational values and cultural principles and to use that to guide their decision making.
It is both the responsibility of the company and the leader to ensure that this high-level business context is understood and effectively communicated across the board. It falls on the organization, though, to ensure that the operating framework is easy to navigate, that the objectives are easy to recount and to provide transparency on what will be measured and rewarded.
To continue on the points above, it’s through the establishment of a well defined set of values and principles that the business is able to distribute decision rights to multiple leaders and overcome the aforementioned loss of coordination in a way that doesn’t lead to different groups working at cross purposes.
One challenge in the single-threaded ownership model is that the high-degree of autonomy afforded to various groups can lead to the creation of disparate cultures within the same organization. This is not a problem, at first, as it is possible for many different cultures to be effective and it may not even be noticed as the emphasis on speed is greater inside the company than the desire for cultural similarity.
There is a long-term problem with disparate cultures, though, that can be best understood as a challenge to the interoperability of the company. Different groups will create different ways of doing things that are not transferable or easily explainable to others. Loyalties will be split, competition will increase and the subsequent mistrust will impair the ability of the organization to work towards a coherent objective.
To solve this problem, an organization must intentionally construct a culture that goes beyond aspirational statements and instead creates a set of shared values and principles that are actually applicable to the tasks at hand: innovating to achieve business and customer objectives.It’s through the explicit outlining of these values and principles that otherwise unconnected teams are able to communicate in a standard language and build trust. Culture, in this way, effectively serves as the operating framework through which decisions can be made in a way that is autonomous and distributed but still consistent with the organization’s defined world views and objectives.
Autonomy is not possible without accountability - in this case, to the business. As decision making is disintermediated by design in this organizational model, it is not possible for a senior executive or the CEO to weigh in on every decision made.
And while the leaders are ultimately responsible for the strategy, direction and outcomes of their groups, the business employs a variety of forcing functions to provide guidance, course correct when needed and hold the leaders accountable for results.
These forcing functions are the tools that enable the culture of autonomous alignment to succeed, as they create a clarity in expectations between the vertical leaders and the business.
• Quarterly roadmaps approval
• Business cases for material projects
• Weekly revenue meeting: meeting every week to discuss progress against the quarterly revenue targets as set out in the team roadmaps
• Weekly servicing meeting: meeting every week to make sure that customers are being serviced adequately post booking
• Periodic product and design reviews: including the product managers and discussing the UX/UI of the features being released.
For our model to continue to succeed, Hopper needs to put its best leaders forward to tackle our toughest challenges. That requires the company to match the talents of its individual leaders with the skills required to succeed at the defined scope of the business or customer problem.
We specifically do not want a culture where only business-school graduates can lead a team or only those with product and design skills can lead. To that end, initiatives, verticals and even whole business units may have leaders from a diverse background of skills regardless of what functional team they may have been on in the past or elsewhere.
Initiatives that will clearly require innovative data solutions should have leaders who come from a data science background; initiatives whose success or failure depends on solving hard technical challenges should have engineering leaders; business-driven initiatives that require Hopper to take on balance sheet risk should have business leaders; and so on.
The takeaway here is that we believe that any type of background can succeed as a leader at Hopper if their talents are correctly matched with the opportunity at hand.
A leader is expected to dive deep and to develop an intimate understanding of problem areas, the customer and the projects being worked on inside their vertical. Writing business cases and narratives are one way for the leader to demonstrate their deep knowledge. Beyond that, though, the leader should have a deep familiarity with the data, design, product requirements docs and the work going into the engineering sprints. The leader is expected to meet with their team to core sample, ideate, course correct and encourage the group to think big and avoid the natural bias towards local maxima.
That said, as the team units are small and focused only on a defined scope of projects, it is inevitable that some initiatives will fail and people on teams will go on to work on other initiatives. When this happens, at times, the leader will go directly on to leading other verticals and establishing new team units. At other times, the leader may find themselves acting as a contributor on a more promising initiative led by another.
Not only does this type of failure not reflect poorly on the leader, but it is an opportunity to show true leadership. Working without ego towards the greater good, learning from mistakes and succeeding in positively impacting our customers are the only qualities that will be rewarded at Hopper. In the face of failure, the leader should enthusiastically do the right thing for the business and customer and use the learnings to further their own development.
We cannot be successful at innovation if we build a culture that is risk averse and driven by fear of personal failure. The only path forward is onward.
For those reading this who aren’t yet in leadership positions, the path to becoming a leader is to act and behave like a leader. To embody the company’s leadership principles, take ownership and deliver positive results for our customers - whether that’s as leader of an entire business unit or the leader of a single initiative within a single vertical.
Innovative ideas are realized through writing business cases that seek to affect meaningful change and then iteratively testing and learning from our customers to make it a reality.
Verticals are created by writing narratives that outline a vision that is too compelling for the business to pass on.
Act like a leader and Hopper will invest into your career development. Become the holder of a powerful idea and Hopper will invest into you as a leader and treat you like a startup CEO.
We intentionally designed our culture to give leaders ownership and autonomy, empowering them to make decisions based on a set of guiding principles and tenets to achieve maximum impact for the customer.Sounds like a fit? Check out our open roles!