How Dailymotion hacked its feature team project model to shorten time to market
An inside look at Dailymotion’s organization transformation to reduce dependency and accelerate delivery three years after implementing a Spotify-like model
Since Vivendi acquired Dailymotion in 2015, we had been relying on a Spotify-like project management model which had served us well, allowing our Tribes to focus on quickly delivering a new product vision. Three years on, our ambitions and constraints had evolved, so we created our own model. Here’s why we made this move and how we now structure our team and work.
The Genesis of our Organization Hacking
After three years of reconstruction and more than 100 talented people recruited, our global priority is now to improve our market competitiveness. For this, we needed to better address the two primary needs of our users and partners: time to market (how to deliver valuable features in a competitive amount of time) and day-to-day (how to ensure we support them with their day-to-day requests when using our products).
Compared to Spotify, we had an obvious scale issue, but more importantly, we needed a model based on our own constraints and ambitions. By organizing ourselves into strong vertical feature teams, we had created a myriad of dependencies that would afflict each project, not to mention that the shape of our projects had changed over time. It was getting more and more difficult for all of our Tribes to follow the same path.
We spent several months benchmarking different models and were particularly inspired by a range of project management thinkers, from Lara Hogan and Général Pierre de Villiers to Basecamp’s Shape Up. We also took the time to listen to our people. We analyzed employee surveys, organized focus groups and came to the conclusion that having a backlog at the company level — and no longer at Tribe level — was going to be the key. This led us to base our new model on a space analogy: build the best team, put them in a space shuttle with a limited amount of fuel, send them into space, ensure they come back safely and willing to embark the next mission.
Our Project Management Model
As a medium-sized company, Dailymotion has to choose its battles carefully, and this is actually the cornerstone of our new organization model. If you’re facing similar resource or scope challenges, you’ll no doubt be interested to learn how we created our new delivery lanes and roles.
Firstly, we kept what was working well, including our existing career paths as they provide a strong basis on which to build our “Crafts” — teams of talent with a specific expertise. Each of the vertical pillars we had defined a few years ago now work on their own time to market and day-to-day delivery lanes (Product, Partner, Design, Data…).
Delivery Lanes
Imagine you work for a space agency. Depending on your profile, you could be working on a mission — with a specific target and team, knowing exactly when you’ll take off and come back to earth — or you could be working on the ground lane — ensuring that day-to-day work is done and easing the life of your co-workers who are flying to space. Following this analogy, we created three new delivery lanes: the Discovery Mission, the Delivery Mission and the Ground Lane. Each with specific goals, timelines, teams and rules.
Missions
Everyone has, at some point in time, been faced with a project where too many people, all from different teams, are trying to solve the same issue in their own way, where there is no real boss in charge and multiple dependencies. The result being you end up spending months or even years working on a sprawling project that never lands…
I believe the key to solving this is to bring together a dedicated group of people for a pre-defined (reduced) timeframe with a view to achieving a common goal. Effective collective experiences — like war rooms or new product launches — have been the climax of Dailymotion’s journey throughout the years, the very moments where working together was the most meaningful and efficient. They gathered all the ingredients of success: all the required people in the same room, clear goals, a defined timeline and a leader dedicated to orchestrate the work and take the decisions. Why limit this format to war rooms and new product launches? Missions are directly inspired by these moments.
Whether Discovery or Delivery the aim of a Mission is to solve Dailymotion’s time to market challenge.
We created two types of mission: the Discovery Mission and the Delivery Mission. The first one lasts for two weeks, the aim being reduce uncertainty and define the work to be done on a future Delivery Mission.
Delivery Missions last for six weeks and allow dedicated teams to meet a specific pre-defined goal. A Delivery Mission always has to deliver value, either directly to the Partner or to the company. It can be a new feature, an improvement or the development of a new internal or external tool. The underlying purpose of a Delivery Mission is to push something into production.
What makes our Missions effective is that we staff them appropriately. Mission members are chosen because they are best placed to help achieve the goal. This means a mission can gather product managers, project managers, front-end and back-end developers, data analysts, designers, copywriters, and others from different Crafts. During the 6 week period, this limited group of talent is 100% dedicated to the Mission. What distinguishes this from a classic feature team organisation is that by time-boxing the exercise we can reshuffle the deck after each delivery.
Ground Lanes
Ground Lanes ensure that the day-to-day business is supported within each Craft. Bugs, small features, SLAs, technical debt, prioritized backlog, a last minute story… whatever the issue the people who work on the Ground Lane are in charge of ensuring the smooth running of the Dailymotion machine.
Staffing the Ground Lanes is also a challenge as we cannot afford to have all our talent working on Missions and neglecting the day-to-day work. Ground Lanes are not only about run, they’re about ensuring that we can deliver small features and optimisations outside of a 6-week Delivery Mission to continually improve our product set. The whole point of separating these lanes is to remain competitive on all aspects of our Partner’s needs.
Cadence: how to preserve teams from exhaustion?
As you can imagine, it’s impossible to stay on a Mission forever or move to the next without a break. Many of you are no doubt all too familiar with an endless series of sprints, and this is something we really wanted to avoid. At the heart of our model is the cadence of our three delivery lanes. Exhausting or discouraging our teams is definitely not part of our Culture, and for this reason, Basecamp’s “Shape Up” inspired us to create cool down periods. Two weeks to enable the teams to regain their footing in “real life”, lower their adrenaline level and work on the Ground Lanes. This period is also used to prepare the work for the next cycle.
… And what if a Mission fails?
Nobody goes undefeated all the time. Although winning together is the goal, and because we are committed to trying new things and fighting challenging battles, we know that failure will be part of the journey. So instead of apprehending it, we try to acknowledge it early on so we can understand what went wrong. Was it because the Discovery Mission didn’t fulfill its role? Was it because the scope was too ambitious? Or because the team didn’t synchronize enough? In any case, failure scenarios are documented to prevent similar issues arising in the future.
Our Roles
After creating new delivery lanes, we had to create new roles to fit this operational model. The notions of trust and efficiency are central to us, as well as the desire to make the best use of everyone’s skills. I would encourage you to read Lara Hogan’s excellent book “Resilient Management” from which we adopted the role of DRI (Direct Responsible Individuals) — an expert 100% responsible for their scope, for taking all the decisions — and the ADL (Accountable Delivery Leads) — a project manager accountable for the team or Mission success. We also created the Craft Leads, in charge of leading an expertise team, and the Stack Owner, a technical expert responsible for a component or system. In a nutshell, DRIs are responsible for the “what”, ADLs for the “when” and Crafts and Stack Owners for the “how” we do things.
What about the “why”? We prioritize the right Missions in a dedicated backlog and manage staffing to systematically deliver on Dailymotion’s strategic roadmap. But this will be explained in a future article.
DRIs (Direct Responsible Individuals)
Some of you may have already encountered this term in software literature; DRIs are responsible for taking decisions on a piece — or several pieces — of our product ecosystem, including technical migration. At Dailymotion, they are the go-to person for a clearly identifiable piece of scope (ex: Homepage, Partner Space, Search…). A DRI can be any employee with the most relevant expertise or knowledge and they remain responsible for their scope for as long as it exists. Before the start of each Delivery Mission, they are responsible for defining a Mission Pitch according to Dailymotion’s strategy. This document enables the company to prioritize the project and the resources that need to be allocated. Throughout the Mission, the DRI takes all decisions and arbitrates within their scope, and by doing so, assumes the decisional pressure thus allowing the operational teams to stay focused on their job. We don’t have technical DRIs, instead we created the role of Stack Owner, as explained below.
ADLs (Accountable Delivery Leads)
ADLs (project managers) are accountable for decisions to be taken within a Ground Lane or a Delivery Mission about priority, scope and dependencies. They ensure the Mission is completed within the six-week time frame by applying agile methodology. Their core mission is to mitigate risk and make sure the ambition of the DRI is consistent with the capacity of the team.
If a DRI wants to send us to Mars, his ADL can remind him that they only have enough fuel to go to the Moon.
Craft Leads
Craft Leads (VPs, Engineering Directors or Program Directors) lead a set of talent within an expertise at Dailymotion. Their job is to develop the talents within the team and beyond when they are assigned to Missions.
Stack Owners
Stack Owners (software or infrastructure engineers) are responsible for either a specific layer in a stack, a horizontal component or a cross-service system. They evangelize the code used in the stack, the process, the documentation and the maintenance of their part of the stack by providing support on the Ground Lanes or when they’re staffed on a Mission.
Regaining a sense of working together
Thanks to this project management shift, every one of our employees is now working in what we believe is their best position. By relieving the operational teams of the decisional pressure, we allow team members to fulfill their dedicated role. No longer having a pilot trying to fly 7 space-shuttles at a time. No longer seeing the pilot trying to fix a reactor when a technician could do it better and faster. As every employee knows exactly what to do and how to do it, the meaning of working as an individual resurfaces, as well as the meaning of working together for Dailymotion’s global mission.
There’s nothing worse than no goal or no achievable goal… and there’s nothing better than achieving a goal as a team
After two completed cycles — and already more than 20 successful Delivery Mission landings — we are beginning to see the benefits of this new organization: clearer roles and ownership within the company, a healthier and more positive pressure on the teams, less intermediaries and bottlenecks, less friction between the Crafts… And as we no longer prioritize resources or projects per team, we now make our strategic choices at a company level, reinforcing our global mission backlog.
We know that no model is perfect, and that we’ll have to continue to iterate on this together again and again — and at some point when we will undoubtedly fail together, we will learn how to further improve this model, as these steps are as key as the deployment of the approach itself (this will probably be the subject of future articles). I truly believe that there’s no such thing as a perfect organization. It’s all about finding the optimal model that best fits your company needs at a specific moment in time.
