How Dropbox Sources, Scales and Ships Its Best Product Ideas

How Dropbox Sources, Scales and Ships Its Best Product Ideas

Dropbox's product engineering leader shares how a hybrid product development approach propels its best ideas.

At some point in the journey, most climbers — and leaders — take in the three rules of mountaineering: it's always farther than it looks; it's always taller than it looks; it's always harder than it looks. A year after joining Dropbox to run product engineering, Tido Carriero found himself looking up from base camp at his Mt. Everest: Dropbox for Business. In order to accomplish this difficult engineering feat, Carriero needed to assemble his best engineers for the big climb: entirely re-architecting the Dropbox platform to support multiple accounts.

His team rallied and successfully shipped the product, but not without sacrifice: the opportunity cost of new ideas from two dozen veteran engineers. This trade-off would paralyze some less experienced managers, but Carriero has learned to take the long view. As an engineering leader at both Facebook and Dropbox, he’s seen how the most successful startups not only identify the products they need to build, but also apply different types of development approaches to usher products into reality.

At First Round’s CTO Summit, Carriero shared the power of a hybrid product development process and the actions young companies can take to navigate it well. Every entrepreneur who wants to know how the best product ideas are sourced, grown and funded stands to benefit from his thoughts.

Aim for a Hybrid Approach to Product Development

To source ideas effectively and grow them into products, Dropbox uses a hybrid product development approach. The method strikes a balance between a bottom-up approach, in which engineers autonomously pursue new product development, and a top-down approach, wherein an engineering or product leader initiates all new product roadmaps. A company doesn’t just institute a hybrid method; it arrives at equilibrium over time by drawing the best from both a bottom-up and top-down approach.

As companies grow, they tend to try on both bottom-up and top-down methods. And they can benefit from recognizing the advantages of each approach. Here’s how Dropbox experienced — and retained the strengths from — each approach to arrive at a hybrid product development best practice.

When you’re early-stage, begin with bottom-up development.

A startup typically develops from a small team of engineers that is heavily — if not directly — influenced by the independent spirit of the founders. So it’s not only wise but also natural to build from the ground floor. “Let's rewind to the very early days of Dropbox, when we relied on bottom-up planning,” says Carriero. “I think every startup begins here. Our server engineer was our co-founder, Arash [Ferdowsi], and our client engineer was our other co-founder, Drew [Houston].”

Carriero is keenly aware of the product development philosophy that preceded him. All of Dropbox’s early planning was very autonomous, even as more people were added to the team. “This type of planning worked well during the early years. There was a freewheeling spirit and the instinct to immediately assess and address whatever needed to get done to serve a rapidly evolving business.”

Shift to a top-down approach when projects are step-changes.

It’s best to switch to the top-down development approach on the eve of a massive endeavor that requires most of your team to accomplish. “Not long after I joined our team, I noticed that with all the projects in motion, it was becoming hard for our group to tackle the biggest challenges,” says Carriero. “What we gained in our diversity of ambition, we lost in collective attention.”

The multi-account project was the behemoth that restructured Carriero’s team and shifted its planning philosophy to a top-down approach. With the majority of users using Dropbox at both home and work, it was critical to make sure that all accounts — whether business or personal — synced and worked concurrently across all platforms. Dropbox already had a sophisticated infrastructure with Desktop, iOS, Android, and Web clients, which made this endeavor particularly complex.

“The multi-account project entailed supporting more than one user at once,” Carriero says. “This project was the straw that broke the camel's back for bottom-up planning. It seemed impossible that we were going to get a team of 25 engineers, which is how many it took to re-architect every single platform to support multiple accounts.”

In my first two weeks, many warned me that this project was too large to tackle. It was huge, but not as big as my smile when we finally shipped.

All that said, drive toward a hybrid development approach as soon as possible.

Once a company is more mature, a hybrid approach — between a bottom-up and top-down development method — is optimal. Splitting the difference between these two strategies is a way to accomplish ambitious all-hands projects and continuously nurture new ideas from individuals. The best startups adopt this approach early on.

Carriero notes the fine balance: “A top-down approach allowed us to execute on much larger projects. However, when leading a team, you can limit your 10x engineers by over-planning them. You miss out on these game-changing ideas from the people who really think about technology a little bit differently.”

Since the multi-account project, Dropbox has embarked on a hybrid approach to product development planning. “Ultimately, we still do most of our planning and the majority of the roadmap setting top-down. We found that that's the only way we can get the focus levels that we need to get. Otherwise, you end up expanding the scope of projects. Yet we also take different tacks to encourage bottom-up ideas from everyone — and I mean all Dropboxers — across the company. These efforts help us keep an open environment for new ideas.”

Embrace Dualities to Keep Your Balance

When combining elements into a hybrid approach, be careful not to tip further in one direction than the other.

Today, Carriero leans on three frameworks to divvy up resources and engineers for any project, from a new initiative to a core business product. These “rules” serve as checks and balances to ensure his team is investing its time and skills to suit the scope of each project. Akin to Pandora’s product prioritization system, Carriero has identified the following frameworks to keep your product planning balanced and on the right track.

Lightweight and heavyweight product reviews. For Dropbox, putting a lightweight idea evaluation before a heavyweight product review has been a game-changer. “We used to only have a full product review process,” says Carriero. “It required writing out every edge case, mocking up each solution, and getting tech leads to determine exact product timelines. Those steps are important to sound product development, but made generating and evaluating ideas at a high level a challenge.”

With this friction, inbound product ideas across the organization slowed to a trickle. So Dropbox created a lightweight — or “Phase 0” review. “The ‘Phase 0’ review improved the entire process. It involves completing a simple one-page template that defines a problem and explains why it's worth solving. Any PM or Dropbox employee can put a ‘Phase 0’ review together in an hour or two, fire it off to a mailing list and schedule time to present the idea. ”

At first, put all ideas in zero gravity. It creates a special atmosphere where anyone can contribute a great idea.

Dropbox's product review system is designed to be incredibly lightweight and encourage lots of product ideas. All ideas start in “Phase 0,” in which someone outlines a challenge and provides evidence that the problem is worth solving. “Phase 0” is incredibly democratic — anyone so inclined can propose a “Phase 0” pitch and present it to Dropbox's product leadership team.

After the product leadership team agrees that it’s a problem worth solving, a Dropbox Product Manager will move through increasingly rigorous phases of product review. “Phase 1” is designed to discuss potential approaches to solving the problem and is often accompanied by tech specs from Engineering and mocks from Design.

Finally, “Phase 2” is more of a traditional product review that locks down all of the product specs in full detail. These sequential phases help move the product along as efficiently as possible, while keeping people from focusing on minute details before the time is right.

Seeds and saplings. After the full review process, Carriero encourages his team to categorize each potential product as an investment, much as a venture capitalist might with portfolio companies. At Dropbox, Carriero applies this concept through a framework called “The Magic Forest.”

“There are seeds and saplings. Seeds are very-early-stage product ideas that we think of as having series A funding. They are prototypes and hypotheses that we want to test. When they have product-market fit, these projects become saplings,” explains Carriero. “These projects are now products that we seek to grow and bring real scale. Lastly, there are trees and roots. Trees are our core products, and roots are the infrastructure that supports them.”

Arboreal imagery aside, the goal is to clearly define runway for each idea that corresponds to its stage of growth. “We give every idea time-bounded and resource-bounded constraints to derive and drive a metric or new discovery.” Without stage-specific constraints, there’s a risk of wasting funds, time and expertise -- often even if you’re aware the project is failing. “It’s very easy to get too attached to your seeds and saplings, and not chop them down. That’s the dangerous downside of funding new projects. Put clear bounds on each project early and you’ll know whether to grow or get rid of an idea.”

Governance and resonance. From their earliest development, products, like startups, should have a board of advisors. “The other thing we've done is we've implemented a board of directors for each of these project groups,” says Carriero. “So they have one consistent set of leaders at the company, much like a startup board.” The teams building the product must report on metrics and solicit feedback from its board. This level of oversight has been really useful in scaling up these little projects within a larger company.

Regardless of informed product boards, Carriero has seen how often employees misunderstand or forget the purpose of early-stage projects. “People know at a high level that there must be a balance of product innovation and more iterative and incremental development. Easy frameworks such as ‘seeds and saplings’ help employees understand investments in projects that aren’t their own but which are a key part of their company’s future.”

Constantly Hack Your Hack Week

In addition to these frameworks and a keen sense of balance, Carriero has seen the value of hackathons for engineers, designers and product managers to jumpstart innovative product development. But startups shouldn’t expect to generate novel ideas with the same stagnant process. Carriero suggests four best practices to design a better hack week that stays fresh and generative.

Commit to a full hack week every year. Don’t bother with Hack Days. Make sure that there’s enough time to really explore an idea and test it seriously. “At Dropbox, we do a full five-day Hack Week. We start on a Monday morning and continue until Friday,” says Carriero. “It’s a beautiful sight to see engineers, designers and PMs given unstructured time to dream up something really awesome to build.”

Experiment with themes. Dropbox has started giving themes for Hack Weeks. Topics have included “Collaboration,” “Growth,” and “Dropbox for Business,” which focused on building out an existing product. “We'll often give high-level themes, but we don't ever assign specific projects because that can constrain the ideas that are generated, and that's exactly what we want to prevent.”

Think Olympics, not summer camp. “We take the projects that come out of Hack Week very seriously,” say Carriero. “Employees at the company vote on who wins awards. For example, Dropbox hack weeks can feature an event called Haxpo (short for “Hacks Exposition”), which is a science fair where everyone puts together booths, posters and demos. Employees from throughout the company speak with presenters about their projects and vote for their favorites. “My favorite award is the #shipit Award. That's the project that's most ready to be shipped right away -- and often it does.”

Hack to the future. All projects that earn a certain number of votes are presented to the Engineering, Product, and Design (EPD) leadership team. Presenters then pitch the project to get it into the long-term road map. “By finishing each Hack Week with a pitch to EPD leadership, we can ensure that all of the efforts during the week aren’t left behind when routine work resumes.”

During his time at Dropbox, Carriero has witnessed some impressive results from Hack Weeks. “Many of our critical projects originated at a Hack Week. My favorite win came from an amazing OSX reverse-engineer hacker named Maxime (nicknamed ‘Haxime’). I’ve never seen the stars align like this: here was a 10x engineer and an idea as crazy as he was.”

Max had the idea of inserting the Dropbox user interface (UI) into Microsoft Office without installing any plug-ins. “While Max didn’t get his idea fully working during the Hack Week, he did manage to get the UI injected in a way that no one thought was possible. That was the birth of what we call ‘The Badge,’ which Dropbox released in early beta to its Dropbox for Business customers in December 2014.”

The best results from a Hack Week may not be finished products, but rather momentum toward the improbable.

Carriero recalls the origin of the multi-account project, “For years, we weren’t sure if the project was even feasible given the technical challenges and platform changes involved. For those who thought it could be done, the discussion turned to which approach might work in under six months.”

During a Hack Week, a desktop-client engineer had had enough talk. “Screw this,” he said, “let’s just get together and do it.” A team of two engineers and a PM assembled, and by the end of the week they had built the first prototype of a multi-account product on the desktop client. “It was awesome,” recalls Carriero. “It propelled the project forward in a way that it wouldn't have gone otherwise.”

Whether it’s through a Hack Week or a lightweight product review, Carriero knows that the best product ideas never travel a typical path to being shipped. It’s fine to take a top-down or bottom-up approach some of the time, but balance is optimal, and can be created only through leaders making conscious and concrete choices.

Ultimately, every idea is an investment. “Idea. Hypothesis. Prototype. Product. Make it very clear which metrics and results you plan to see at each stage of an idea’s growth. Set expectations so it’s clear to everyone which ‘investments’ receive ‘funding’ and why. This is very difficult, but totally worth the extra effort up front,” says Carriero. “As you fund new product innovation, everyone needs to have the right expectations for and investment in what’s next.”