Paul Graham's wrote "One of the most common types of advice we give at Y Combinator is to do things that don't scale" in his influential post from 2013. He published it at a time when many startup founders in his program had overvalued the mere existence of their products and services, leading to torched funding, missed opportunities, and a lack of customers. A business's survival can't be attributed to only the unscalable work, but it's pretty clear that the majority of tech's success stories include such a component.
And yet, Paul's advice is limited to the context of his work at Y Combinator. He is speaking to small teams with small cash reserves who must understand the realities of bringing new products to market. It's amazing how quickly that this advice is disqualified once business leaders watch their revenue growth explode. In fact, doing things that don't scale is precisely what keeps businesses alive long enough to break their inevitable growth plateau.
Join me as I explore some of the reasons why escaping the first growth plateau is so difficult and, despite the challenges, why I believe that Paul Graham's advice to "do things that don't scale" provides a perfect starting point for companies looking to expand their product roster.
How scale creates lock-in
Writing policies, hiring for highly-focused roles, and building digital systems in place of human interaction are all requirements for handling the pressure created by scale. The focus shifts from developing a product which provides consistent value to distributing and configuring that product to as many new customers as possible. Product teams still have a roadmap to execute on, but they do so in between rolling out reliability improvements like fixing bugs, upgrading systems, and paying down tech debt.
At some point, executive leadership recognizes the need to expand the product roster. Sometimes the right answer (and often the safest answer) is to acquire smaller companies with related products and promising traction, but an option with far more potential upside is to roll the dice and spin out a new team to lead product development. This new team will exist in an environment very different from when the company was a baby.
Long gone are the days where the CEO, lead developer, and new sales rep sat next to each other in an office the size of a closet. Companies growing to hundreds of employees are eventually forced to tame the chaos by setting up organizational structures like departments. Since teams are now separated by business function and layers of management, everyone is at the mercy of corporate life: communication silos, department cultures, budget spend reporting, performance reviews, internal politics, and arbitrary deadlines.
An environment full of rigid structures makes it difficult to bring new products and services to market with the same caring touch as before. At best, new product development wades through the internal controls at a snail's pace, and at worst, the innovation teams never break away from the established corporate norms. It takes a lot of coordination to test prototypes, talk to potential customers, and develop software without stepping on the toes of leadership in other departments.
Cooking up innovative (or at the very least, competitive) new products from inside an established company is not impossible, but the people doing the cooking need to work in an environment which allows them to take risks and make mistakes. Internal innovation is, for the most part, a shift in mindset.
"Of course we're innovating. We have an innovation lab."
As someone who has worked inside one myself, I am painfully aware of the way innovation labs miss the point. Most of these labs exist in a corporate environment which celebrates the spirit of innovation but balks at all of the opportunities to take a risk on a team who gets traction.
It is very rare that you find a leadership team that has thought through the implications of opening the lab. The first symptom of this is the lack of a clear innovation strategy. (source)
It feels as though these labs are budgeted and built by leadership who hasn't changed their mindset about innovation. Spaces are full of bean bag chairs and whiteboards but the entrepreneurs within have no cultural ties with the parent company, which is usually unprepared to absorb a team, tech, or both. Since the innovation lab was a tactic to begin with, a few years with negative balance sheets and no acquisitions makes it easy for leadership to conclude that enabling entrepreneurs is the wrong choice.
Build like a startup, live like an enterprise
Internal startup-lookalikes made up of talented full-time employees are worth the gamble if everyone involved has the same expectations.
Entrepreneurial employees, despite being surrounded by the symptoms of scale, are lucky to get their hands dirty with a hundred-million-dollar logo hanging over their head. Established corporates have tons of cash and miles of reach. Outside of HQ, most startup founders usually have to start from zero and do the hard stuff like recruit and fundraise on their own.
Having access to all of those resources comes with a caveat: the parent company is now the board. Just like how investors need to understand how their money is being spent, executive stakeholders need to know the same.
However, executive stakeholders need to have an exit plan. Venture capitalists understand that they are in the game until their shares can be exercised. Since the CEO isn't buying a stake in your new internal startup, they better have a handle on what success looks like. Innovating for the sake of serving a market is quite different from innovating to enable a large company to serve a market. Any story where innovation teams are scrapped because their product might cannibalize the parent's cash cow is rooted in a lack of clarity from leadership about what they were expecting from their startup program.
Assuming the innovation strategy is in a healthy state, the second shift in mindset is to accept the reality that any startup-like team is not out there building a genuine, standalone startup. New tech will eventually have to integrate with the rest of the platform (and ideally look on-brand). New products or services have to be related to the parent brand in a way that seems natural. Sales, marketing, tech, and support personnel will need to roll into the larger departments without losing steam despite physical separation. Early prospects need to be marked so that sales and marketing from the mothership doesn't poach. It's a balance between total innovative freedom and staying near the established norms.
And this is where Paul Graham's advice deserves to be loudest. The looming inevitable future of being absorbed into a larger company's structures puts pressure on the team to prepare for scale starting from day one. Pushing away that feeling as much as possible is the only way to avoid building a dud.
Up close and personal
As a rule of thumb, things that aren't scalable are things that require you to be there, like conducting meetings, doing manual data entry, or going on-site to onboard a customer. In the same spirit, Graham's essay lists recruiting, finding users, obsessively delighting customers, getting feedback, purposely limiting growth, doing consulting, building things from scratch, or not building things at all as unscalable tactics.
The entire goal for internal innovation teams looking to cook up their company's next smash hit product is to answer the right questions. Specifically:
- Is the product addressing an untouched market need or just allowing the company to crush competition in a defined space?
- How similar do the new product's users look when compared to users of the core product?
- Can the new product stand on its own or does it require integration with the parent company's core product to provide enough value?
- What proof exists that the new product will not only delight its users but will make money in a predictable, well-understood manner?
In other words, everything that the startup team should be doing ought to be unscalable.
The parent company, now at the apex of its growth, has established a track record of effective execution; turning assumptions into infrastructure is what they do best. If the internal startup team hands the executive team an exciting story but too many unverified assumptions, budgeting millions of dollars to lock in those assumptions can result in what looks like when Colgate tried to sell frozen meals.
Perhaps this is what a real "innovation lab" looks like; teams handed a short list of constraints are being paid to try new things to solve people's problems in the name of understanding. The more they understand, the less risky it is to give the team more resources to scale up the operation.
Unscalable at scale
As of the time of writing this post, the most "innovative" tech companies have one thing in common: culture. Content from Apple, Netflix, Shopify, Amazon and Google shows how each, in a strikingly similar way, lays the groundwork for innovative behavior.
The recipe, it seems, has most of the same ingredients as when old-school enterprises build successful innovation teams. Employees need freedom to explore and build whatever they feel is right, as long as they stay within certain lines. Failure is expected and accepted as long as the attempts are carefully considered. Emphasis is placed on understanding users, including but not limited to getting face-time and digging into product analytics. Going above and beyond for users in the name of your company can be rewarded with more resources to expand the side-project into a full-time gig.
The benefits extend all the way to the top. Some leadership teams realize that, as growth slows, they have to unlearn some habits which developed as a result of scale. After creating the role of SVP of Customer Success, "HubSpot's CEO, Brian Halligan, [now sits] with the support team every week listening in to calls". Although time-consuming, this behavior keeps executive leadership from losing touch with the market that they serve.
So does every company need to start by writing a Googly culture code to get out of an innovation rut? Of course not. What matters most is that, somewhere in the company, there is an open-ended environment which rewards people for turning customer intimacy into actionable insight.
Paul Graham said it best:
"The unscalable things you have to do to get started are not merely a necessary evil, but change the company permanently for the better."