Should This, Not Can This, Product be Built?
(Part 2 of 2) The counterintuitive business of building startups
Startup-building is hard. New ventures operate in uncertainty. So it is inevitable that there’ll be waste but there are many levels of waste. The trick is to find the most acceptable level of waste by following a lean approach.
No one told you this but your job as an intra- or entrepreneur is not to ensure timely launch of products and services and timely delivery of new features and timely and growing new revenue streams. It is to first ensure that your assumptions about the market are correct and whatever model is in your head is worth bringing to life at all.
Until you validate your assumptions you will have little to show as output except things you and the team have learned—insights, if you may. If that’s a metric your investors or the management are queasy about, it is plausible that you will be prodded to show real results. Once you start optimizing for it—and it is common to—it is also plausible that you will have built products or invested in distribution for the same or bought expensive tech that no one wants.
To avoid getting sucked into this cycle, Eric Ries formalized the lean startup methodology. Just like the agile methodology breaks down the software development process into manageable cycles, the lean startup system does something similar for startup-building.
But there are misconceptions along the way. These misconceptions have survived because starting up is counterintuitive. It is counterintuitive because for a lot of us the reference point is the certain way things get done in established companies whereas the default setting in a startup is uncertainty.
To borrow Tim Urban’s analogy, a startup is a place for chefs. A corporation is a place for cooks. Chefs are in the pursuit of the next great recipe that no one has created yet. Cooks spend almost all their time wondering how to turn an old recipe into a dish quicker, cheaper, better.
One works in a lab, the other in a kitchen. There’s a big difference between the two.
A lesser known truth about startup-building
Most think that a singular way you can goof up while starting up is by not pinning down the unmet customer needs your product is going to meet. That is not true. Once it’s clear that your offering does a job for a set of customers that they are willing to pay for, there’s almost always a set of linked choices that you have to make around how you will deliver customer value and turn a profit 🥶😧
To find product-market fit, you have to get your entire business model right and that has dimensions other than just proving there’s value for a set of customers. This is an often-ignored aspect about venture-building. Undergirding this is the simple fact that launching a new venture is learning to wrestle with uncertainty.
I continue to learn this lesson. Like a collector on the lookout for memorabilia, I’ve gathered odds and ends as and when. The set of lessons isn’t complete, probably will never be. But if my lessons help you at all, I’ve done my job here.
Learning the lucky way
I work in science communication. I help scientists around the globe get funding for their innovative ideas, publish their research in the best scientific journals, and take their published work to new audiences for wider adoption and application. Within this setup, I’m an intrapreneur.
Many years ago, when I was new to this funny business of launching new things, my boss and I were looking for a black box—a secret-sauce algorithm that would help scientists find the most appropriate journal based on the specifics of their research.
We assumed that this particular job was tricky for scientists. With science becoming increasingly un-siloed from traditional disciplines, we had a hunch that the trickiness quotient would only shoot up. There were more than twenty-five-thousand journals worldwide and that number continued to rise year on year. We figured this was a job worth doing for our scientist friends.
We went hunting for an algorithm. We found a Silicon Valley company that used AI (back when AI was not quite the buzzword it is today) to scan a research manuscript and suggest a ranked list of journal matches for it.
We wanted their proprietary IP and we wanted no one else to be able to lay their hands on it. Talks around a potential acquisition, escrow, and the works ensued. Weeks and months passed. Finally, we ended up buying a three-year license for use of the software. We launched a tool to help scientists select the right journal for their manuscripts. It never took off.
We got off lucky. We could’ve paid the millions for an acquisition and be forced to cling on to the project to justify that kind of money. We could have built an entire kitchen, bought the fanciest equipment, hired a bunch of cooks, trained a school of waiters, printed out new menus, and spruced up the restaurant without ever checking if there were diners. 🚩🚩🚩
The prediction we had made about the market had turned out wrong. I had learned my lesson, or so I thought. Turns out there were a few more in the offing.
Understanding business models for startups
Just like an architect’s model shows the key structural elements in a proposed design, a business model captures the key components of a business that will help it stand on its own. It identifies the key assumptions that will sustain the business. These assumptions, contrary to received wisdom, are not limited to figuring out if the business would bring value to a defined set of customers. They cover all the various tests a business needs to pass to get on the path to self-sufficiency.
🔺Customer Value Proposition - Will the business create value for customers?
🔺Go-To-Market - How will the said customers discover the product/service?
🔺Technology & Operations Management - How will you build the product?
🔺Profit Formula - What profit will the business make per unit of product sold?
These components are the labels of Professor Thomas Eisenmann from Harvard Business School where they are taught as part of a class on launching technology ventures.
If you’re groaning about the need to nerd out on business models instead of working on your fledgling venture, I hear you. Yet, look closely 🔎 and you’ll see that the four components each correspond to:
👉Strategy ➡ Customer value proposition
👉Operations ➡ Technology and operations management
👉Marketing ➡ Go-to-market, and
👉Finance ➡ Price formula
These functions are the load-bearing parts of any business. Whether you are an intra- or entrepreneur, you need to be on firm ground in each of these domains. But let me not throw any more theory at you. Let me share with you how I learned this lesson.
Learning the hard way
Scientists are always looking for the best vehicle to take their research places. This vehicle is not always the same. It depends on the cargo to be ferried. If the scientific cargo is valuable it needs a certain class of vehicles (the top scientific journals in the field). If the cargo needs to reach a diverse audience, it needs another class (multidisciplinary journals). And so on.
No two scientific loads are the same. Each has a different and unique use. The cargo, unlike what you know about it, is not a commodity. On the average, transport companies refuse half the ferrying requests that come to them. And the cargo owners? Their careers depend on their ability to find the right transport company. They can be rather persistent.
It pays both the cargo owner (the scientist) and the transport company (the journal publisher) to find themselves a good match to be each successful. At the time I’m talking about, there was no better way for this matchmaking to happen than a cargo owner checking with transport companies, one at a time based on their judgment, if they would ferry their scientific load. It cost the transport company non-trivial effort to check every item of cargo. It was inefficient and becoming even more so every passing year because the volume of scientific load was on the rise.
I came upon an idea to smooth this wrinkle in the scientific world. I imagined a two-sided marketplace that would match cargo owners with transport companies, without needing either party to commit time or effort into it. A matching engine would make a fingerprint of each new scientific manuscript, do the same for each scientific journal there was, and then—ta da!—propose a match between the two.
It would then come down to one of two outcomes—both parties agreed to a transaction or they passed the opportunity and asked for another match, courtesy of the engine.
This time round, we had the matchmaking tech. Having learned from the earlier algorithm experience, we turned our attention to customers. We spoke to customers on both sides, checked for any prior patents for the idea, and then filed for and got a patent for my idea. At ease about the value proposition of the venture, we committed to it.
To make a long story short, here’s a sample of what we learned over the next 12-18 months, as we tried to cold-start an article marketplace. I’ve tagged it below by the relevant business-model component:
🚩Go-to-market
Cold-emailing scientists to cold-start the article marketplace didn’t work. The idea of an algorithm doing the job for them didn’t play to their fantasies.
Getting journals and publishers to join the marketplace was hard too. They were cagey about sharing their content for fingerprint generation, worried about being tricked by fraud science.
You may wonder why am I putting these problems under GTM and not under Customer Value Proposition? Well, that’s because most customers appraised our idea well. But when the time came to join the marketplace, they hesitated. Demand generation proved much harder than predicted. In a way, competition would’ve cured doubt. Because the idea was so novel, it gave them cold feet. They stayed with the status quo.
🚩Operations Management
Because we couldn’t get the scientists to join in on their own, we had to try picking up their intellectual property from wherever it sat.
Scraping preprint servers (online repositories where early versions of research papers are hosted for free before they’re published and legitimized) was harder than anticipated. Manuscripts were structured in a variety of ways that made it hard for the fingerprint engine to come up with an accurate fingerprint for each article. Imagine trying to read a poem and an essay the same way to understand what each means.
The fingerprint engine worked best on file format X and majority of the manuscripts were available in format Y. There was almost no file-format converter that accurately reproduced the content.
🚩Profit Formula
Securing partnerships with preprint servers that allowed us to scrape content raised fixed costs. Content was not commodity and there was just so much content and so many preprint servers.
The path to making money was fuzzy. Should we absorb the costs and offer the matcher free to both sides of the marketplace? What was the right price? Should we open up the platform to third-party developers with services to offer on both sides and take a cut from the developers? The questions came thick and fast.
It seemed impossible to fully validate our business model until we reached a critical mass of platform users, and we could not put off monetization until that point because it was too expensive to just keep things running.
A fun coda
A startup may have to eventually invest copious capital in customer acquisition or operational infrastructure and that’s not a sign of bad business-model management. But the timing of such an infusion is telling. It makes sense to pump in men, money, and machines only after the business model has been validated through repeated testing.
Here are some key insights behind some famous businesses. I wouldn’t go so far as to call them testable hypotheses because I don’t know these businesses any better from where I am. But these are, I believe, the non-obvious theories that these businesses formed early on that put them on the path to success. I must mention that these theories made sense for the time and the reality behind these theories may have changed several years on.
SpaceX🚀👩🚀
Problem: How can we dramatically lower the cost of space travel?
Hypothesis: A rocket is made of aluminum, titanium, copper, carbon fiber. The raw material cost of all components is 2% of what a rocket costs, so if we can approximate bringing down the costs of building a rocket to the cost of the raw materials, space travel (and by extension, space colonization) would become orders-of-magnitude more affordable.
Tesla🔌🚗
Problem: How can we make long-range automobile travel sustainable?
Hypothesis i: Barring a small number of environmentalists, people don’t care about emissions. But a car that looks slick, has a heap of cargo space, and can do 0-60 mph in 4 seconds is irresistible.
Hypothesis ii: Most of the cost of batteries today comes from the cost of middlemen, not raw materials. Cutting out middlemen could make batteries cheaper by X% and EVs by Y% in Z time.
Spotify🎵▶
Problem/Question: How can we make customers out of casual listeners of music?
Hypothesis: Casual listeners don’t prefer music by genre so much as they like to listen to music if it accompanies the occasion/use case (gymming, driving, walking). Given an equal choice between traditional musical genres and use-case-based playlists, X% of the casual-listener base will opt for the latter.
If this list piques your interest, I suggest you sit down with a pen and paper and do the same for the idea in your head you’re looking to launch or the project you’re already immersed in.
For Part I of this two-part series, go here.
If your experience resonates with my observations or runs counter to them, post a comment and tell me how. You will help me improve my current mental model.
Until next week…👋