Companies get big in remarkably few ways. If you ask companies with >$100M in revenue how they got there, you’ll never hear an answer like “we did these 20 things pretty well and it all added up”. Instead, you’ll hear “we did one or two things, and we did them exceptionally well”.
Why is that the case? There is a central metaphor that can both explain what is happening here, and guide growth teams toward where they should focus.
It’s the concept of a perpetual motion machine: something that can do work indefinitely without an energy source. In the physical sense that’s impossible, of course, due to the problematic first law of thermodynamics. But at the core of all great growth stories, there is at least one perpetual motion machine: a loop in which the inputs are directly turned into a greater output of the same form without the need for external stimulus.
The (short) list of perpetual growth machines
Perpetual growth machines all follow the same pattern: they start with some initial asset, which attracts users, who through some conversion process generate more of the original asset.
I count only four types of machine that describe the vast majority of successful growth companies. That number is low because there are very few types of assets that are ubiquitous enough to attract users at scale: marketing and sales, word of mouth, content, and physical space.
1. Marketing and sales machines
The first type of machine is perhaps the most common and widely understood. It starts with marketing and/or sales, which is used to acquire customers, some % of whom drive revenue, which can then be reinvested in more marketing or more salespeople.
There are many B2C examples with large paid marketing budgets (eBay, Credit Karma) and many B2B examples with large sales teams (Salesforce, Twilio).
2. User machines
Here, the asset which attracts new users is the user itself. Another term for this kind of machine is virality, and it can often drive explosive growth, as in the case of WhatsApp or Twitter.
This virality can happen indirectly (i.e. word of mouth, which has been around forever) or more recently directly in the product, as in the case of LinkedIn users directly inviting their connections to the platform.
While paid customer referrals (e.g. “both you and your friend get $20”) can also be effective, I wouldn’t include them here, because that operates more like a marketing machine than a viral machine. You just happen to be paying your own customers instead of ad networks, and unlike truly viral products, you’ll still need deep monetization to keep paying out for referrals. In this post, Andrew Chen explains why true virality is getting harder and more companies are relying on incentivized referrals.
3. Content machines
This is essentially a specialized form of the user machine. Instead of users directly acquiring more users, they instead create content (e.g. Yelp reviews) that is then surfaced to potential new users, most powerfully via organic search.
Examples of this kind of machine are SEO powerhouses: Pinterest, TripAdvisor, Houzz.
4. Physical space machines
This last kind of machine is less familiar in the tech world, but I saw them all the time when I was an investor in mostly offline consumer businesses at TSG. The initial asset is physical space (e.g. retailer shelf space, or a retail location itself), which attracts customers, who pay you and allow you to invest in more space.
One example is the retail location with great unit economics (Starbucks, SoulCycle), which drives profit that can be used to fund the next location. Another is a consumer product that flies off of retailer shelves (Vitamin Water, Halo Top), funding the increased distribution, retail slotting fees, and other expenses needed to get more shelf space.
Implications for growth teams
What does this mean for growth teams, or anyone thinking about how to drive sustainable growth? A few things:
One machine is likely to drive almost all of your growth
Yes, multiple machines running simultaneously can be incredibly powerful. Amazon’s customers are not only powerful advocates working on its behalf, but it has the depth of monetization to support massive paid marketing. LinkedIn not only has direct viral growth but a powerful content production machine that gives it a broad SEO footprint.
However, for most companies, especially in earlier stages, one machine drives ~80% of all new customer growth. That is because products are naturally built for a specific kind of machine. Marketing/sales machines require significant depth of monetization. Virality requires that users both can share with lots of people (the product is relevant to a high % of the population) and that they want to (it makes them look good, the product gets better when more people use it, etc.)
As a result, once a company has exhausted the unscalable methods of getting its initial user base, the best path to a big business is finding and exploiting one core growth machine. In this phase, most of a growth team’s time should be dedicated to making this machine run exceptionally well.
Efficiency of machine = rate of growth
What do we mean by making the machine run well? We can think about it on two dimensions:
Leverage – the ratio of output to input after a full run of the machine. For marketing/sales, this is the LTV/CAC ratio, for a viral machine it is k-factor, or the number of incremental users each new user drivers.
If this ratio is <1, you don’t have a perpetual growth machine at all, and without outside stimulus (e.g. cash or press), your growth rate will trend toward 0. And to really have something interesting on your hands, this ratio needs to be significantly higher than 1. For example, if you’re running a marketing machine, it probably needs to operate at leverage of > 3:1 LTV/CAC unless you want to do violence to your P&L.
Velocity – the speed at which the machine generates output. Independent of leverage, the speed at which you can re-invest in another turn of your machine is a major determinant of growth rate. Consider two subscription services, both with a customer acquisition cost of $100 and average annual profit per customer of $250. But company A bills annually up-front, company B bills monthly. They will grow at very different rates!
Metrics which combine both leverage and velocity are a great measure of how quickly a company can grow. This is why, for example, the payback period of sales and marketing spend (which accounts for both leverage and velocity) is usually more telling than raw LTV/CAC ratio (which accounts only for leverage).
Efficiency of machine = a good way to prioritize growth work
Most of what we think about as part of the growth and marketing functions can be linked to efficiency of machine. Ad buying efficiency improves the conversion of dollars to your initial asset. Conversion rate optimization increases the rate at which an asset generates new customers. Email and push notifications can be used to encourage users to create more content (leave a review!) or spend more money (buy these shoes!). Even less performance-driven activities like brand building can be linked, e.g. by providing a better hook in the customer’s mind to enhance word of mouth.
And given that the impact of most of these activities on machine efficiency can be expressed quantitatively, you can build a model to figure out which pieces are the highest leverage.
Breakout growth is often predicated on a wildly more efficient machine
The above discussion is mostly about incremental improvements to efficiency. But many of the biggest growth success stories are something else entirely – they have a unique wedge on the market that results in a vastly more efficient machine than incumbents are using.
In the payments processing space, Square brought a self-serve product coupled to a marketing-driven machine to a fight where everyone else brought a manual setup product and a sales-driven machine. As a result, Square had both higher leverage (more efficient use of the initial asset) and higher velocity (shorter time from outreach to revenue). This opened up the SMB market to them in an unparalleled way, while incumbents were not efficient enough to target customers with lower LTVs.
Or take the example of WeWork. Clearly one of the core tenants of their business model is to get more leverage out of square footage by chopping it up into smaller pieces to achieve higher premiums and better utilization. But they also increase velocity by getting landlords to share in up-front buildout costs in exchange for profit sharing down the line. As a result of both of these advantages – at least according to a 2015 pitch deck – they have a 7-month payback period on new units vs. an industry average of 22 months.
Cash can make your machine run faster, but it can’t buy you a new one
There are many ways an infusion of cash can help a machine run faster to drive growth. Obviously, it can allow a company to invest in marketing and sales ahead of payback. It can help a company more quickly expand into new markets or open new locations. It can hire a team which can then go work on the efficiency of your machine.
But cash is never a substitute for a growth machine that works on its own, and without one the cash will eventually dry up. That’s why smart investors, (at least at series A+) look for machines that already have signs that they could work well – good payback periods, high customer satisfaction, etc. – and would benefit from some extra fuel to get the machine turning faster.
When the machine is running, you’ll feel it
At the start of this great series of articles, Brian Balfour compares what he calls Smooth Sailors (companies for which growth seems to come easy) vs. Tugboats (companies that always seem to be scraping for more growth).
That analogy certainly resonates with my experience working with consumer tech companies. Growing at venture-backed rates is never “easy”, but when one of the above machines is really working, it certainly feels more like the case of the Smooth Sailor. You can mess up a lot of things, but the company still grows. And even more importantly, there is a repeatable process with well-understood inputs and outputs that surfaces high leverage opportunities to improve that growth trajectory.
So rather than a scattershot approach of testing everything, try articulating your perpetual growth machine and each of the steps that determine its efficiency, and get to work on tuning your machine.