The Blog

Which business models will fail?

The undoing of lose-money-to-earn hardware models 

Business model innovation is a combination game, linking partners, suppliers, and other actors to create value for everyone in an ecosystem. 

But when things fall apart, failed business models suddenly appear visible when they had formerly been hidden in plain sight. While many hardware-as-a-service models will endure, the most fragile connected devices were those that depended on losing money today hoping for dreamy future cloud revenue. The axe is coming for products and companies that got hardware-as-a-service wrong: 

Alexa: earn-per-use never made it past small talk

The original vision for the voice-activated Alexa product was to sell the device at or below cost,  betting that Amazon customers would increase their use and spend on products and services. 

Alexa made a big splash in 2014, and with excitement about voice as the next level of screen-free interaction. Critics flagged the risks of surveillance and intrusion of privacy. At one point, Amazon filed patents claiming Alexa’s ability to detect coughs or emotional abnormalities based on voice analysis, suggesting ways the device could play an instrumental role in digitally-enabled care. 

But in the end, Amazon’s customers only ended up making small talk with the devices, asking about the weather and music repeatedly, but never getting to know us any better, and almost never using the devices for commerce. Attempts to create brand-based “skills” never materialized into substantial revenue. Alexa devices were merely unprofitable consumer electronics with no substantial cloud value to reap into rewards. 

The company’s Alexa division has been a steadily increasing money pit, losing $5 billion in 2018, and was on track to lose $10 billion this year on Alexa and the devices. These losses were hidden in the upwardly increasing profit and free cash flow days of Amazon during the pandemic, but now with substantial headwinds and exposure, the division has been targeted for a substantial proportion of Amazon’s 10,000 estimated staff cuts. 

Bird Scooters: price-per-ride the wrong unit economic


Bird was the fastest company to earn “Unicorn” status, moving from zero to $1 billion in valuation in six months from its start in September 2017 to a funding round the following May, 2018. Founded by a former Lyft and Uber exec Thomas VanderZaden, the company’s business model was price-per-ride: a fixed fee to unlock the scooter, and a second fee based on a per-mile charge. Customers could pre-load their wallet for friction-free ease-of-use. 

To the company’s first investors, pay-per-use was a familiar pattern: it looked just like software-as-a-service, priced to increase speedy trial and adoption. Indeed, adoption grew swiftly as Bird targeted cities with weaker legislative muscle and aimed for international scale. VanderZaden even attempted to show that Bird’s unit economics were more profitable than those of other car-dependent business models like Uber and Lyft, even though their Special Purpose Acquisition (SPAC) documents showed operating losses – see the tweet below.


But in the end, the framing of the Bird unit economics model was misleading. VanderZaden stepped down earlier this year, and during the most recent earnings call the current CEO admitted to prior revenue misstatements due to pre-loaded wallet balances which never were redeemed, creating phantom revenue. The company is not only at risk of being delisted from the NYSE, they have also concluded that they do not have effective “internal control over financial reporting” and they are swiftly designing and implement new controls.

Bird’s management changed their unit economics from framing pay-per-ride margins to a more meaningful utilization number, or percent of time the scooter is used. Lesson learned: utilization was the unit economic Bird should have used form the start since the scooters live on the balance sheet, not in the cloud. But Bird’s early investors have likely not bothered to learn the lesson, since they made money on the SPAC Initial Public Offering, leaving long-haul public markets investors and employees owning a greatly devalued asset.

Latch: when a door lock is not a platform business model

Latch launched in 2014 by selling connected door locks to residential apartment building owners and charging a monthly subscription fee for building management software. Run by 32-year-old Luke Schoenfelder, the company claims to be an “enterprise SaaS” business model, insisting that the value is in the software, not the hardware.

On top of these two revenue streams, the company added the possibility of charging revenue to apartment renters as well – through integrations with companies like Sonos, Honeywell, Lev and ecobee thermostats, and Leviton light switches. Schoenfelder framed their business as a “platform model,” describing the software as LatchOS or operating system, as if they were akin to the Apple OS. At the time of their SPAC IPO, supported by venerable NYC real estate company Tishman Speyer, Latch claimed that one out of every 10 buildings in the US was being built with their products embedded. 

But like Bird, Latch seems to have fooled itself that it was, indeed, a software company, touting SaaS KPIs like LTV:CAC ratios (lifetime value as a ratio to cost of customer acquisition, a metric that was popular during the SaaS boom and encouraged companies to spent against rules of thumb rather than core product margins). 

How could Latch have failed with not one, not two, but three potential revenue streams from the same core hardware experience? They were overly bullish on revenue projections, expecting $100 MM in the year after their IPO, when they generated $49 MM. Their frictionless software revenue was in fact heavily dependent on friction-filled building contractors and supply chain shortages that have plagued the industry since the pandemic. 

Like Bird, the company said that its financial statements for 2021 and the first quarter of 2022 are unreliable warning of “material errors and possible irregularities.” The company is a target of a shareholder lawsuit for misstating revenues, and the NASDAQ sent a notification of non-compliance for failing to issue both their second and third quarterly reports of this year. 

The slide below is taken from Latch’s SPAC IPO document showing their framing of the business model as classic Software-as-a-Service, a function of LTV/CAC metrics, not acknowledging the company’s dependence on hardware, and deeper dependence on construction timelines that ended up as the rate-limiter for the company’s growth.


Pattern Matching to the Wrong Models

So in sum – how did these companies mess up the connected hardware model so badly? It’s one thing for two charismatic startup founders  hungry Venture and SPAC-marketers that their old-school friction-filled hardware companies had the margins and growth trajectories of software. But notice – it happened to Amazon, too. 

One primary reason: these companies patterned matched against the wrong business model. 

Amazon pattern matched against the razor/razorblade model common in razors, water filters, and game cartridge products. The idea: you sell the hardware at a loss, but make it up on a high margin consumable. In this case, the consumable cloud revenue, or up-selling, never materialized to recover those costs. 

Bird pattern matched against Uber and Lyft – marketplace models that had no substantial hardware on the balance sheet, whereas Lyft did. Listing per-use margins without recognizing utilization of hardware gave a distorted understanding of the economics at a unit level. 

Finally, Latch thought it was a Software-as-a-Service company. But hardware is dependent on things you can touch, and rate-limiting factors like supply chains and building timelines. The company’s framing of LTV/CAC was also distorting the potential growth trajectory. 

To the hardware and device makers in my tribe: I’m not as worried about you. You are focused rebuilding your supply chains, strengthening your regional manufacturing ecosystems, and you are investing in apprenticeship to train the next generation of builders. You’ve avoided faulty combinations and when you have adopted hardware-as-a-service, you have focused on reconfiguring to create better value for your customers and your stakeholders. It may still take many months before the zombie companies stop sucking up the air time, but then you’ll just have more clarity as you build your ecosystem with investors and partners who are no longer intoxicated with faulty as-a-service mirages.

Give us a call if you are still struggling to de-SaaS your brain and figure out how to make your hardware model profitable.