Eight Sleep's Sleep Hack Revolution

Good morning. Ever since I started working with connected devices (and vehicles), I have been obsessed with this new stream of unexplored sensor data. Everything that collects and uploads new forms of high-frequency data could give us so many insights about patterns that we didn’t see before—a whole new world.

We connect everyday objects to the Internet, add sensors to track specific interactions and movements, and suddenly, we have a range of data that can better explain how the world works.

It clicked for me when we started attaching gyroscopes and accelerometers to vehicles. Suppose we understand every signal that comes from these sensors. In that case, we can detect very valuable events and information without having to watch hundreds of hours of dashcam footage - we would just look at the footage where interesting things happen.

I learned this in the vehicle space, but the same principle applies to fitness trackers.

How to build a fitness tracker in a nutshell

Garmin, Fitbit, and Apple make devices that include sensors that translate movement into forms of data. They merge that with data about how our body reacts from other sensors (heart rate, heart rate variability, blood oxygen, skin temperature, etc.).

Then, they synchronize data streams (which, let me tell you, is quite a challenge in itself) and add filters to find patterns. Connect body data with movement data, and information emerges slowly. Because training these models takes time, optimizations and iterations can lead to valuable new insights.

But since they are all operating businesses, the challenge is to build “good enough” models that are valuable to paying customers and then fund further research and development through this business.

Silicon Valley’s New Obsession

Today, we’ll discuss Silicon Valley’s obsession with better sleep and how Eight Sleep built a business around it. They used a combination of factors that made their business one of the hardest to succeed: consumer hardware combined with AI. However, they were able to make it work by using a playbook that had been used before.

Silicon Valley Hustle Meets the Need for Sleep

Biohacking in Silicon Valley is not new. It's a long-running movement of optimizing one's body and habits for maximum performance. It’s really about experimenting, and Silicon Valley loves experimenting.

Biohacking is experimenting with how you treat your body and health and the outcome of such treatment, with the goal of optimizing for maximum performance, focus, and productivity and being the most optimized, least human human.

But like with everything, these extreme experiments yield some interesting insights that more humans can apply to improve their lifestyles.

And it turns out that one of the biggest factors is healthy sleep.

Even if you say, " Why are you surprised?” It’s more like a relief. No more feeling intimidated or guilty when certain professionals mention that they are proud of their low sleep and constant 20-hour workdays. Or founders who have an extra large stake in telling people how little they sleep, regardless of whether it’s true or not.

We can finally all admit that sleep is good.

Global Sleep Crisis

ResMed’s 2024 Global Sleep Survey recently uncovered that our world is in a sleep crisis.

• 36,000 people across 17 markets responded to the annual survey

• 40% of respondents get no more than three nights of good sleep per week

• More than 33% have started to track their sleep patterns actively and are working on adjusting their sleep patterns.

We are becoming more and more aware of the fact that we should focus on improved sleep. And luckily we are entering a phase with more advanced tech that seems to bring actual results.

Don’t take it from me; listen to Zuck.

or Elon…

or Lewis Hamilton…

Tracking all the things

I, too, became curious about data trackers that could help me identify ways to optimize my sleep and performance. So, last year, I got a Whoop.

A Whoop is a wristband without a display, just a sensor that tracks certain biometric data points. What I loved about it was that I could inform Whoop via a daily journal about what I did that day—making sure that some model was trained with my personal data to find patterns.

Imagine having a team of machine learning engineers analyze sensor data from your body and find patterns based on your behavior and what you did the day before. Over time, these patterns show what is good and bad for your personal health. I loved that idea, and I was willing to participate in this experiment (by paying a one-year subscription and filling out a daily 1-minute checklist).

After one year, I can say that I am more aware of a few things that directly impact my sleep and recovery. Although there were no huge surprises, it was impactful to see quantified data about my recovery and sleep performance.

And I can tell you, celebrating my little local Oktoberfest I am already worried to see my sleep score in deep red the following day.

So I conclude, putting a sensor on my wrist gave me some interesting data to learn from.

But having that data is only “half-useful”. I also need to act accordingly.

What if my surroundings could somehow be altered based on the insights from my health data so that the outcome is peak performance and well-being for me—without me having to do anything!?

Eight Sleep: The Self-Driving-Car of Sleep-Tech

This is where Eight Sleep comes in.

A bedsheet is a large surface that can fit a ton of new sensors that can’t fit on our wrists. Okay, cool. It can collect even more data. It’s like a sleep lab at home. Nice.

This is where Eight Sleep completely changes the game. Their products can alter your environment based on what your body needs. If you have a longer deep sleep with a colder bedsheet, Eight Sleep will learn and automatically adjust that. Could it be true?

Eight Sleep designs and creates accessories like a bedsheet that can adjust temperature and a bed stand that can change elevation. These are connected to a pod that controls and adjusts these based on what’s best for you.

So it can run experiments with you while you are asleep.

• Change temperature

• Change the elevation of certain parts of the mattress

• Measure sleep impact

• Then, learn from these changes and iterate

But if you start with this idea and no data? How would you know what people need? And how do you build these models? And then, how can you sell that mattress on the hypothesis that it could help you? That makes the go-to-market quite complicated, doesn’t it?

The Market Entry Model

Eight Sleep needed to solve a problem—finding people to use their product when the models and maybe the sensor layout are not that robust and optimized yet.

Never mind the hardware - the value kicker will be in the data and resulting models.

Their market entry model is not new: Build a premium brand, deliver a high-end product, charge premium prices for your first products to fund continued R&D, collect lots of data, and work on better, predictive models.

Then, start coming down the cost curve with mass-market products and derivatives that promise everyone the best sleep they ever had. This promise has already been confirmed by influencers, celebrities, and raving customers in the premium segment.

Then, add more products that improve sleep to your product line.

All you have to do is find that first product that’s good enough to get customers excited.

How It Started: Getting Investors to Bite.

The promise to investors was the invention of a “sleep performance machine”.

The vision was to build an enclosed space that can completely control your environment.

The founder Matteo Franceschetti wrote up a memo that he called “Waking up in the future”.

His main point about sleep was that society had pills, mattresses and wearables. But these were passive and one-size fits all. He knew that a personalization is key. People have their own needs for a perfect sleep.

The investor reaction… Meh.

They applied to Y Combinator and were initially rejected. This led them to start an Indiegogo campaign, which secured over $1 million. This was more than enough to confirm that they were onto something.

They developed a prototype and went back to YC with that. This time, they got in.

Following YC they raised a seed round of $6 million and a $12.9 million Series A.

In the meantime, pre-orders grew, but the product wasn’t ready. Another round of capital was needed to get to market. They raised another $10 million in Series B to finalize the product that would be deemed good enough to launch. Still, not enough, they had to raise another $40 million in Series B2. Consumer hardware… I am telling you, it’s no a joke.

Finding an Audience of Enthusiast Multipliers

Eight Sleep knew exactly who they needed to target for their first products. They needed to find the people who care about peak performance and are outspoken about it. They were looking to find top athletes and business professionals who are role models in their world and that would influence others by announcing how this product help them improve on their goals.

Running online ads didn’t immediately take off for them. At least not in a sustainable manner. The cost of acquisition was way too high. Understandable. Aren’t we all spoiled already with all kinds of mattress ads promising better sleep? Why would we trust this one?

They needed word of mouth and they could only get that once they satisfied their early customers. They needed to become believers.

So they decided that the temperature regulation was just good enough to prove their product-market fit and create enough impact on a customer’s sleep that they would become believers. The product would understand your deep sleep and restful sleep and learn how it could be improved by changing the temperature of your sheet. That was it. The temperature changes made a huge impact on users who were complaining about non-restful sleep.

It worked.

Over time, their customers started talking about it. Influential people in the biohacking industry became aware, and word of mouth brought them attention. Now, over 30% of their revenue comes from word of mouth.

I’ve shared some pretty heavyweight endorsements at the beginning of this post from Mark Zuckerberg, Elon Musk, and seven-time F1 world champion Lewis Hamilton.

Sidequest - Dashcam Collision Detector

I remember when leading the SaaS platform for a company that designed and built connected dashcams; we won our customers over in a similar way.

Our cameras would analyze sensor data based on vehicle movement. With an accelerometer and gyroscope trace, it wasn’t hard for our models to detect a big collision in near real time. It’s a high-g-force event. So this was one of the first features we launched our beta with. We would notify customers (fleet owners via notification and video footage) about collisions in their fleet.

In a lot of cases, that footage helped exonerate their drivers and save them millions of dollars. After an event like this, these fleet owners became our biggest multipliers.

Building a Sustainable Business

Having found their product market fit, Eight Sleep is now expanding their product portfolio with all kinds of add-ons and even a not so much liked subscription model.

Subscription models are a way to fund continuous research and development. Machine learning and AI engineers are expensive, and without high profits, it will be hard to keep them employed. On top of that, computing to run these models on high-performing GPUs and CPUs is not exactly a steal either.

It was a struggle for Eight Sleep to find that profitability at first. But constant iterations on pricing and product portfolio made them ….

Now the key for them becomes building their ecosystem further and keeping their customers in it.

Adding products that can further enhance sleep - lights, sound and scent devices, supplements it can all impact sleep and it can all be added to the Eight Sleep ecosystem.

For this, they can look closely at Apple’s playbook, which we discussed last week.

Conclusion and Key Takeaways

One thing to take away is that there are untapped opportunities, even in markets as established and seemingly mature as the sleep industry. Pairing these with technology can reveal some unexpected innovations.

• A combination of sensors and models that extract valuable information from everyday products can lead to finding golden nuggets of how someone’s life can be improved. Wrapping this into a “take-home sleep lab” is a low barrier for (price-insensitive) customers to improve their sleep based on data literally while they sleep… (I couldn’t help it…)

• Building a data driven ecosystem that has a self-reinforcing learning loop enables the data flywheel. But building it on a personalized basis that enables your product to actually react to the customers preferences leads to very high customer satisfaction.

• They looked beyond comfort and innovated towards outcomes measured in performance, well-being, and improvements that customers could feel the next day.

• Leveraging the premium segment in a very specific market of peak performance seekers (at all costs) and driving measurable impact helped them create word-of-mouth promotion.

• Building a data-driven ecosystem made their product not just a one-time purchase but an ongoing service that they can monetize.

Their strategic vision to position themselves as performance enhancers helped them build a loyal customer base. Through word-of-mouth, they established their position in the center of the lucrative sleep-tech economy.

Entering the hardware consumer market is hard—it might be one of the hardest to master. Building out a machine learning operation that constantly improves the product for customers on a personalized basis is a monumental undertaking. I almost want to talk about this more… however, this post is already getting beyond the “5-minute-read” promise, and I might put it in a part two post for next week.

Have a great rest of the week and I hope you have a great sleep too.

Best,