Our Lean LaunchPad for Life Science class talked to 2,355 customers, tested 947 hypotheses and invalidated 423 of them. They had 1,145 engagements with instructors and mentors. (We kept track of all this data by instrumenting the teams with LaunchPad Central software.)
This post is one of a series of the “Lessons Learned” presentations and videos from our class.
Sometimes a startup results from a technical innovation. Or from a change in regulation, declining costs, changes in consumers needs or an insight about customer needs. Resultcare, one of the 26 teams in the class started when a resident in clinical medicine at UCSF watched her mother die of breast cancer and her husband get critically injured.
A team of 110 researchers and clinicians, in therapeutics, diagnostics, devices and digital health in 25 teams at UCSF, has just shown us the future of translational medicine. It’s Lean, it’s fast, it works and it’s unlike anything else ever done.
It’s going to get research from the lab to the bedside cheaper and faster.
Our class talked to 2,355 customers, tested 947 hypotheses and invalidated 423 of them. They had 1,145 engagements with instructors and mentors. (We kept track of all this data by instrumenting the teams with LaunchPad Central software.)
In a packed auditorium in Genentech Hall at UCSF, the teams summarized what they learned after 10 weeks of getting out of the building. This was our version of Demo Day – we call it “Lessons Learned” Day. Each team make two presentations:
2 minutes YouTube Video: General story of what they learned from the class
8 minute Lessons Learned Presentation: Very specificstory about what they learned in 10 weeks about their business model
In the next few posts I’m going to share a few of the final “Lessons Learned” presentations and videos and then summarize lessons learned from the teaching team.
Magnamosis Magnamosis is a medical device company that has a new way to create a magnetic compression anastomosis (a surgical connection between two tubular structures like the bowel) with improved outcomes.
Their initial idea was that making an anastomosis that’s better, faster and cheaper will have surgeons fighting to the death to get a hold of their device. They quickly found out that wasn’t the case. Leak rates turned out to a bigger issue with surgeons and a much larger market.
Their slide deck is below. Don’t miss the evolution of their business model in the Appendix.
If you can’t see the presentation above, click here
The best summary of why Scientists, Engineers and Principal Investigators need to get out of the building was summarized by Dr. Harrison below. After working on his product for a decade listen to how 10 weeks of the Lean LaunchPad class radically changed his value proposition and business model.
As our Lean LaunchPad for Life Sciences class winds down, a good number of the 26 teams are trying to figure out whether they should go forward to turn their class project into a business.
Given that we’ve been emphasizing Evidence-based entrepreneurship and the Investment Readiness Level, I guess I shouldn’t have been surprised when someone asked, “After we figure all this data out, should we pursue our idea based on the numbers?”
I pointed out that the “data” you gather in 10 weeks (talking to 100+ customers, partners, payers, etc.,) are not the first thing you should look at. There are three more important things you should worry about.
Now that you’ve gotten to know your potential channel and customers, regardless of how much money you’re going to make, will you enjoy working with these customers for the next 3 or 4 years?
One of the largest mistakes in my career was getting this wrong. I used to be in startups where I was dealing with engineers designing our microprocessors or selling supercomputers to research scientists solving really interesting technical problems. But in my next to last company, I got into the video game business.
My customers were 14-year old boys. (see 1:30 in the video) I hated them. It was a lifelong lesson that taught me to never start a business where you hate your customers. It never goes well. You don’t want to talk to them. You don’t want to do Customer Development with them. You just want them to go away. And in my case they did – they didn’t buy anything.
So you and your team need to feel comfortable being in this business with these customers.
2. Is this a scalable business? And if not, are you Ok with something small?
Is it a lifestyle business while you’re keeping your other job? Is it a small business that hits $4 million in revenue in four years and $8 million in ten years? Or is it something that can grow to a size that will result in an acquisition or some liquidity event?
You need to decide what your personal goal is and how it matches what you think this business can grow into. And you and your cofounders need to have that discussion to make sure that all the co-founders’ interests are aligned – before you make any decision to start the company. If one of you are happy making $500K/year and the other has visions of selling the company to Roche for a billion dollars, you have very different goals. Without clear alignment, one or both of you will be really unhappy later when you try to make decisions.
3. If I Didn’t Make Any Money After 4 Years, Did I Still Have A Great Time?
If your company fails, would you still say you had one hell of a ride? Founders don’t do startups because they’re searching for a huge financial windfall. They do it because it’s the greatest invention they can imagine. Most of the time you will fail. So if you’re not going to have a great time with your team and learn and build something you are truly excited about – don’t do it.
We talk a lot about Customer Development, but there’s nothing like seeing it in action to understand its power. Here’s what happened when an extraordinary Digital Health team gained several critical insights about their business model. The first was reducing what they thought was a five-sided market to a simpler two-sided one.
But the big payoff came when their discussions with medical device customers revealed an entirely new way to think about pricing —potentially tripling their revenue.
We’re into week 9 of teaching a Lean LaunchPad class for Life Sciences and Health Care (therapeutics, diagnostics, devices and digital health) at UCSF teaching with a team of veteran venture capitalists. The class has talked to ~2,200 customers to date. (Our final – not to be missed – Lessons Learned presentations are coming up December 10th.)
Among the 28 startups in the Digital Health cohort is Tidepool. They began the class believing they were selling an open data and software platform for people with Type 1 Diabetes into a multi-sided market comprised of patients, providers, device makers, app builders and researchers.
The Tidepool team members are:
Aaron Neinstein MD Assistant Professor of Clinical Medicine, Endocrinology and Assistant Director of Informatics at UCSF. He’s an expert in the intersection between technological innovations and system improvement in healthcare. His goal is to make health information easier to access and understand.
Howard Look, CEO of Tidepool, was VP of Software and User Experience at TiVo. He was also VP of Software at Pixar, developing Pixar’s film-making system, and at Amazon where he ran a cloud services project. At Linden Lab, delivered the open-sourced Second Life Viewer 2.0 project. His teenage daughter has Type 1 diabetes.
Brandon Arbiter was a VP at FreshDirect where he built the company’s data management and analytics practices. He was diagnosed at age 27 with Type 1 Diabetes. He developed a new generation diabetes app, “nutshell,” that gives patients the information they need to make the right decisions about their dosing strategies.
Kent Quirk was director of engineering at Playdom and director of engineering at Linden Labs.
A Five-sided Market In Week 1 the Tidepool team diagramed its customer segment relationships like this:
Using the business model canvas they started with their value proposition hypotheses, articulating the products and services they offered for each of the five customer segments. Then they summarized what they thought would be the gain creators and pain relievers for each of these segments.
Next, they then did the same for the Customer Segment portion of the canvas. They listed the Customer Jobs to be done and the Pains and Gains they believed their Value Proposition would solve for each of their five customer segments.
It’s Much Simpler Having a multisided market with five segments is a pretty complicated business model. In some industries such as medical devices its just a fact of life. But after talking to dozens of customers by week 3, Tidepool discovered that in fact they had a much simpler business model – it was a two-sided market.
They discovered that the only thing that mattered in the first year or two of their business was building the patient-device maker relationship. Everything else was secondary. This dramatically simplified their value proposition and customer segment canvas.
So they came up with a New Week 3 Value Proposition Canvas:
And that simplified their New Week 3 Customer Segment Canvas
Cost-based Pricing versus Value-based Pricing While simplifying their customer segments was a pretty big payoff for 3 weeks into the class, the best was yet to come.
As part of the revenue streams portion of the business model canvas, each team has to diagram the payment flows.
The Tidepool team originally believed they were going charge their device partners “market prices” for access to their platform. They estimated their Average Revenue per User (ARPU) would be about $36 per year.
But by week 6 they had spoken to over 70 patients and device makers. And what they found raised their average revenue per user from $36 to $90.
When talking to device makers they learned how the device makers get, keep and grow their customers. And they discovered that:
device makers own customers would stay their customers for 10 years (i.e. the Customer Life Time (CLT))
and the Life Time Value (LTV) of one customer over those 10 years to a device maker is $10,000
These customer conversations led the Tidepool team to further refine their understanding of the device makers’ economics. They found out that the device makers sales and marketing teams were both spending money to acquire customers. ($500 per sales rep per device + $800 marketing discounts offered to competitors’ customers.)
Once they understood their device customers’ economics, they realized they could help these device companies reduce their marketing spend by moving some of those dollars to Tidepool. And they realized that the use of the Tidepool software could reduce the device companies’ customer churn rate by at least 1%.
This meant that Tidepool could price their product based on the $1,800 they were going to save their medical device customers. Read the previous sentence again. This is a really big idea.
The Tidepool team went from cost-based pricing to value-based pricing. Raising their average revenue per user from $36 to $90.
There is no possible way that any team, regardless of how smart they are could figure this out from inside their building.
If you want to understand how Customer Discovery works and what it can do in the hands of a smart team, watch the video below. The team ruthlessly dissects their learning and builds value-pricing from what they learned.
This short video is a classic in Customer Discovery.
Investors sitting through Incubator or Accelerator demo days have three metrics to judge fledgling startups – 1) great looking product demos, 2) compelling PowerPoint slides, and 3) a world-class team.
We think we can do better.
We now have the tools, technology and data to take incubators and accelerators to the next level. Teams can prove their competence and validate their ideas by showing investors evidencethat there’s a repeatable and scalable business model. And we can offer investors metrics to play Moneyball – with the Investment Readiness Level.
We’ve spent the last 3 years building a methodology, classes, an accelerator and software tools and we’ve tested them on ~500 startups teams.
A Lean Startup methodologyoffers entrepreneurs a framework to focus on what’s important: Business Model Discovery. Teams use the Lean Startup toolkit: the Business Model Canvas + Customer Development process + Agile Engineering. These three tools allow startups to focus on the parts of an early stage venture that matter the most: the product, product/market fit, customer acquisition, revenue and cost model, channels and partners.
An Evidence-based Curriculum (currently taught in the Lean LaunchPad classes and NSF Innovation Corps accelerator). In it we emphasize that a) the data needed exists outside the building, b) teams use the scientific method of hypothesis testing c) teams keep a continual weekly cadence of:
Hypothesis – Here’s What We Thought
Experiments – Here’s What We Did
Data – Here’s What We Learned
Insights and Action – Here’s What We Are Going to Do Next
LaunchPad Central software is used to track the business model canvas and customer discovery progress of each team. We can see each teams hypotheses, look at the experiments they’re running to test the hypotheses, see their customer interviews, analyze the data and watch as they iterate and pivot.
We focus on evidence and trajectory across the business model. Flashy demo days are great theater, but it’s not clear there’s a correlation between giving a great PowerPoint presentation and a two minute demo and building a successful business model. Rather than a product demo – we believe in a “Learning Demo”. We’ve found that “Lessons Learned” day showing what the teams learned along with the “metrics that matter” is a better fit than a Demo Day.
“Lessons Learned” day allows us to directly assess the ability of the team to learn, pivot and move forward. Based on the “lessons learned” we generate an Investment Readiness Level metric that we can use as part of our “go” or “no-go” decision for funding.
NASA and the Technology Readiness Level (TRL) In the 1970’s/80’s NASA needed a common way to describe the maturity and state of flight readiness of their technology projects. They invented a 9-step description of how ready a technology project was. They then mapped those 9-levels to a thermometer.
What’s important to note is that the TRL is imperfect. It’s subjective. It’s incomplete. But it’s a major leap over what was being used before. Before there was no common language to compare projects.
The TRL solved a huge problem – it was a simple and visual way to share a common understanding of technology status. The U.S. Air Force, then the Army and then the entire U.S. Department of Defense along with the European Space Agency (ESA) all have adopted the TRL to manage their complex projects. As simple as it is, the TRL is used to manage funding and go/no decisions for complex programs worldwide.
We propose we can do the same for new ventures – provide a simple and visual way to share a common understanding of startup readiness status. We call this the Investment Readiness Level .
The Investment Readiness Level (IRL) The collective wisdom of venture investors (including angel investors, and venture capitalists) over the past decades has been mostly subjective. Investment decisions made on the basis of “awesome presentation”, “the demo blew us away”, or “great team” is used to measure startups. These are 20th century relics of the lack of data available from each team and the lack of comparative data across a cohort and portfolio.
We collect this data into a Leaderboard (shown in the figure below) giving the incubator/accelerator manager a single dashboard to see the collective progress of the cohort. Metrics visible at a glance are number of customer interviews in the current week as well as aggregate interviews, hypotheses to test, invalidated hypotheses, mentor and instructor engagements. This data gives a feel for the evidence and trajectory of the cohort as a whole and a top-level of view of each teams progress.
Next, we have each team update their Business Model Canvas weekly based on the 10+ customer interviews they’ve completed.
The canvas updates are driven by the 10+ customer interviews a week each team is doing. Teams document each and every customer interaction in a Discovery Narrative. These interactions provide feedback and validate or invalidate each hypothesis.
Underlying the canvas is an Activity Map which shows the hypotheses tested and which have been validated or invalidated.
All this data is rolled into a Scorecard, essentially a Kanban board which allows the teams to visualize the work to do, the work in progress and the work done for all nine business model canvas components.
Finally the software rolls all the data into an Investment Readiness Level score.
MoneyBall At first glance this process seems ludicrous. Startup success is all about the team. Or the founder, or the product, or the market – no metrics can measure those intangibles.
Baseball used to believe that as well. Until 2002 – when the Oakland A’s’ baseball team took advantage of analytical metrics of player performance to field a team that competed successfully against much richer competitors.
Statistical analysis demonstrated that on-base percentage and slugging percentage were better indicators of offensive success, and the A’s became convinced that these qualities were cheaper to obtain on the open market than more historically valued qualities such as speed and contact. These observations often flew in the face of conventional baseball wisdom and the beliefs of many baseball scouts and executives.
By re-evaluating the strategies that produce wins on the field, the 2002 Oakland A’s spent $41 million in salary, and were competitive with the New York Yankees, who spent $125 million.
Our contention is that the Lean Startup + Evidence based Entrepreneurship + LaunchPad Central Software now allows incubators and accelerators to have a robust and consistent data set across teams. While it doesn’t eliminate great investor judgement, pattern recognitions skills and mentoring – it does provide them the option to play Moneyball.
One of the great innovations of the 21st century are products that are cloud-connected and update and improve automatically. For software, gone are the days of having to buy a new version of physical media (disks or CD’s.) For hardware it’s the magical ability to have a product get better over time as new features are automatically added.
The downside is when companies unilaterally remove features from their products without asking their customers permission and/or remove consumers’ ability to use the previous versions. Products can just as easily be downgraded as upgraded.
It was a wake up call when Amazon did it with books, disappointing when Google did it with Google Maps, annoying when Apple did it to their office applications – but Tesla just did it on a $100,000 car.
It’s time to think about a 21st Century Bill of Consumer Product Rights.
Google – Well It Looks Better In July 2013 Google completely redesigned Google Maps – and users discovered that on their desktop/laptop, the new product was slower than the one it replaced and features that were previously available disappeared. The new Google Maps was worse then one it replaced – except for one key thing – its User Interface was prettier and was unified across platforms. If design was the goal, then Google succeeded. If usability and functionality was a goal, then the new version was a step backwards.
Apple – Our Code Base is More Important than Your Features In November 2013 Apple updated its operating system and cajoled its customers to update their copies of Apple’s iWork office applications – Pages (Apple’s equivalent to Microsoft Word), Keynote (its PowerPoint equivalent), and Numbers (an attempt to match Excel). To get users to migrate from Microsoft Office and Google Docs, Apple offered these iWorks products for free.
Sounds great– who wouldn’t want the newest version of iWorks with the new OS especially at zero cost? But that’s because you would assume the new versions would have more features. Or perhaps given its new fancy user interface, the same features? The last thing you would assume is that it had fewer features. Apple released new versions of these applications with key features missing, features that some users had previously paid for, used, and needed. (Had they bothered to talk to customers, Apple would have heard these missing features were critical.)
But the release notes for the new version of the product had no notice that these features were removed.
Translated into English this meant that Apple engineering recoding the products ran out of time to put all the old features back into the new versions. Apple said, “… some features from iWork ’09 were not available for the initial release. We plan to reintroduce some of these features in the next few releases and will continue to add brand new features on an ongoing basis.
Did they think anyone wouldn’t notice?
Decisions like this make you wonder if anyone on the Apple executive staff actually understood that a “unified file format” is not a customer feature.
While these examples are troubling, up until now they’ve been limited to content or software products.
Tesla – Our Problems are Now Your Problems In November 2013 Tesla, a manufacturer of ~$70,000 to $120,000 electric cars, used a software “update” to disable a hardware option customers had bought and paid for – without telling them or asking their permission.
One of Tesla features is a $2,250 “smart air suspension” option that automatically lowers the car at highway speeds for better mileage and stability. Over a period of 5 weeks, three Tesla Model S cars had caught fire after severe accidents – two of them apparently from running over road debris that may have punctured the battery pack that made up the floor pan of the car. After the car fires Tesla pushed a software release out to its users. While the release notice highlighted new features in the release, nowhere did it describe that Tesla had unilaterally disabled a key part of the smart air suspension feature customers had purchased.
Only after most of Telsa customers installed the downgrade did Tesla’s CEO admit in a blog post, “…we have rolled out an over-the-air update to the air suspension that will result in greater ground clearance at highway speed.”
Translation – we disabled one of the features you thought you bought. (The CEO went on to say that another software update in January will give drivers back control of the feature.) The explanation of the nearly overnight removal of this feature was vague “…reducing the chances of underbody impact damage, not improving safety.” If it wasn’t about safety, why wasn’t it offered as a user-selected option? One could only guess the no notice and immediacy of the release had to do with the National Highway Safety Administration investigation of the Tesla Model S car fires.
This raises the question: when Tesla is faced with future legal or regulatory issues, what other hardware features might Tesla remove or limit in cars in another software release? Adding speed limits? Acceleration limits? Turning off the Web browser when driving? The list of potential downgrades to the car is endless with the precedent now set of no obligation to notify their owners or ask their permission.
In the 20th century if someone had snuck into your garage and attempted to remove a feature from your car, you’d call the police. In the 21st century it’s starting to look like the normal course of business.
What to Do While these Amazon, Google, Apple and Tesla examples may appear disconnected, taken together they are the harbinger of the future for 21st century consumers. Cloud-based updates and products have changed the landscape for consumers. The product you bought today may not be the product you own later.
Given there’s no corporate obligation that consumers permanently own their content or features, coupled with the lack of any regulatory oversight of cloud-based products, Apple’s and Tesla’s behavior tells us what other companies will do when faced with engineering constraints, litigation or regulation. In each of these cases they took the most expedient point of view; they acted as if their customers had no guaranteed rights to features they had purchased. So problem solving in the corporate board room has started with “lets change the feature set” rather than “the features we sold are inviolate so lets solve the problem elsewhere.”
The result is that consumers in the 21st century have less protection then they did in the 20th.
What we can hope for is that smart companies will agree to a 21st Century Bill of Consumer Product Rights. What will likely have to happen first is a class-action lawsuit establishing consumers’ permanent rights to retain features they have already purchased.
Some smart startups might find a competitive advantage by offering customer-centric products with an option of “no changes” and “perpetual feature rights” guarantee.
A 21st Century Bill of Consumer Product Rights
No changes to content paid for (whether on a user’s device or accessed in the cloud)
Notify users if an update downgrades or removes a feature
Give users the option of not installing an update
Provide users an ability to rollback (go back to a previous release) of the software
The product you bought today may not be the product you have later
Manufacturers can downgrade your product as well as upgrade it
You have no legal protection
Update: a shorter version of the post was removed from the Tesla website forum
This post is an update of what we learned about life science revenue models.
Life Science/Health Care revenue streams differ by Category For commercialization, the business model (Customers, Channel, Revenue Model, etc.) for therapeutics, diagnostics, devices, bioinformatics and digital health have very little in common.
This weeks topic was revenue streams – how much cash the company can generate from each customer segment. Revenue streams have two parts: the revenue strategy and the pricing tactics.
Figuring out revenue strategy starts by gaining a deep understanding of the target customer(s). Setting a revenue strategy starts with understanding the basics about the customer segments:
who’s the user, the recommender, buyer, and payer
How the target customer currently purchases goods and services and how much they currently pay for equivalent products
Their willingness to pay for value versus lowest cost?
How much budget they have for your type of product?
Revenue strategy asks questions like, “Should we offer cost-based or value-based pricing. How about demand-based pricing? Freemium? Do we price based on hardware sales or do we offer hardware plus consumables (parts that need to be disposed or replaced regularly)? Do we sell a single software package or a subscription? These strategy hypotheses are tested against the target customer segment(s).
Once you’ve established a revenue strategy the pricingtacticsfollow. Pricing is simply “how much can I charge for the product using the selected revenue strategy?” Pricing may be as simple as setting a dollar value for hardware or software, or as complicated as setting a high price and skimming the market or setting a low price as a loss leader.
You can get a feel for how each of the cohorts address the Revenue Streams by looking at the Revenue lectures below – covering the therapeutics, diagnostics, devices and digital health cohorts.
At the end of the lectures you can see a “compare and contrast” video and a summary of the differences in distribution channels.
Week 5 Todd Morrill Instructor
If you can’t see the presentation above click here
Week 5 Abhas Gupta Instructor
If you can’t see the presentation above click here
Week 5 Allan May Instructor
If you can’t see the presentation above click here
Week 5 Karl Handelsman Instructor
If you can’t see the presentation above click here
Life Science and Health Care Differences in Revenue Streams ” This weeks lecture and panel was on Revenue; how much cash the company can generate from each customer segment – and the strategy and tactics to do so. Therapeutics, diagnostics, devices and digital health use different Revenue Strategies and Pricing Tactics, in the video and the summary that follows the instructors compare and contrast how they differ.
Therapeutics revenue is from drug companies not end users
18 months to first revenue from a deal
Predicated on delivering quality data to a company
Deal can be front-end or back-end loaded
Quality of the data has to be extremely high for a deal
Diagnostics (Starting at 4:10)
Diagnostic revenue is from end users: a hospital or clinical lab
You need to figure out value of your product but…
Pricing is capped by your reimbursement (CPT) code limits
Reimbursement strategy is paramount, design to good codes avoid bad ones
Find a reimbursement code consultant
Don’t do cost-based pricing… go for value-based pricing
Medical Devices (Starting at 8:23)
There really is no such thing as a perfect First Generation Medical Device
So Medical Device companies often start with a Volkswagen product and then build to the Ferrari product
Revenue models are typically direct product sales
Don’t do cost-based pricing… go for value-based pricing, especially where your device lowers the treatment costs of the patient
In most cases, pricing is capped by your reimbursement (CPT) code limits
Or pricing can be capped by what competitors offer, unless you can demonstrate superior cost savings
In a new market there is no reimbursement code but if you show high cost-savings you can get a high reimbursement rate
A risk in device hardware is getting trapped in low-volume manufacturing with low margins and run out of cash
Digital Health (Starting at 10:35)
Digital Health revenue models are often subscription models to a company per month across a large number of users
Intermediation fees – where you broker a transaction – are another source of revenue (i.e. HealthTap)
Advertising is another digital health revenue model, but requires at least 10 million users to have a meaningful model, but can be lower if you have higher value uses like specialist physicians because you can charge dollars not cents
Don’t do cost-based pricing… go for value-based pricing
Value-based pricing is based on the needs you’ve learned from the customer segment and the strength of your product/market fit
the sum of customer needs + product/market fit = the pricing you can achieve
Each of these Life Science domains has a unique revenue strategy and pricing tactic
In therapeutics revenue comes in lump milestone payments from drug companies based on quality data
Diagnostics revenue comes value pricing to hospital or clinical lab
capped by reimbursement (CPT) code limits
Device pricing starts by offering an initial value-priced base product and then following up with a fully featured product
capped by reimbursement (CPT) code limits
Digital health products use subscription value pricing. Alternatively may use advertising revenue model