Rebecca Lynn - Finding Your Customers - [Invest Like the Best, EP.309]
What this covers
Rebecca Lynn, co-founder of Canvas Ventures, talks with the host about the go-to-market craft she learned at Procter & Gamble and how it carries over to startups. The core question is how to learn what customers want. She explains why surveys mislead: people have no reference point for a new concept, and they answer behavioral questions, like whether they balance a checkbook, with the answer they think they should give. Her alternative is the concept and use test, plus splash pages and minimum viable products that measure behavior. She also pushes back on the story of Steve Jobs as an intuitive oracle, noting he tested extensively, and says P&G's expert bets often finished last.
The middle of the conversation turns to organizational mechanics. Lynn argues that incentives explain behavior, citing the HITECH Act and the sales-to-customer-success handoff, where compensation rewards signing deals rather than fit. On positioning, she says defining what a company is not forces focus. She favors vertical messaging over horizontal platform claims and treats the number of customers an early-stage founder has turned down as a diagnostic.
The episode then wanders. Lynn gives opinions on health care that she flags as non-expert and that are contested: obesity as the main cost driver, dietary fat, and voice technology reaching low-literacy users. She also covers venture cyclicality, the 2009 opening that led her to LendingClub, and her investing test of whether she would join the company herself.
Lynn argues that go-to-market success depends on testing products with real customers rather than trusting surveys or executive instinct, and on designing incentives, positioning, and customer focus so that companies serve a defined customer and are willing to say no to the wrong ones.
- Surveys fail on judgment-laden questions and on products consumers have never experienced; in-hand concept and use tests work better
- Founders routinely misidentify their customers, and sales-to-success handoffs break because incentives reward signing deals rather than customer fit
- Positioning works by deciding what the company is not, and early-stage firms should verticalize their message and turn away poor-fit customers
This asset isn't compiled yet
You're seeing its claims, ranked. Compile it to build the argument threads, weight them, and check each claim against your library — the full view.
Venture capital is cyclical and has been declared dead roughly every ten years; those downturns, when capital is scarce and talent is available, are often the best times to build great companies and invest.
“every 10 years, venture capital is dead. And those are the best times to be in it”
Consumers cannot meaningfully react to a brand-new concept in a survey because they have no reference point, so teams must let people see, touch, and use a product through concept and use testing; surveys only work for incremental improvements to existing products.
“you can't ask people what would you think of X if it doesn't exist. For example, what would you think of a coffee cup with two handles?”
Quantitative surveys mislead when they ask about behavior that implies a judgment, such as whether someone pays bills on time or balances a checkbook, because respondents answer with the response they think they should give, so what people say diverges from what they do.
“whenever there's a judgment about how somebody should think, feel or act, you stay away from surveys”
Brilliant CEOs and brand experts cannot reliably predict consumer preferences; at P&G, the bets of the engineer or senior brand person often finished last in product tests, so consumer testing should override instinct.
“whatever the engineer or the really smart brand person would bet on, usually would come in the last”
The hardest kind of business is one that must re-acquire the customer every time, so companies should design for the whole customer lifecycle, since the true cost is acquiring the customer.
“The hardest company to be in, is the company where you have to re-acquire the consumer every single time”
What people say they do, such as balancing a checkbook or tracking their diet, differs from what they actually do, so self-reported behavior should be checked against actual behavior or real-world testing.
“do you balance your checkbook, what people will say they do and what the reality is are two different things.”
Rather than asking people whether they would use a product, companies should put out splash pages or minimum viable products and measure real behavior, because stated enthusiasm is unreliable.
“let's just run with a minimum viable product and just get some feedback because a lot of times, people will say, oh, do you like this? Would you use this?”
The sales-to-success handoff breaks because sales compensation rewards signing deals while customer success teams are not incentivized in the same way, so the fix is to reward sales for post-live customer success and bring success staff into the sales process as a fit check.
“people do what they're incentivized to do. And so you really have to be clear about your incentive program”
When trying to understand why people did something, look at the incentive they were offered, because people do what they are incentivized to do, and government programs can change behavior, as when the HITECH Act drove EMR adoption.
“You incentivize the behavior and if you're ever asking why people did something, look at the incentive they were offered”
The popular story of Steve Jobs knowing what consumers wanted by instinct ignores that he did extensive consumer testing, so the CEO-as-oracle narrative misreads how great products emerge.
“Steve Jobs was brilliant and amazing and had an incredible instinct that he did do a lot of consumer testing”
A UCSF physician friend told Lynn that the real cost in health care is the roughly 10 percent of people who are not literate, not quantified self tools, which is why voice-enabled technology matters.
“she's like the real cost in health care is not this quantified self-thing, it's really the 10% of people who aren't literate.”
Companies frequently conclude that marketing campaigns did or did not work based on flawed testing, such as too-small sample sizes, inconsistent placement, or poor copy, so the conclusions are often unreliable.
“a lot of the conclusions they have drawn, especially for what didn't work were not based on solid testing methodology”
The conventional advice to avoid fat is wrong, since the brain and nerves require fat, and animal fats and avocado oil are beneficial while vegetable oils should be avoided; Lynn disclaims being a doctor.
“we're told, oh, avoid fat, that's really bad for you, but it's really not bad if your brain is 60% fat”
Obesity is the main cost driver in US healthcare because it causes hypertension, diabetes, and sleep apnea, and for sleep apnea the primary intervention is losing about 10 pounds.
“Sleep apnea, the number one care for sleep apnea is lose 10 pounds. That's the big one”
A key diagnostic of an early-stage company's focus is how many customers it has turned down, since strong founders can name specific rejected customers and explain why they were not ideal yet.
“the really, really, really hard, the critical thing for everyone to do is say no to some people. That's really the second question I ask is, how many customers have you said no to?”
During the 2008 to 2009 financial crisis, when banks stopped lending, a pocket of super-prime and prime borrowers remained safely lendable in any economy, which created an opening for LendingClub to invest in Q1 2009.
“in any market, in any economy, there is a pocket of super prime and prime that you can lend to very safely”
Voice technology, especially AI-enabled voice, can serve populations with low literacy and in any language, opening markets that were previously considered unserviceable, so it is a democratizing technology.
“voice just breaks that barrier because you can have voice technology in any language and it really is sort of a democratizing agent”
Abstract platform claims like solving all data science needs are meaningless to buyers, whereas messaging that says we helped companies just like you, with the same size, problems, and vertical, is meaningful and persuasive.
“companies who look just like you, same size, same problems, same vertical, we helped them and we can help you too. That's really meaningful”
The first investment criterion is the CEO: Lynn wants a founder she would enjoy working with for five to ten years, with a driving motivation and a history of overcoming adversity with a good attitude.
“First and foremost, it's the CEO. The CEO has to be somebody that I'd love working with and vice versa for the next 5 to 10 years”
Great positioning makes a company focus by forcing executives to decide what they are not and who they serve; horizontal claims to serve everyone are not meaningful, even for platform businesses.
“great positioning makes the company focus. It makes the executives decide what they are not”
Quantified self products primarily serve the worried-well end consumer, the same group that balances its bank account each month, which is a small pocket of people.
“you're serving the same pocket of people that balance their bank account every month, which is not a lot. You're sort of serving what they call almost like the worried well often.”
Canvas invests in areas where its team has expertise, and her personal test for a deal is whether she would join or start the company if she were not a VC.
“if I wasn't doing this job of venture investing and coaching, that is kind of how I see it, would I join or start this company?”
Her key investment barometer is whether she would join or start the company if she were not a venture investor, and she likes companies where she thinks 'isn't somebody already doing that?' yet no one is.
“my biggest barometer is, if I wasn't doing this job of venture investing and coaching, would I join or start this company?”
Consumer products should be brain-dead simple, since any decision a consumer must make reduces adoption; Apple succeeds because it removes choices, and its designs reflect what thousands of people said they wanted.
“if you make them think or you make them make a decision, you will shoot yourself in the foot”
Canvas formalized a Go-To-Market Council of roughly a couple hundred vetted go-to-market experts, matched to portfolio company needs, who advise Series A and B companies part-time because those companies cannot afford them full-time.
“we interviewed, oh my gosh, a couple of hundred people actually that were known to be experts in their field”
Lynn finds quantified self personally fascinating but is not sure it is investable, because it serves a mostly worried-well consumer segment that is not high-value, so personal interest and investment potential must be separated.
“I find it personally interesting. I'm not sure if it's investably interesting”
Survey design matters greatly: framing questions as do you like this versus do you like this, or as a simple yes-or-no, can change negative responses by about 10 percentage points.
“Doing things like saying, do you like this versus do you like this, or not, will actually result in about a 10% difference in negative responses”
Early-stage founders usually hold wrong assumptions about who their customers are, and the mismatch can be discovered by analyzing actual usage data rather than relying on the founder's stated beliefs.
“oftentimes, they have an assumption of who their consumer is and it's not right”
A fintech startup believed its audience was millennial men, but a usage survey showed the average user was a woman in her late 40s or early 50s managing household finances, and the marketing was then retargeted to that audience.
“The average age was late 40s or early 50s, female had financial household”
Surveys are always helpful for existing customers who use the product and for discovering who the customer is, but they are poor at predicting what new features the customer will want next.
“If there's a customer using your product, 100% we can go to him and ask him a survey.”
The most common break in customer funnels is the handoff from sales to implementation or customer success, which typically fails because of miscommunication or misunderstanding and slows customer setup.
“The most common break is that handoff from sales to customer implementation like success, always broken”
The venture industry's long period of rising valuations was an abnormality that should have reset years earlier, so a downturn is a normal correction rather than a sign that the model is broken.
“abnormality is that we were so up and to the right for so long. We should have had a reset a few years before it happened”
Scaling from $10 million to $100 million in revenue almost never relies on a single channel, so companies need a mix of channels and multiple levers so that if one channel fails another can be used.
“that's almost never done in a single or even a couple of different channels. What I did at NextCard and what I've done in my life is you need to have a channel mix”
A company's headline customer count can be misleading, since some companies with many customers have low-value ones (outside the US, too young, or otherwise unattractive) while others with few customers have very high-value accounts.
“we cracked that open and they're not interesting customers, right? Because they're in a bracket, they're maybe not in the U.S.”
Gabi had a mass-market product but weak marketing, so Canvas improved its funnel, brought in a direct-to-consumer executive, and used television advertising before its Series B; the company later sold to Experian.
“We ended up taking Gabi on television as pre-Series B, believe it or not. And it worked.”
In consumer categories, companies advertise their highest-tier product and biggest benefit as an umbrella concept, with other products riding under it, such as a dog-running business promising an aspirational fitness outcome while the actual service is walking.
“you typically lead with this umbrella concept, like your highest tier product. That's always what people advertise”
During past venture hibernation periods, so few carry checks were written that when Reid Dennis asked an audience who had ever seen one, nobody raised a hand, even though those were the periods when great companies were built.
“I remember one of them asked, I think Reid Dennis asked people in the audience to hold up their hand if any of them have ever seen a check. Nobody raised their hand, nobody.”
Nutrition education should begin in elementary school, because Americans have been given an inaccurate picture of diet, such as the claim to avoid fat, when the brain needs fat and vegetable oil is the problematic fat.
“Now what you don't need is like vegetable oil, that's really a bad idea.”
Asking how many drinks someone has is a classic judgment question, and people rarely answer it honestly, as Lynn and Patrick joked that only a zero answer is reliable.
“How many drinks do you have?”
In 2009 only about 6 percent of medical records were digital, and Canvas invested in Practice Fusion, the first cloud-based EMR, which was later sold to Allscripts, after government incentives such as the HITECH Act and the Affordable Care Act made the market.
“back in '09, 6% of medical records were digital, believe it or not”
The founder of Intuit and Meg Whitman of eBay both came from Procter & Gamble, and their work shows P&G's touch.
“The founder of Intuit was from Procter & Gamble, Meg Whitman as well, with eBay, was from P&G.”
At Procter & Gamble, mocked-up products were placed in consumers' hands for real use, followed by questions, which gave a far better read on new products than asking hypothetical questions.
“P&G had some really great concept in use test. So we would send mocked up product out into the hands of consumers”
Anyone claiming that option X beats option Y in quantitative testing must know the confidence interval and base size behind the comparison, a discipline P&G instilled and Silicon Valley teams typically lack.
“you damn well better know what your confidence interval and your base size is if you're talking to anybody about X versus Y winning”
Consumer preferences for basic household products like laundry detergent and toilet paper differ significantly across countries and even regions within the US, so products must be localized rather than assumed universal.
“in Mexico versus Japan versus Germany versus Canada and even in different parts of the United States, you had very different consumer preferences”
When researching customers, teams should sample the full spectrum of people, including those who love, like, dislike, or have never tried the product, rather than only support-call users or enthusiasts.
“it's really important to get a spectrum of people who love you, who just kind of like you, who are pissed off at you, who haven't tried you”
Silicon Valley companies tend to sell features and growth rather than the higher-level customer need, while Lynn pushes companies to articulate the benefit behind the product.
“It's just like feature, feature, feature in terms of when people are trying to sell their product and message it”
In health, patients are largely on their own, as doctors are not available to manage day-to-day issues, so people must use quantified self tools to understand their own condition.
“So if you're sick, it's you and you better figure it out because your doctor is not going to come at night”
When evaluating marketing results, Lynn asks to see every piece of copy, every test cell, the number tested, placement, and time of day, because the details often reveal why something failed.
“I want to see every piece of copy, I want to see every test cell and how many did you test? Where was it placed? What time of day?”
Teaching a startup course forces students to work through go-to-market fundamentals such as supply and demand, channels, and customer promise, and because these building blocks are the same across businesses, it almost does not matter what the business idea is.
“it almost doesn't matter in a way what your idea is because you're going to learn so much just by going through that process”
Nutrition education should start in elementary and preschool, with children learning to feed themselves, as in Japan where kids help make school food, because small-town Midwest grocery stores offer limited produce and abundant carbohydrates.
“I would make sure the kids are fed and they know how to feed themselves, for goodness' sake”
Startup leaders frequently have a wrong assumption about who their customer is, and the first step is to ask how many consumers they have spoken to, which ones, and what surprised them.
“Usually, my first question is how many consumers have they spoken to and which consumers and what did they learn?”
Going through the process of launching a startup teaches transferable skills such as meeting discipline, decision-making, accountability, channel testing, and go-to-market sequencing, so the specific idea matters less than the learning from the process itself.
“it almost doesn't matter in a way what your idea is because you're going to learn so much just by going through that process”
Engineers often equate marketing with PR, but marketing should be defined by how many units or cases of the product are sold.
“for me, marketing is how many cases out the door, how many units of your product are you selling, right?”
Companies should set an umbrella positioning that states the highest-level promise, then verticalize the message at market entry by naming the specific customer problem and proof from comparable customers.
“you have like your umbrella positioning, which is, okay, we're this whizbang, AI data science platform”
Founders should have a vision of who they will become, because a company can serve everyone in five or six years even if it must serve a single vertical today.
“you got to have a vision of not only who you are today but who you're going to be later because that helps them understand that you can be that everything to everyone in 5 or 6 years”
The kindest thing anyone did for Lynn was when a friend's mother let her live with them for a year during high school after her parents moved, which reshaped what she thought was possible for her trajectory.
“my friend's mom let me live with them for a year”
P&G marketers and product people, most of whom were chemical engineers, built large models to determine which variables had the biggest effect on units sold, and even physically modeled results, a practice Lynn says she rarely sees in Silicon Valley.
“we actually developed pretty big models to model out which variable had the biggest factor effect”
Beyond the CEO, a good investment needs a large enough market, a product that an average person can adopt and understand without many decisions, and a CEO who designs for ease of use; Casetext is an example.
“There has to be a big enough market and the product has to be designed in such a way that people can pick it up and use it”
Canvas deliberately avoids trying to invest at all stages and in all verticals, and focuses on Series A and B companies in fintech, healthcare, and marketplaces where its go-to-market team can add value.
“We know what we're good at and we're really good at fintech and health care and marketplaces”
Telling an early-stage CEO that a large prospective customer such as Walmart, Apple, or TikTok is a bad fit right now is difficult, but necessary when timelines and needs do not align.
“It's hard to talk to an early-stage CEO who has a massive company like Walmart or Apple or TikTok that wants to come work with them and tell them that that's a really bad idea right now”
Canvas formalized the go-to-market council to systematize help it was already giving portfolio companies, by identifying what drove success at companies such as LendingClub, Doximity, and Luminar and matching portfolio needs to vetted experts.
“The go-to-market council came about, and it really has always been about, but we just formalized it”
When recruiting for quantitative marketing roles, Lynn looks for economics backgrounds because many engineers are weak at statistics.
“I actually will really look for someone with an economics background because you'd be surprised how many engineers really can't do stats”
Canvas steps in when a company shows early glimmers of product-market fit but limited traction, then spends days or weeks with the CEO and team to find the best channel and hires the team to scale from zero to $10 million and then to $100 million revenue.
“we're seeing some glimmer and then we go in and spend a good amount of time with the CEO and the team”
When matching go-to-market talent, the council prioritized people who were not too senior but had seen rapid growth in their past, such as Graham Chalfant from One Medical and Allan Ng from Livongo.
“we found people that not only weren't too senior, but also had seen just a crazy growth in their past”
Asking consumers for their credit score yields a usable ballpark range because it carries little judgment, unlike questions about bill-paying or checkbook habits.
“if you ask them something more hardcore, like, hey, what's your credit score? They can kind of give you a ballpark range”
Personal interest in a topic such as quantified self is distinct from investing potential, and investors need to determine whether the company is likely to succeed.
“Just to separate and make sure the distinction between interest and investing potential is an interesting one, right? One doesn't mean the other.”
It is normal for early-stage companies to get their ideal customer wrong because they are new and may not know who the ideal customer is yet.
“everyone gets it wrong initially because you're starting a new company”
Voice interfaces have become a default expectation, since her children assume a voice interface exists and are shocked when it does not, a change she observed after investing early in Siri.
“if it doesn't have the voice interface, they're kind of shocked”
Because many companies are laying off staff, the tide is turning so that founders can now hire excellent talent, making this a good time to build companies.
“you look at how many companies they are laying people off, and that tide is turning in terms of being able to hire just amazing talent”
The gut microbiome will be the big unlock for neuroscience going forward.
“I love the microbiome. I think that's really what's going to be the big unlock for neuroscience going forward.”
Canvas formalized an informal go-to-market network it had been using to help portfolio companies, and it plans to keep doubling down on it as a core part of the firm.
“We were doing it intuitively. We were helping people solve their go-to-market issues with this network that we had, but we hadn't formalized it”
Large government spending such as the transportation infrastructure bill changes spending habits and creates investment opportunities, which she tries to find lines up with Canvas's focus.
“When you look at all the money that's just got into this transportation infrastructure bill, it's unbelievable.”
Customer life cycle and account management become critical in early-stage and later-stage companies once they have enough customers to actually manage.
“customer life cycle management, which is also incredibly important to focus on in an early-stage and a later-stage company once, importantly, you have enough customers to actually manage”