Lydia Jett - Investing in E-commerce - [Invest Like the Best, EP.282]
What this covers
<invoke name="think"> </invoke> Lydia Jett of SoftBank Investment Advisers talks with Patrick O'Shaughnessy about e-commerce and about her own change as an investor alongside Masayoshi Son. She leads SoftBank's consumer and internet investing and has sat on the boards of Coupang and Flipkart. The conversation moves between country-level market structure, what separates e-commerce companies that outgrow their rivals, and what Vision Fund 1 taught the team about capital and founders.
The market-structure section is where the specifics are. Jett says she first assumed US or China models would carry over to India, Korea, and Indonesia, then found that penetration differs sharply by country. Korea is under-retailed, with big offline chains acting more like landlords to small merchants, which she calls almost a perfect setup for online. Most Western investors passed on Coupang because they thought Korea was too small. She also notes that users in China, Indonesia, and India prefer dense, many-service landing pages over the clean US interface.
On company building, she argues e-commerce starts from a worse position than offline because delivery margin is negative, so vertical integration matters: owning the whole chain lets a company redesign packaging, van density, and returns. She holds that efficiency has to be proven first, on a small scale and measured by contribution margin. Her harder claims are contested: she says Uber, Alibaba, and Amazon do not offer a replicable scale-first template, that debt suits volatile early-stage companies badly, and that venture outcomes owe more to luck than public or private-equity outcomes do. She also says investors who demand a large TAM misunderstand how consumer companies begin. Other threads cover founder selection, the value of an early finance-minded right-hand person, and why every investment's forecast turned out wrong.
Jett argues that durable e-commerce returns come from tailwinds, vertical integration, and efficiency built before scale, and that SoftBank's abundant capital over-funded growth and destroyed the discipline that makes these businesses work.
- E-commerce penetration will keep rising in every economy, but local structure (Korea's under-retailing, fragmented US logistics, Asian trust gaps) determines where and how it wins
- Owning the full fulfillment stack (Coupang) beats marketplace models, but only when efficiency is proven before scaling
- Over-capitalization in Vision Fund 1 caused lack of discipline, fragile businesses, and bloat, and the lessons point to strategic finance, unit economics, and small-scale proof
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.
Investors who require a big TAM misunderstand how great consumer companies start, because none had a big established TAM at launch; if a large market existed, someone would already have captured it, so the better question is what would have to be true.
“I can't even think of a big consumer company that had a big established TAM when it got started. It doesn't logically make sense. If there was a big TAM, somebody would've eaten it”
Venture capital is driven far more by luck and good fortune than public-market or private-equity investing, because it is a people business where company success depends on aggregating talent, which is very hard to predict.
“it is so much more driven by luck and good fortune than traditional investing is, public market investing, private equity investing”
Models that worked in the US or China cannot simply be transplanted to India, Korea, or Indonesia; countries differ in drivers and penetration rates, with China and the US highest, Korea next, and India, Indonesia, and Russia very low.
“you could just take models that worked in the US or in China, and you'd transplant them to India or Korea, Indonesia, and you'd see very similar curves”
Jett initially assumed US or China e-commerce models would transplant to India, Korea, and Indonesia with similar adoption curves, but found that countries differ in drivers, so penetration rates vary dramatically.
“you could just take models that worked in the US or in China, and you'd transplant them to India or Korea, Indonesia, and you'd see very similar curves”
Even though no companies in emerging markets are bigger than banks or mining companies today, SoftBank bets that the value transition to technology will be clear because consumer benefit is so much greater, so technology companies should become bigger than traditional industry.
“We're willing to make those fundamental bets that the technology companies should be bigger than what we've seen in traditional industry before”
Uber, Alibaba, and Amazon are often cited as examples of scaling first, but their success depended on special circumstances in time, and it is very hard to replicate them in a market flooded with capital.
“People love to use examples of Uber or Alibaba or Amazon, but the reality is, it's a very special set of circumstances in time that allow to build those companies”
Putting debt on the balance sheet of a fast-growing early-stage business is hard because the business is too volatile, and covenants will likely lead to default, so venture-style capital should not be structured like debt for these companies.
“it's really hard to put debt on the balance sheet of a fast growing early stage business. Their businesses are still too volatile”
Coupang drives toward efficiency first before scaling, whereas scaling first and then trying to push efficiency through a large organization is next to impossible, which is why Western companies building logistics networks struggle to operate them.
“They really drive towards efficiency first before they scale, as opposed to scale and then try and figure out how to push efficiency through a big organization, which is next to impossible”
E-commerce businesses start from a worse position than offline ones because offline giants like Walmart and Target run single-digit margins efficiently, while digital businesses have negative delivery margin, are human-capital-intensive, and lack automation, so they must reach significant scale.
“once you get past product margin and then delivery margin is negative, you have to get to significant scale”
Venture capital outcomes are more driven by luck and good fortune than public or private equity investing, so the people termed great investors are often those who have stayed longest, and longer tenure makes investors humbler because their impact is smaller than they think.
“the people we term to be great investors often are the people who've been standing the longest”
Margin viability should be proven first on a small scale, such as within one geography or with one customer, using contribution margin, because the business should improve from there; layering adjacencies onto a core model that does not yet work sets up structural pain.
“It is first and foremost, can we prove out margins on small scale? So can we prove it out in a specific geography? Can we prove it out against one specific customer?”
Consumers will not stick with a company that is only marginally better, so the key diligence question is whether a business can improve the customer experience by a factor of five or ten times or meaningfully change its economic value proposition.
“What levers do you have to be able to pull to either improve a customer experience by a factor of five or 10 times? Because a consumer isn't going to stick with you because you're marginally better”
Son can pivot a large organization quickly when he believes a theme is fundamentally changing, whereas traditional investors are sclerotic and bound by covenants with LPs and multi-fund strategy commitments.
“Organizations tend to be pretty sclerotic. You say, "This is who I am." It's why you don't see that much change in venture outside of new entrance. Masa's the opposite”
Korea is a very under-retailed geography, where offline big retailers function more like real estate companies leasing space to small merchants, leaving a limited skew count and high prices, which makes it almost a perfect setup for e-commerce.
“They're more like real estate companies that are leasing space to small merchants. You have a really under retailed economy with a very limited skew count with very high prices, which is almost just a perfect setup for e-commerce”
The LTK model works by taking a very small cut from creators, which funds their small businesses, while also serving retailers as a distribution partner that provides advertising budget, and matching both sides is what makes the economics work.
“how do you take a small cut, very small cut, almost the infrastructure layer like FinTech assets have established, from creators that allows them to build their own small businesses become economically viable”
Owning the full chain lets a company rethink packaging to increase van density, eliminate boxes, and redesign returns networks, improvements that are very hard to achieve when focused only on last-mile delivery, sortation, or fulfillment in isolation.
“It really is that integration that allows you to rethink form factor, it allows you to rethink experience”
Vision Fund 2 invested earlier at about a quarter of Vision Fund 1's check size, mostly in Series B companies, whose founders had raised on charisma and vision but had not learned that they are making financial commitments to later investors to hit a plan.
“series B companies had only ever really raised in the past by off the basis of that founder's charisma”
Jett estimates that if the portfolio produces about 20 exceptional founders out of 300 who build $20 to $40 billion companies, the strategy would be a success, because few people can operate across vision and quantitative detail.
“if we walk away from this grand experiment with 20 exceptional founders out of a 300 founder portfolio”
Founders should invest early in a financially oriented right-hand person who can help them see around capital allocation corners and read the balance sheet, because many board members, despite being visionary, pay little attention to balance sheets.
“no one knows how to read a balance sheet, no one's paying any attention to the balance sheet of these businesses”
In every great investment, the underlying drivers have not mapped to the forecast built at the time of investment, so the team has been surprised every time, which means the forecast is a starting bet rather than a predictor.
“in every great investment, the underlying drivers and how they have operated have not mapped to the forecast that we built at investment. We've been surprised in every case”
Consumers in China, Indonesia, and India prefer a single dense landing page offering many services, while the US has moved toward clean UI, so the Western aesthetic was wrongly assumed to be universally correct.
“countries really establish a consumer preference in a single landing page that offers them all of the consumer services”
Every decision a founder makes will have unintended consequences in about three directions, and brittle businesses are those whose leaders do not pay attention to the consequences coming their way.
“every decision you make as a founder will have an unintended consequence in probably three”
Video-based content drives much higher engagement, and businesses using Firework's video integration see conversion rates that are a significant multiple of non-video purchasing, even though video commerce in the US will not move as fast as in China.
“The conversion rates that we're seeing are significant multiple of what they're seeing for the non-video based purchasing”
SoftBank's consumer e-commerce team tends to back founders who are more often immigrants with a wider world view, sometimes from China or India, because they are naturally inclined to look outward and model trends from abroad to the US.
“a founder that more often than not is an immigrant, does have a wider world view”
SoftBank initially believed Alibaba, Baidu, and Tencent would capture most of the Chinese technology market, but it was wrong, because the scale of ambition and velocity of company building in China was far beyond expectations.
“We believed that we would accrue the value of China through our position in Alibaba. And we thought between Alibaba, Baidu, and Tencent. That would be most of the Chinese technology market. And man, were we wrong”
The last few years amounted to the Vision Fund in distributed form, as crossover funds, corporates, LPs, individual investors, syndicates, and SPVs all piled into late private rounds, meaning hundreds of players were doing what Vision Fund 1 did alone.
“The last couple years have basically been the Vision Fund in distributed form”
Most Western investors passed on Coupang because they believed Korea was not big enough, and Coupang then pitched Korea's size to satisfy them, a misjudgment Jett believes reflects pattern-matching to developed economies.
“most Western investors had passed on the opportunity with the fundamental belief that Korea just wasn't big enough”
Across Asia the creator model that emerged was a long tail of influencers who are good at sales, placed in front of consumers to speak authentically about product quality, which is less relevant in the US because of structural differences in trust.
“It was effectively taking people who were just good at sales and putting them in front of consumers with product to authentically speak about the quality”
E-commerce is a beautiful economic flywheel because consumers are poorly served by a fragmented long tail of offline retail, and moving online creates efficiencies that deliver more product at lower prices.
“it is a beautiful economic flywheel that the consumer globally today is not well served by a really fragmented long tail of offline retail”
Marketplaces that list long-tail merchant assortment without handling fulfillment deliver a subpar experience compared with a fully owned channel like Coupang's, so Coupang should keep compounding market share on speed, assortment, and price.
“That is fundamentally a subpar experience to a fully owned channel like Coupang has built and invested and developed in”
In the US, speed is limited by separate supply chain operators like FedEx, UPS, USPS, and 3PLs, so the key question is integrating those operators through technology to remove handovers, whereas Coupang in Korea has a fully integrated supply chain.
“how do you start to integrate more of that supply chain and drive through technology in a way that can just shave some of the handover between these separate operators”
Live and social commerce in the East is powerful because of a lack of high-quality physical retail, a lack of trust in existing brands, and a culture of people selling directly to consumers, conditions that are less present in the US, so live shopping probably will not replicate there in the same way.
“there is a lack of the high quality physical retail and there's a lack of trust with existing brands. I think there's a culture of people selling to you in markets that we don't have in the US”
Coupang's vertical integration was prefunded with about $3 billion of SoftBank capital, and it cannot succeed unless the whole organization focuses on it, because building fulfillment and logistics networks is so hard.
“We had $3 billion of capital put into the ground on the thesis that it was the right thesis”
Son's simplified slides and aggressive build style are instinctively aligned with generating fund returns, because a fund of this diversified portfolio needs each company to return more than two to three times its investment.
“you can't build a fund of this diversified set of companies without believing that you're going to generate more than two to three times return on each individual company”
Vision Fund 1 with about $100 billion backed more than 80 companies at an average check of $600 million, and the widely held belief that too much capital causes lack of discipline, fragile businesses, and bloat was largely correct.
“Too much capital causes a lack of discipline, which causes fragile businesses. It causes bloat”
Growth equity investors start by examining core economics, including unit economics, cohort retention, and how engagement changes over time, and they expect companies to organize data early and track what levers matter against forecast.
“We are always spending time looking at unit economics. We are always spending time looking at cohort retention”
Masayoshi Son distilled that tailwinds, meaning large fundamental technology shifts, are what actually protect businesses, so the fund should only invest in businesses riding tailwinds.
“it is tailwinds that fundamentally provide protection for businesses. So we should only ever be investing in businesses with tailwinds”
Capital providers over the last few years made a mistake by focusing companies too much on revenue growth, which built large, inefficient companies that are very hard to fix later because margin improvement requires cultural and personnel shifts.
“I think we have made a mistake, those of us in capital provider seats, over the last handful of years in focusing companies too much on revenue growth”
As companies gain scale, negotiating power improves, and combined with operational efficiencies this became a bigger lever than expected for turning unit economics positive, a result Jett says was surprising even though it looks obvious in retrospect.
“as you amass scale, your negotiating power gets better in combination with the operational efficiencies which are core to the heart and lifeblood of these companies has actually been a bigger lever”
Vision Fund 1 backed capital-intensive businesses on the premise that venture had never funded them but SoftBank had enough money, and without strong operational controls these companies consumed capital very fast.
“without very strong operational controls, it is a beautiful way to consume a lot of capital fast”
E-commerce CEOs must be extraordinarily disciplined because they start with impaired margins, so they must focus on a few things that matter rather than being distracted by adjacencies.
“you have to be really disciplined because you are starting from a position of having impaired margins or an impaired economic profile”
Technology has caused a broad decay of fee streams in financial services, so Jett prefers companies that can extract some economics onto their own balance sheet rather than passing all of it back to customers.
“you've just had a big, broad decay of fee streams into these organizations, which is what technology has brought”
The most extreme efficiency pickup Jett has seen is in product cost, because COGS is the majority of cost structure for these businesses, so negotiating leverage at scale matters more than any single operational fix, which tends to be incremental.
“it has been in extracting further margin on product. It's just negotiating leverage when you gain scale. That's because the businesses I work with fundamentally, their COGS end up being the majority of their cost structure”
Jett screens investments on three questions: whether the company is on the curve of growth, whether it has identified and defended its market position, and whether it is fundamentally improving its business model over time.
“Are you on the curve of growth, have you identified your position in market, and are you fundamentally improving your business model over time”
Jett rejects the common founder narrative that traditional operators such as insurers and retailers are inefficient, arguing they are extraordinarily good and quantitatively adept, so new entrants must compete against strong organizations.
“I actually think these companies are extraordinarily good. They've built very, very efficient businesses”
Ant Financial's $14 billion raise in 2018 was a watershed because it was the first time there was a list of more than a page of investors who could write a $200 million check, showing the market had grown to absorb very large checks.
“when Ant Financial raised $14 billion back in 2018. That was the first time I'd seen a list of investors that was more than a page long who could write a $200 million check”
SoftBank's aggressive model is not the right fit for the vast majority of companies, and neither is venture capital, so founders must know who they are and what they target because a SoftBank ride narrows their options.
“SoftBank is just not the right fit for the vast majority of companies. And I don't think venture capital necessarily makes sense for the vast majority of companies”
Founders increasingly need to see their company as a product, with investors as buyers, especially as capital becomes more expensive and scarce, which requires a clear financial narrative in addition to product and distribution.
“your company is a product and investors are the buyer in many ways”
There is rarely a magic number that growth investors want; they evaluate a company in the context of the overall picture, team, and product, and they need evidence that the company can compare to other assets through confidence in its numeric orientation.
“there is just very rarely a magic number that we're looking for”
Founders who openly share mistakes, learnings, and changes build the most trust with investors, while founders who stick rigidly to metrics they promised cannot build the relationship needed in a down market.
“the founders who lean in and share the mistakes and the learnings and how they're changing are the founders that I build the best trust with”
During the pandemic Amazon was not capable of fulfilling its promise, which created an opening for others such as Shopify's full-stack approach and Walmart's rethought fulfillment to invest meaningfully.
“what we saw in the pandemic was Amazon not capable of fulfilling that promise that they've been really good at fulfilling”
SoftBank spent significant time in 2021 on food because it is among the least penetrated categories, with proven economics abroad, and in the US only prepared food delivery through DoorDash or Uber Eats has been adopted, so food purchasing will change over the next decade.
“We've really only done that in the construct of prepared food delivery, i.e., DoorDash or Uber Eats”
DoorDash has shown that consumers are willing to pay far more for delivered food than anyone expected, and the consumer demand stretched further than anticipated, which was an extraordinary learning.
“DoorDash has shown that consumers are willing to pay far more than anyone thought they would for delivered food”
Son grants trust freely to his investing team and stands up publicly when he is wrong, as seen in WeWork and recent earnings calls, which lets founders think big because people operate from optimism rather than fear.
“he is the first person to stand up when he makes a mistake and say "I was wrong." We see it over and over in our earnings calls. We saw it in the case of WeWork”
One company decided it was subscale in six regions and chose six hubs to scale, planning to rationalize along the way, which illustrates how a unit-negative business can be shaped by picking a few locations to reach scale.
“we are subscale in six regions that we have determined to have the ultimate ability to become scalable. And we've decided these are our six hubs”
Coupang in Korea has a fully integrated supply chain, and Flipkart and Amazon India have a similar setup, while the US and other established economies do not yet have that degree of integration.
“Korea where you have a fully integrated supply chain in the context of Coupang. You have the same very similar setup in India with Flipkart and with Amazon India”
Every business is messy, including large corporations like SoftBank, Amazon, and Google, so leaders must identify what really matters and focus on it, accepting that they will get some things wrong.
“Every single business I work with is a mess, including big corporate businesses, including SoftBank, including Amazon, including Google”
The kindest thing Jett experienced was the many people who gave her feedback, both positive and negative, which built her confidence and taught her to pivot, and she now tries to give feedback to her own team.
“every person along the way who pulled me aside and gave me feedback, positive or negative”
Despite tough headlines, SoftBank's brand is extraordinarily strong with founders, who value Son's trust and belief in them and his focus on building big things, creating an intoxicating position for the firm.
“I feel like our brand has been extraordinarily strong with founders.”
Apple and Google security changes are impairing the customer relationship and conversion, and video commerce could recapture that lost conversion and surpass previous levels.
“Apple's security changes and Google security changes are impairing a customer relationship”
Modern influencer networks let retailers specify the demographic, region, and age of the influencer they want, replacing broad offline media, which only targeted by city or region at best.
“What these new networks allow is for you to actually quantify and qualify what you're looking for in the influence you work with”
A company can be dramatically unit-economics negative while being subscale in several regional hubs it has chosen to scale, and it rationalizes along the way, which makes operational efficiency the core challenge.
“we are subscale in six regions that we have determined to have the ultimate ability to become scalable”
Jett does not expect US video commerce to reach the same velocity as China, because the forcing factors that drove China's adoption are not present in the US.
“Do I think we're going to see the incredible velocity of video commerce in the US that you've seen in China? Not to the same degree. I don't think they're forcing factors are the same”
Operational efficiency gains in e-commerce come from daily small improvements that compound over a 48-month period, rather than from single dramatic moves, because the businesses face constant crises in human capital, inflation, and staffing.
“It is the cost, the hard one cost of every day getting incrementally better in very small ways and then getting to look back at a 48 month period”
Son has historically focused on technology improving efficiency and driving down total cost of producing goods, and this has generated the enterprise value seen across the technology sector.
“technology improving efficiency and therefore driving down the total cost of producing goods”
Healthcare is complicated because incentive systems are illogical and payer-provider dynamics are messy, so SoftBank tried to avoid that relationship and instead addressed fundamental needs such as loneliness.
“The incentive systems aren't logical in any rational way. There's not only regulation of course, but there are payer provider dynamics that are just messy”
Jett's operating principle is to understand the market you are in because you have to learn how to operate within it, and Son is able to build conviction in big ideas supported by fundamental economic tailwinds.
“understand the market you're in because you have to learn how to operate to it”
Despite very different current penetration rates, e-commerce penetration should continue rising everywhere because the economic benefit to consumers is significant, and the real question is how much more value can be added rather than whether the transition is finished.
“the question to me is not are we done, but how much more value can we continue to add. I don't see a stopping point realistically”
Investors should spend substantial time defining what moat a business has and how it sustains market position in a volatile market flush with cash, and the aim is funding models that improve through automation rather than decaying the economic curve.
“we're making sure that we are funding business models that are fundamentally improving through automation and technology, as opposed to just decaying the economic curve”
Experience is the most obvious competitive frontier in e-commerce, and it is really hard to compete against speed across every region, though in value-driven economies like India and Indonesia, low-cost assortment matters more.
“It is really hard to compete against speed. I think that is generally true across every region I look at”
The key question for the creator economy is whether an economic model can support a long tail of fragmented creators, and SoftBank made a significant investment in that category after exploring it.
“is there an economic model that supports a really long tail fragmented set of creators in an exciting way”
Cameo is a conceptual analog to LTK, but Cameo is one-to-one content for a single fan, while LTK lets an influencer speak to an entire network, so the scalability question is reaching a broader audience than one-to-one.
“Cameo of course is a one to one model to my understanding”
SoftBank's roughly $900 million investment in Fanatics, one of its first Vision Fund investments, is a validation of the core strategy because the management team built extraordinary assets and a real moat.
“we put almost $900 million into the company. And that is an extraordinary mature management team that has built an extraordinary set of assets and a real moat”
Jett believes it is less obvious that Amazon will be everything to everyone, which makes the current period a more interesting time for innovation in e-commerce.
“I think this is a less obvious point in time that Amazon is going to be everything to everyone”