LLenny's Podcast
← All episodes
Benjamin Lauzier (Lyft, Thumbtack, Reforge)29 September 2024

How marketplaces win: Liquidity, growth levers, quality, and more

8Frameworks
15Insights

Frameworks in this episode

Insights & moments

The myth-busts, hot takes, explainers, and tools worth keeping.

Myth Buster· 1

Myth Buster27:30

Sidecar's Mistake: Giving Users Too Much Control

Sidecar, an early ridesharing competitor to Lyft and Uber, tried to differentiate by giving users total control — filters for car year, driver rating, even letting drivers set their own prices. In theory it was pro-choice; in reality it hyper-fragmented the marketplace and wrecked ETAs. Ask for a 2020-or-newer car and you might lose 10 minutes waiting, passing up a great driver in a 2018 Honda Civic. Product teams instinctively build these toggles from user feedback, unknowingly fragmenting supply.

  • Sidecar gave users filters for car year, driver rating, and even driver-set pricing
  • The theory (more choice = differentiation) hyper-fragmented the marketplace in practice
  • Filters silently carve out supply and blow up ETAs
  • Product teams build these toggles from user feedback without seeing the marketplace-health cost
  • Sidecar eventually shut down

in reality it just fragments your Marketplace like even further right and you have this like hyper fragmentation of your Marketplace

Ben Lauzier · 28:30
#ridesharing#product-mistakes#user-feedback

Hot Take· 2

Hot Take08:00

Pre-Product-Market-Fit? Forget the Marketplace Stuff

Ben's most common advice to early founders who are eager to nerd out on supply/demand ratios and economic papers: if you don't have product-market fit and a good enough growth strategy for at least one side, ignore all the marketplace dynamics. Nail the core exchange of value on one side first, and use a crutch or hack for the other side for now.

  • Founders pre-PMF get distracted by the intellectually fun marketplace dynamics
  • Go deep on one side and rely on a hack/crutch for the other side for the time being
  • Airbnb and Thumbtack both leaned on Craigslist early to jumpstart growth
  • Nail the basics of product-market fit before working the marketplace problem

if you don't have product Market fit and if you don't have a good enough growth strategy for at least one side of your Marketplace…

Ben Lauzier · 08:30
#product-market-fit#startups#founders
Hot Take21:30

A Marketplace Has Two Product-Market Fits

Ben argues product-market fit is largely independent of marketplace dynamics — you should measure it the way you would for any company (e.g. the Sean Ellis 'how disappointed would you be if this went away' test), but on both sides separately. Often founders find PMF on the demand side but the value proposition isn't compelling enough for suppliers because margins are too high.

  • Product-market fit is measurable in a traditional way, independent of marketplace dynamics
  • Use the classic Sean Ellis disappointment test on each side
  • You must have a compelling value proposition on both sides
  • A common failure: PMF on demand but not enough for supply because margins are too high

you have two product Market fits essentially you want to make sure that you have like a compelling enough value proposition on both sides of…

Ben Lauzier · 22:30
#product-market-fit#supply#demand

Explainer· 6

Explainer03:30

What Actually Makes a Company a Marketplace

Ben defines a marketplace as two or more distinct sides that provide value to each other, with an intermediary facilitating the exchange in the middle. How involved that intermediary is defines how 'managed' the marketplace is, on a spectrum from hands-off (Craigslist) to semi-managed (Lyft), where the platform significantly shapes the transaction.

  • A marketplace is two or more distinct sides that provide value to each other, plus an intermediary in the middle
  • The company does not own the supply — that's what separates a marketplace from just selling stuff
  • How involved the intermediary is defines how 'managed' the marketplace is
  • Craigslist is super hands-off/unmanaged; Lyft is semi-managed and shapes the transaction

it's two or more sides that are distinct uh you know from one another and they provide value to each other and then you have…

Ben Lauzier · 03:30
#marketplaces#definitions#managed-marketplaces
Explainer10:30

Jumpstarting Supply: 'Play One-Player Mode'

A common early tactic is to tap into an existing channel where one side of your marketplace is already latent — Ben calls this 'play one-player mode.' Countless businesses were built off Craigslist this way. Other levers: leverage job boards, and build value-added services early to retain supply (like OpenTable did with restaurant services).

  • Tap channels that already have one side of your marketplace latent
  • Thumbtack quietly posted jobs to Craigslist to pull in browsing contractors
  • Lyft and Thumbtack leveraged job boards to source supply
  • Build value-added services early to retain supply (e.g. OpenTable's restaurant tools)

play one player mode is uh what it's Al also called sometimes but you know try to find a way to tap into existing channels…

Ben Lauzier · 10:30
#supply#growth-tactics#cold-start
Explainer16:30

Liquidity Is How Marketplaces Win

Liquidity is your ability to match buyers and sellers efficiently — how quickly people find what they're looking for. Ben pictures it as a Venn diagram: one circle is what supply wants to sell, the other is what demand wants to buy, and liquidity is the overlap. It's a direct multiplier on marketplace efficiency and the ultimate engagement loop: more supply means more choice, more transactions, more repeat use.

  • Liquidity = ability to match buyers and sellers efficiently
  • Think of it as the overlap between what supply wants to sell and what demand wants to buy
  • For Lyft/Uber: of everyone who opens the app intending to book, how many actually get a ride
  • It's the ultimate engagement loop — more supply drives more choice, transactions, and retention

to me liquidity is how marketplac win right it's it's this measure of your ability to match buyers and sellers efficiently right

Ben Lauzier · 16:30
#liquidity#metrics#network-effects
Explainer18:30

Find the 'Market Health' Metric That Predicts Liquidity

Liquidity itself (fill rate of intentful demand) is a lagging output metric distorted by exogenous factors like weather and competition. Ben prefers a more actionable 'market health' metric: the best proxy that predicts liquidity. At Uber/Lyft it was ETA — if the closest driver was three minutes away or closer, you'd almost certainly book; beyond that you'd start checking alternatives. A supply team can then act against that predictor directly.

  • Fill rate of intentful demand is the true output metric but is laggy and noisy
  • A 'market health' metric is the best predictor of liquidity and is far more actionable
  • At Lyft/Uber that predictor was ETA
  • Under ~3 min ETA you convert; at 5 min you check Uber, walk, or take the bus
  • Supply teams can measure whether adding 100 drivers actually reduces ETA

and for Uber it was etas uh so we knew that if we had uh if the closest driver was at least three minutes away…

Ben Lauzier · 19:30
#metrics#liquidity#operations
Explainer24:30

Three Signs an Idea Is Right for a Marketplace

Ben names three signals that a marketplace model fits an idea: high fragmentation (a long tail of buyers and sellers, no handful of big players), a relatively uniform set of needs so supply can be commoditized, and a high enough barrier in matchmaking or vetting. Services marketplaces like Thumbtack are tricky because supply is fuzzy — an electrician only wants certain jobs, only if available that day, and may cancel for something better.

  • High fragmentation: a long tail of buyers and sellers, no dominant players
  • Uniform needs so supply can be commoditized (eBay's clean inventory vs Thumbtack's fuzzy supply)
  • A high barrier in matchmaking/vetting — the harder it is to find and vet each other, the bigger the opportunity
  • Nobody says 'I'm building Airbnb for X' — the model should fit the problem, not the reverse

one high fragmentation I think you want this long tale of buyers and sellers without a handful of big players controll in the market

Ben Lauzier · 24:30
#marketplaces#business-models#idea-validation
Explainer32:30

The Three Biggest Reasons Marketplaces Fail

Ben names three failure modes. One: failing to reach liquidity — running out of time or money before hitting enough density on both sides. Two: ignoring one side and operating too long as a one-sided business (running demand ads, treating supply as an afterthought); marketplaces are laggy, so network effects die before you notice. Three: quality — being a marketplace implies curation, and there's a constant pull to lower the bar for more supply.

  • Failing to reach liquidity/density before running out of time or money
  • Ignoring one side — running only the demand funnel and treating supply as an afterthought
  • Marketplaces are laggy, so you realize you've neglected a side too late
  • Quality: being a marketplace implies curation; resist lowering the bar just to add supply

marketplaces are very laggy so once your network effects start to die down in terms into this moment of panic of oh shoot we forgot…

Ben Lauzier · 34:00
#marketplaces#failure-modes#quality

Story· 5

Story30:30

The Smoke Machine Problem: Filter vs. Ranking

At Thumbtack, a checkbox let people booking wedding DJs request a smoke machine — and lots of people enthusiastically checked it. But only 5% of DJs had one, so the filter silently carved out 95% of supply. When asked, users said they didn't actually care that much. The lesson: make preference checkboxes affect ranking, not hard filtering, so you honor the preference without destroying liquidity.

  • A 'smoke machine' checkbox for wedding DJs was popular with users
  • Only 5% of DJs had one, so the filter removed 95% of supply
  • Users admitted it wasn't actually a deal-breaker
  • Fix: let the checkbox affect ranking, not filtering — be smart enough to know it's not a deal-breaker

yeah hell yeah I want a smoke machine in my wedding and unknowingly to them obviously like only 5% of our DJs had a smoke…

Ben Lauzier · 31:00
#thumbtack#product-design#supply
Story37:00

When Control Backfires: Thumbtack's Direct Bookings

Thumbtack originally sold leads to pros, only a fraction of which converted. To improve pro ROI (~20%), they shifted to selling direct bookings — and pros hated it. Subconsciously they liked the 'thrill of the sale' and the customer contact, and overestimated their own closing ability. No matter what the data said, the pros didn't feel it. Ben's takeaway: any attempt at control can be tricky and backfire unpredictably, plus in the US controlling supply risks employee reclassification.

  • Thumbtack moved from selling leads to selling direct bookings to lift pro ROI ~20%
  • Pros hated it — they liked the 'thrill of the sale' and customer contact
  • Pros overestimated their closing ability; being kept busy felt like hustling
  • Data saying earnings rose 20% didn't change how pros felt
  • In the US, controlling supply risks legal reclassification as employees

and Pros hated it uh they hated it because they actually subconsciously like they like the spill of the sale right they love this contact…

Ben Lauzier · 38:00

any attempt at control can be really tricky and backfire in ways that are unpredictable

Ben Lauzier · 39:00
#thumbtack#managed-marketplaces#supply
Story40:30

How Toptal Upholds Quality: The 3% Pass Rate

Ben cites Toptal as a talent marketplace (mostly engineers and designers) that upleveled quality without becoming fully managed — claiming only the top 1-3% of talent via rigorous checks and processes. A funny detail: their advertised 3% pass rate is reportedly higher than the real number, because they feared that stating the true, lower figure would sound fake and discredit them. This only works when supply is so abundant it competes to join.

  • Toptal claims only the top 1-3% of talent, mostly engineers and designers
  • Heavy vetting up front plus coaching and education to maintain quality
  • They advertise 3% because the real pass rate is even lower and would sound fake
  • Only viable when supply is abundant and eager to join the platform

they advertise something like maybe like a 3% uh you know sort of a pass rate for their talent so they only like on board…

Ben Lauzier · 41:00
#toptal#quality#talent-marketplaces
Story42:30

How Lyft Built a Rental Company to Manufacture Supply

After GM invested half a billion in Lyft, Ben got a Christmas Eve call and they decided to build a rental company. GM had off-lease vehicles it was forced to auction; Lyft had a huge supply gap because ~50% of job seekers and welfare recipients in the US don't own a car. By renting cars, Lyft could manufacture its own supply — dialing it up or down surgically by market and price. In three months they built it; within 18 months it was the fourth-largest rental fleet in the US, with highly loyal drivers who could be paid for the car only if they didn't drive for Uber and drove 30+ hours a week.

  • GM invested ~$500M and had off-lease vehicles it couldn't offload
  • ~50% of US job seekers and welfare recipients don't own a car — Lyft's supply gap
  • Renting cars let Lyft manufacture supply surgically by market and price
  • Fourth-largest US rental fleet within 18 months
  • Drivers stayed loyal — car paid for only if they didn't drive for Uber and drove 30+ hrs/week

by renting cars we could essentially manufacture our own Supply right we could dial this up and down we could be very surgical about like…

Ben Lauzier · 43:00
#lyft#supply#managed-marketplaces
Story46:30

Lyft's Mentor Program: Competing With Uber at 1/10th the Cost

In 2014-15, Uber was ~30x Lyft's size, so Lyft had to be 10x more efficient per person just to survive. Instead of Uber's approach of opening offices for group onboarding, Lyft paid its best drivers $35 per 'mentor session' to inspect cars, check documents, and take new drivers on a test ride. The benefits were huge: mentors were brand evangelists who shared real tips (drowning out the company's marketing emails), it was massively scalable, and being a mentor became a recognition and earnings lever that boosted retention of top drivers. This let Lyft match most of Uber's footprint at a tenth of the resources.

  • Uber was ~30x Lyft's size, so Lyft had to be 10x more efficient per person
  • Lyft paid its best drivers $35 per mentor session to replace office-based onboarding
  • Mentors were brand evangelists who shared real tips that beat the company's own marketing copy
  • Being a mentor was a recognition and earnings lever (~$70/hr) that boosted top-driver retention
  • Matched most of Uber's footprint at a tenth to a twentieth of the resources

we would pay our best drivers $35 per Mentor session and a mentor session was essentially replacing this onboarding flow

Ben Lauzier · 49:00

this this allowed us to match most of Uber's footprint with like you know a tenth or a 20th of the the resources at you…

Ben Lauzier · 51:00
#lyft#supply#growth-tactics#onboarding

Takeaway· 1

Takeaway09:30

Pick the Hardest Side — It's Supply 80-90% of the Time

When starting a marketplace, focus on the side you have no idea how to grow. Teams intuitively know which side is hard — often they can get demand but can't source supply. Ben says supply is the hardest side 80-90% of the time, with rare exceptions like Rover (dog walkers) and TaskRabbit (taskers) where supply came easily.

  • Pick the hardest side — the one you have no reliable growth strategy for
  • Outsource, subsidize, or hack the easy side; build a real engine for the hard side
  • Supply is the hardest side roughly 80-90% of the time
  • Counterexamples where supply was easy: Rover (dog walkers) and TaskRabbit

yes I would say um Supply is the hardest side maybe like 80 to 90% of the time

Ben Lauzier · 12:30

focus on on the side that you have no idea how to grow like that's that you should have a reliable growth strategy for for…

Ben Lauzier · 12:00
#supply#demand#growth-strategy