Marketplace Growth: When to Push Supply vs. Demand

Most marketplace founders are burning money on the wrong problem.

You're trying to grow both sides at once — pouring budget into supply acquisition while simultaneously chasing demand — and watching qualified users bounce because the marketplace is empty, or suppliers churn because you can't deliver volume. The classic "chicken-and-egg" problem. But here's what nobody tells you: it's not actually a chicken-and-egg problem. It's a sequencing problem. And if you get the sequence wrong, you will light money on fire.

I've scaled five two-sided marketplaces from zero to eight- and nine-figure outcomes. The playbook isn't complicated, but it is precise: build density-first supply, seed the harder side, geo-fence your launch, then unleash targeted demand generation. And steer the entire operation by one metric — match rate — because it's the only number that tells you whether both sides will actually stick around and compound.

Third-party sales are projected to account for 59% of all global e-commerce by 2027, per Sharetribe, and the marketplace model is growing at a 10.6% CAGR through 2035 according to Market Research Future. The model works. But most founders die in the cold-start phase because they're optimizing for the wrong metrics and chasing both sides with equal, unfocused fervor. Let me show you how to avoid that.

The Cold-Start Problem Is a Liquidity Problem

An empty search result is a death sentence.

When I joined Shiftgig as Growth Marketing Leader, the company was scaling two businesses at once: a worker marketplace (B2C) and an enterprise staffing platform (B2B). If there weren't enough workers in a market, businesses would bounce after one bad experience. If there weren't enough businesses hiring, workers would uninstall the app and never come back. We couldn't afford to launch nationally and hope liquidity emerged organically — we had to engineer it, city by city.

Here's what we did. We identified the harder side — in our case, business demand (B2B) — and built supply density first. We used the scalable, lower-cost channels (Facebook, SEO, PPC) to grow the worker side to over 1 million users, then geo-fenced the launch market by market so we never opened a city to businesses until we had enough workers to guarantee fulfillment. We cut B2B acquisition cost in half with targeted email and lead gen, and we honestly subsidized early supply where we had to so the demand side never saw an empty marketplace.

The result: the team scaled from $0 to over $60 million in ARR, grew to 1 million+ users and 21,000+ businesses (including 1,200+ mid-market and enterprise accounts), and built the marketing function from 1 to 40+ people. But the real unlock wasn't the total user count or the GMV — it was that we filtered every decision through match rate: the share of demand-side requests we could successfully fulfill.

Because match rate is what makes a marketplace compound. Users don't stay for your brand story; they stay because the marketplace works every time they need it.

The Cold-Start Playbook: Build Density Before You Open the Floodgates

Most founders try to grow both sides simultaneously and end up with a marketplace that's half-functional everywhere and fully functional nowhere. A marketplace fully functional in 5 cities beats one half-broken in 50.

Here's the tactical playbook I've run multiple times:

Identify the harder side.

In most marketplaces, supply is harder to acquire than demand. It's easier to get consumers to browse than it is to get businesses or service providers to commit inventory, time, or labor. But not always — in some cases (like Airbnb in its early days), you need to prove demand first to convince supply to join. The harder side is whichever one requires more trust, more onboarding friction, or more upfront commitment.

At Shiftgig, the worker (B2C) side was easier — we could scale it with Facebook ads, SEO, and PPC because the cost per acquisition was lower and the volume was higher. The business (B2B) side required longer sales cycles, more trust-building, and higher touch. So we built worker supply first.

Subsidize the harder side honestly — and get specific about how.

Uber famously paid drivers even when they had no riders to ensure sufficient supply density before opening a city to rider demand. This isn't a trick; it's an investment in liquidity. You're buying time to prove the marketplace works.

At Shiftgig, we took a multi-lever approach to building worker density in Chicago before opening the business side:

Paid acquisition with geo-targeted creative. We ran Facebook and Instagram campaigns aimed specifically at hourly workers in Chicago — service, retail, hospitality, warehouse — with messaging that spoke to shift flexibility and same-day pay. We tested dozens of creative variants and optimized for cost per completed profile, not just app install, because an incomplete profile was worthless supply.

Referral incentives for early supply. We offered cash bonuses to workers who referred other workers, but only if the referred worker completed at least one shift. That kept the program from attracting freebie-seekers and ensured we were building real, engaged supply.

Subsidized early shifts. In the first few weeks of a new market, we occasionally guaranteed a minimum hourly rate to workers even if the business demand wasn't there yet. We absorbed the cost so workers had a positive first experience and would come back when real demand ramped up. This was temporary and targeted — we didn't run it indefinitely — but it bought us the time to prove the marketplace worked.

Manual outreach to high-density worker pools. We partnered with community colleges, workforce development programs, and local staffing agencies to recruit workers in batches. This wasn't scalable long-term, but it got us to critical mass faster than relying solely on paid digital.

The key move: we never opened the business side in a market until we had enough worker supply that we could promise a 90%+ fill rate on every shift request. That threshold became the forcing function. If we didn't have it, we didn't launch demand. Period.

Geo-fence the launch and earn density city by city.

National launch sounds ambitious, but it kills liquidity. Instead, pick one market, get supply density right (enough suppliers that demand-side users see multiple high-quality options for every search), prove match rate is high, then open the next city.

Grubhub did this brilliantly when they entered new markets. They would manually gather local restaurant takeout menus, scan them for their website, and call restaurants to map delivery boundaries, even if the restaurants weren't initially online with Grubhub. They built comprehensive supply in one market, owned demand, then moved to the next. That manual, non-scalable work created the perception of completeness — when a user searched, they saw every restaurant that delivered to them, which made Grubhub the default discovery tool.

At Shiftgig, we used this exact sequence. We launched Chicago first, built worker density with paid acquisition, proved match rate was consistently above 90%, then opened the business side in that market. Only after Chicago was liquid did we move to the next city. The result was sustainable growth instead of a national launch that collapsed under its own weight.

Seed your own supply if you have to.

Airbnb's co-founders famously went door-to-door in their early markets, photographing listings and onboarding hosts in person. It didn't scale, but it created just enough supply that the first wave of demand-side users had a real marketplace experience. That manual hustle bought them the liquidity they needed to kick-start the network effect.

If you can't afford to subsidize or if your supply is too scarce, you manufacture it manually until organic supply can take over. At PartnerSlate — a B2B food and beverage co-manufacturing marketplace I worked on — we couldn't wait for manufacturers to sign up organically, so we manually onboarded them. We called, we emailed, we offered free listing optimization, and we positioned the marketplace as a lead-gen channel they didn't have to pay for upfront. That hands-on work got us to critical mass, and once brands saw a fully stocked marketplace, organic manufacturer sign-ups accelerated.

The lesson: density-first supply isn't about waiting for both sides to show up; it's about engineering liquidity on one side so the other side has no choice but to stick around.

Match Rate Is Your North Star — Not Users, Not GMV

Here's the metric that actually matters: match rate — the percentage of demand-side requests you successfully fulfill.

Not total users. Not GMV. Not page views or app downloads. Match rate is the only number that tells you whether your marketplace is compounding or bleeding users.

At Shiftgig, we filtered every operational and growth decision through match rate. If a new city's match rate was below our internal threshold (we aimed for 90%+), we paused business-side acquisition and invested more in worker supply. If a vertical (say, hospitality vs. warehouse) had low match rate, we didn't scale it — we fixed the supply problem first. Match rate was the forcing function that kept us honest about liquidity.

Why? Because a marketplace with high match rate compounds. Both sides have a good experience, so they stay, they refer, and they transact again. A marketplace with low match rate — where demand-side users search and see nothing, or where supply-side users join and get no volume — bleeds on both sides. You can paper over the problem with aggressive acquisition spend, but you're just accelerating churn.

Most marketplaces obsess over GMV because it's the number investors want to see. But GMV without match rate is a vanity metric. Marketplaces typically retain 20-30% of GMV as revenue (the "take rate"), per Eric Andrews Startups, but if your match rate is low, that GMV is one-time transactions from users who will never return. You're renting transactions, not building equity.

Here's the diagnostic I run on every marketplace I touch:

Match rate by market. Pull it city by city or region by region. If one market is consistently below 80%, you have a supply problem there — either pause demand acquisition in that market or double down on supply-side growth until you hit the threshold.

Match rate by vertical or category. If you're a horizontal marketplace serving multiple verticals (like Shiftgig serving hospitality, retail, warehouse, etc.), match rate by vertical will tell you which categories are liquid and which are dead weight. We discovered early on that our hospitality vertical had great match rate but our warehouse vertical was under-supplied in certain markets. That insight let us deprioritize warehouse acquisition in those markets and focus budget where liquidity was proven.

Time-to-match. How long does it take to fulfill a request? Faster is better, but only if quality stays high. If time-to-match is increasing, you're either under-supplied or your marketplace is getting too broad (you're trying to serve too many verticals without the density to support them). At Shiftgig, we tracked time-to-fill for every shift request. If it started creeping up, that was an early warning sign that we needed more workers in that market or vertical.

If your match rate is below 80%, you have a liquidity problem, not a growth problem. Stop spending on acquisition and fix the supply side first.

Intelligent Demand Generation: Scale Only After Supply Is Primed

Once you have supply density and proven match rate in a market, demand generation becomes a math problem.

At Shiftgig, we grew the worker (B2C) side with the scalable channels — Facebook, SEO, and PPC — because the cost per acquisition was lower and the volume was higher. We could afford to test, iterate, and scale those channels quickly. The business (B2B) side was harder and more expensive, so we used targeted outbound, industry events, staffing-association partnerships, and a sales-enabled lead-gen engine.

Here's the key move: marketing enabled sales. We didn't let marketing and sales operate in silos. The lead-gen and SDR teams warmed leads to the point of closing so the sales team could focus only on closing, never on prospecting. That's how we cut B2B acquisition cost in half — we made every dollar work twice as hard by handing sales only qualified, warmed leads.

The lead-gen engine worked like this: we ran targeted LinkedIn and email campaigns to operations managers, HR directors, and staffing coordinators at mid-market and enterprise companies in verticals we knew we could serve (hospitality, retail, logistics). We used gated content (workforce trend reports, shift-scheduling best practices) to capture contact info, then passed those leads to an SDR team that qualified them with a short discovery call. Only after a lead confirmed budget, authority, need, and timeline did we hand them to the closing sales team. That multi-touch qualification process meant the sales team closed at a much higher rate because they weren't wasting time on unqualified leads.

But the bigger unlock was building owned channels so we weren't renting all our demand from paid platforms. At CodaPet — a national in-home pet end-of-life care marketplace where I serve as Head of Marketing — we built what I call the Local Presence Machine: over 200 fully optimized Google Business Profiles, one per market and practitioner, with automated location-specific content triggered by operational events.

Here's how it works in practice:

One Google Business Profile per practitioner per market. CodaPet operates across 170+ markets in 50+ states. Each market is served by local, licensed veterinarians who are independent contractors. We created a unique Google Business Profile for each vet in each metro they serve, so when someone in Denver searches 'mobile vet euthanasia near me', they see a local Denver vet with a local phone number, local reviews, and a Denver-specific description — not a generic national brand.

Automated, event-triggered content. Every time a vet completes an appointment, an automated system triggers a review request to the pet owner (with appropriate timing and sensitivity, given the category). Every time a vet adds a new service area, the system updates the relevant Google Business Profiles with fresh location-specific content. Every time we add a new vet to a market, the system spins up a new profile with pre-populated schema markup, hours, service descriptions, and photos. None of this is manual — it's all triggered by operational events in our CRM.

Reviews as a ranking and conversion engine. We track review velocity (reviews per profile per month) and average rating by market, and we treat reviews as a first-party growth lever, not a nice-to-have. Reviews drive local ranking, and they drive conversion — when a pet owner in crisis lands on a profile with 50+ five-star reviews and sees a local vet's face and credentials, the emotional barrier drops. That's why we built the review-request automation: it turns every completed appointment into a compounding ranking and trust asset.

The result: we grew the business roughly 225% from baseline in about a year, operate across 170+ metros, and now drive about 80% of demand through non-paid channels. Organic clicks grew roughly 5x and impressions jumped over 300% in about a year, while we cut blended paid CPA about 27% and sustained strong ROAS. We're running the function lean — one full-time lead plus a few contractors, using agencies and AI as leverage — because the systems are engineered to compound.

Paid acquisition is rent; owned channels are equity. At CodaPet, we restructured Google Ads into a performance-tier system and leaned into Performance Max where it dramatically out-converted Search. But the real work was building the local SEO infrastructure so the business isn't held hostage by any single platform's algorithm or CPM.

That's the demand-generation sequence: prove liquidity with supply density and match rate first, then scale demand through a mix of paid (for speed) and owned (for compounding). If you try to scale demand before supply is ready, you'll burn budget acquiring users who bounce on their first search.

The In-House Data Infrastructure You Actually Need

You can't optimize a marketplace through an agency black box.

When I consulted for The RealReal — a luxury consignment marketplace with over $100 million in revenue — they were spending heavily on paid acquisition but without the efficiency to match. The core problem wasn't the budget; it was that they were running acquisition through an external agency that didn't live in the granular, day-to-day nuance of their customer data.

Here's what the team did. We rebuilt the paid acquisition function and brought every capability in-house using a specific transition playbook:

Run agency and in-house in parallel for 30-60 days so nothing breaks. You can't flip a switch and shut off an agency without risking a revenue cliff. We ran both simultaneously — the agency continued managing the accounts while the new in-house team shadowed them, absorbed the account structure, learned the naming conventions, and documented every optimization rule and audience segment. This overlap period cost more in the short term but de-risked the transition entirely.

Hire a senior paid lead first — someone who can shadow the agency and migrate one channel at a time. We didn't hire a junior coordinator and expect them to reverse-engineer a multi-million-dollar paid program. We hired a senior performance marketer who had scaled paid social and paid search at another e-commerce company, gave them full access to the agency's dashboards and reporting, and had them migrate one channel at a time starting with the highest spend (Facebook). Only after Facebook was fully transitioned did we move to Google, then programmatic, then everything else.

Build a unified customer profile joining ad, web, email, and purchase data so you can see what actually drives LTV. The agency was optimizing for last-click conversions, which meant they were pouring budget into retargeting and bottom-of-funnel search because those channels got credit for the sale. But we had no visibility into which channels were actually creating new demand vs. which were just re-acquiring existing customers.

We built a unified data warehouse that joined ad platform data (impressions, clicks, conversions by campaign and creative), web analytics (sessions, page views, behavior flow), email engagement (opens, clicks, purchase attribution), and purchase data (first purchase, repeat purchase, LTV). That let us see the full customer journey — for example, that Instagram-acquired customers engaging with handbag content had 3x the average LTV compared to customers acquired through generic retargeting. That insight let us chase that audience aggressively and deprioritize the channels that were just harvesting existing demand.

Replace last-click attribution with a proper multi-touch model. Last-click lies — it hands all the credit to the final touch and none to the channels that created the demand. Once we could see the full journey, we reshuffled the channel mix entirely. We discovered that paid social was responsible for a huge share of first touches and consideration-stage engagement, but last-click attribution was giving all the credit to retargeting and branded search. The multi-touch model revealed the real incrementality of each channel, which let us reallocate budget toward the channels that were actually driving new customer acquisition.

The result: the team cut CAC by 40%, lifted LTV by 40%, achieved a 4x return on the paid program, and doubled the monthly revenue driven by paid social. We scaled paid social revenue from under $5 million to over $100 million.

But the real unlock wasn't the migration itself; it was owning the data and the infrastructure. We built real-time daily dashboards for CAC, LTV, and ROAS by channel, audience, and creative variant, so creative fatigue surfaced in 48 hours instead of a 30-day agency report. That speed let us kill losers fast and scale winners before the platform's algorithm caught up.

If you're running your marketplace through an agency and wondering why you can't hit your targets, the answer is simple: you can't optimize what you don't own. Bring it in-house, build the data infrastructure, and let the math tell you what to scale.

When Emotional Weight Drives the Category, Trust Is the Growth Strategy

Most marketplace advice assumes rational, transactional categories. But some marketplaces live in emotionally weighted, high-trust, high-consideration spaces where the purchase decision is driven by fear, loss, or life transition.

CodaPet is the first national at-home pet euthanasia marketplace. It's a brand-new category — people don't know to search for it — selling an emotionally devastating service where trust is everything. There's no existing demand to capture, so we had to create it.

Here's what worked: we treated category creation as demand creation, not intent capture. We built discovery-led acquisition (paid social, content, local SEO) for a service people don't know exists, and we made trust and community the core growth levers, not soft extras.

The Local Presence Machine I described earlier — 200+ Google Business Profiles, automated location-specific content, event-triggered review requests — does double duty: it ranks us locally for the few high-intent searches that exist ('mobile vet euthanasia near me'), but more importantly, it builds trust at scale. When a pet owner in crisis lands on a Google Business Profile with 50+ five-star reviews and sees a local vet's face and credentials, the emotional barrier drops. That's why reviews aren't a nice-to-have; they're a ranking and conversion engine.

We also rebuilt the funnel around time-to-conversion, not just conversion rate. In an emotionally raw category, every extra day in the purchase cycle is ad-spend bleed and a chance for the customer to ghost. Most pet owners who contact CodaPet are in acute crisis — their pet is suffering, and they need help within 24-48 hours. So we compressed the decision window with urgency (most appointments happen within 48 hours of first contact), used email automation to nurture the small percentage who need more time, and made the owned channels carry the trust-building work so paid acquisition could focus purely on discovery.

Here's the specific funnel architecture:

Paid social for discovery. We run Facebook and Instagram campaigns with empathetic, story-driven creative aimed at pet owners who are starting to consider end-of-life care. The creative doesn't sell the service directly; it normalizes the decision, addresses the guilt and fear, and positions at-home euthanasia as a dignified, compassionate option. The goal is to get people who didn't know this service existed into the funnel.

Local SEO for high-intent capture. The Google Business Profiles and local landing pages rank for the small set of high-intent searches ('mobile vet euthanasia', 'at-home pet euthanasia', 'in-home dog euthanasia near me'). When someone lands on one of these pages, they see a local vet, local reviews, and a simple booking flow. The emotional barrier is already lower because they're actively searching, so conversion rate is much higher than on paid social.

Email automation for the long tail. A small percentage of users aren't ready to book immediately — they're researching months in advance, or they're hoping their pet's condition improves. We built an email nurture sequence that delivers educational content (what to expect during at-home euthanasia, how to know when it's time, how to talk to kids about pet loss) over 2-4 weeks, with periodic check-ins and a soft CTA to book when they're ready. This sequence converts at a lower rate than the immediate-intent funnel, but it ensures we're not losing people who need more time to make the decision.

Review requests timed for empathy, not speed. We don't send the review request immediately after the appointment — that would be tone-deaf. We wait 7-10 days, send a condolence message with grief resources, and then gently ask if they'd be willing to share their experience to help other pet owners. The response rate is high because the request is framed as helping others, not as a transactional favor.

The result: we're operating a serious national marketplace with a lean team, growing at roughly 225% from baseline, and driving 80% of demand through channels we own. In emotionally weighted categories, trust and community are the growth strategy. If your marketplace lives in a space where the customer decision is driven by fear, loss, or a major life transition — luxury resale, end-of-life care, health and wellness — empathy-driven growth isn't a nice-to-have. It's the entire engine.

The Metrics That Actually Predict Compounding

If you're running a marketplace and optimizing for the wrong metrics, you will scale the wrong things.

Here's the short diagnostic I run on every marketplace I touch:

Match rate — the percentage of demand-side requests you successfully fulfill. If it's below 80%, you have a supply problem, not a growth problem. Stop spending on demand acquisition and fix liquidity first.

Time-to-match — how long it takes to fulfill a request. Faster is better, but only if quality stays high. If time-to-match is increasing, you're either under-supplied or your marketplace is getting too broad (you're trying to serve too many verticals without the density to support them).

Repeat transaction rate — the percentage of users who transact more than once. This tells you whether the first experience was good enough to earn a second one. If repeat rate is low, your match rate might look fine on paper, but the quality of matches is bad. At Shiftgig, we tracked repeat shift-fill rate for workers (did they come back and work another shift within 30 days?) and repeat business usage for employers (did they post another shift within 30 days?). Both numbers had to stay high or we knew we had a quality problem, not just a liquidity problem.

Cohort retention — are users from month 1 still transacting in month 6? If not, you're replacing churn with acquisition spend, which means you're renting growth, not compounding it. At CodaPet, we don't have traditional repeat transactions (most users only need the service once), so we track referral rate and review rate as proxies for satisfaction and compounding. If a user refers a friend or leaves a review, they had a good experience, and that compounds into future demand even if they don't transact again.

LTV-to-CAC ratio — the only number that tells you whether your unit economics are sustainable. If your LTV:CAC is below 3:1, you're not profitable at scale. If it's above 5:1, you're under-investing in acquisition. At The RealReal, the team rebuilt the customer data model to calculate true LTV (not just first-purchase value) and discovered that certain acquisition channels (Instagram, influencer partnerships) had dramatically higher LTV than others (generic retargeting, affiliate). That insight let them reallocate budget toward the channels that actually drove long-term value.

Most marketplaces I audit are tracking users, GMV, and conversion rate — all lagging indicators that tell you what happened, not why. The metrics above are leading indicators. They tell you whether the marketplace is compounding or bleeding out, and they tell you early enough to fix it.

Key Takeaways

Sequence matters more than speed. Build density-first supply in one market, prove match rate, then scale demand. A marketplace fully functional in 5 cities beats one half-broken in 50. At Shiftgig, we launched Chicago first, built worker supply to the point where we could guarantee 90%+ fill rate, then opened the business side. Only after match rate was proven did we move to the next city.

Match rate is your north star. Not users, not GMV. Match rate is the only metric that guarantees both sides stick around and compound. Pull it by market and by vertical. If it's below 80%, stop spending on demand acquisition and fix the supply problem first.

Subsidize the harder side honestly — and get specific. Identify which side is harder to acquire (usually supply), seed it with paid acquisition, referral incentives, and manual outreach, and geo-fence your launch so you never open demand until liquidity is proven. Uber paid drivers even when they had no riders. Grubhub manually gathered restaurant menus to build supply before opening demand. Do the unglamorous, non-scalable work to engineer liquidity.

Own your data and your channels. You can't optimize through an agency black box. Build first-party data infrastructure, bring acquisition in-house, and invest in owned channels so you're not renting all your growth from paid platforms. At The RealReal, the team brought paid acquisition in-house, built a unified customer profile, and replaced last-click attribution with multi-touch — the result was a 40% reduction in CAC and a 4x return.

In emotionally weighted categories, trust is the growth strategy. If your marketplace serves a high-consideration, trust-driven, life-transition category, empathy-driven growth isn't optional — it's the entire engine. At CodaPet, we built a Local Presence Machine (200+ Google Business Profiles, automated review requests, location-specific content) that serves as both a ranking engine and a trust-building mechanism. Reviews aren't vanity; they're the conversion lever.

Frequently Asked Questions

What is match rate and why does it matter more than GMV?

Match rate is the percentage of demand-side requests your marketplace successfully fulfills. It matters more than GMV because it's the only metric that tells you whether both sides are getting value — and whether they'll come back. You can have high GMV from one-time transactions, but if your match rate is low, you're bleeding users on both sides and papering over the problem with acquisition spend. High match rate means the marketplace compounds; low match rate means you're renting growth.

Should I launch my marketplace nationally or city-by-city?

City-by-city, every time. National launch sounds ambitious, but it kills liquidity. You end up with supply and demand spread so thin that neither side gets a good experience. Pick one market, get supply density right (enough suppliers that every demand-side search returns multiple high-quality options), prove match rate is consistently high, then open the next city. At Shiftgig, we launched Chicago first, proved 90%+ match rate, then moved to the next city. A marketplace fully functional in 5 cities beats one half-broken in 50.

How do I know which side to prioritize first — supply or demand?

Prioritize whichever side is harder to acquire, which is usually supply. The harder side is the one that requires more trust, more onboarding friction, or more upfront commitment. In most cases, it's easier to get consumers to browse than it is to get businesses or service providers to commit inventory, time, or labor. Build that side first, seed it if you have to, and only open demand once you have enough supply that match rate will be high from day one.

What's the fastest way to diagnose if my marketplace has a liquidity problem?

Pull your match rate by cohort, by market, and by vertical. If it's below 80%, you have a supply problem, not a growth problem — stop spending on demand acquisition and fix the supply side first. Also check time-to-match (how long it takes to fulfill a request) and repeat transaction rate (the percentage of users who transact more than once). If time-to-match is increasing or repeat rate is low, your marketplace looks liquid on paper but the quality of matches is bad.

How do I build owned channels instead of relying on paid acquisition?

Start with local SEO if you're a geo-based marketplace. At CodaPet, we built 200+ Google Business Profiles (one per vet per market), automated location-specific content triggered by operational events, and implemented event-triggered review requests. The result: organic clicks grew roughly 5x and impressions jumped over 300% in about a year, and we now drive 80% of demand through non-paid channels. Owned channels compound — every review, every piece of content, every local ranking improvement is an asset that keeps working without ongoing spend.