A marketplace with no supply and no demand can’t spend its way out of the gap. 2026 CAC benchmarks put supply-side acquisition around $150 against roughly $30 for demand, a five-to-one spread. Pouring budget into the cheap side first just fills the funnel with people who find nothing to transact on. Sequencing supply and demand in the right order, one narrow slice at a time, closes the gap instead.
What Actually Breaks Growth Spend in a New Marketplace
Most early marketplace budgets go toward demand, because demand is the cheaper, faster-reporting side to acquire almost everywhere. That’s exactly the problem. When supply is thin, every dollar spent pulling in buyers produces a buyer who opens the app, finds an empty category, and leaves, an acquisition cost the dashboard still logs as a signup.
Marketplace CAC data compiled in 2026 by Lech Kaniuk, author of The Two Numbers, puts blended marketplace acquisition between $30 and $150, with demand-side acquisition running near $30 and supply-side acquisition closer to $150 . Supply costs roughly five times more to acquire than demand in a typical marketplace. Spending the marketing budget on the cheap side first produces a reported signup number and no real progress toward liquidity.
Separate 2026 research on B2B acquisition costs found CAC up 40 to 60% since 2023 (Data-Mania, July 2026). Misdirected spend costs more to recover from now than it did two years ago, which raises the price of guessing wrong about which side to seed first. The same sequencing mistake shows up across nearly every early-stage channel decision, marketplaces included, and Voxturr’s guide to digital marketing for startups walks through the general version of this discipline: prove one channel before adding the next.
The SLICE Framework
SLICE is the five-step sequence we run at Voxturr to get two-sided marketplaces liquid: size the smallest real slice of the market, land real supply into it directly, broker the first trades by hand, confirm they repeat without a manual push, then replicate into the next slice. Demand and supply get read as one system throughout, not two separate reports that never talk to each other.
This sequence sits inside Voxturr’s step-by-step pre-launch marketing guide and applies specifically to platforms still proving their first transactions, not brands already selling on an established marketplace like Amazon or Flipkart.
| Step | Goal | Primary Tactic | Signal It’s Working |
| Size | Define the smallest real transaction unit | Localized keyword and search-intent data | Real, specific search volume inside the slice |
| Land | Recruit supply directly, inside the slice | Direct outreach where sellers already are (Instagram, local directories) | A confirmed count of active sellers in the slice |
| Instigate | Broker the first trades manually | Concierge matching over DM or WhatsApp | First completed transaction |
| Confirm | Verify real liquidity | Track transactions with no manual push | Two or more repeat transactions, unprompted |
| Expand | Replicate into the next slice | Repeat Size through Instigate, one slice at a time | The next slice reaches Confirm faster than the first did |
Size the Slice
A slice is the smallest unit where supply and demand could plausibly meet: one neighborhood, one sub-category, one price band, not a city and not a category nationwide. Aggregate demand numbers hide the real gap, because a country-level search volume can look healthy while the specific radius a buyer expects supply from is empty.
Run the narrow, localized version of the query through a keyword or search-intent tool before assuming demand exists; KWfinder is a common one for this. Not “furniture rental” but “furniture rental Bandra.” Not “ceramics subscription” but “handmade ceramics subscription Austin.” A narrow query with real, if modest, volume is a usable signal. A narrow query with almost nothing behind it means the slice picked is too small, or the category isn’t there yet in that specific pocket, and that’s cheaper to learn from a search tool than from three months of paid acquisition into it.
Search volume is intent, not commitment. Pair it with a small manual test, the same pre-order and beta-signup validation Voxturr’s lean eCommerce startup guide walks through, before treating a slice as validated.
Land Real Supply
Recruit sellers directly, inside the same slice sized in the previous step, instead of publishing a sign-up form and waiting for supply to arrive on its own.
For physical-goods and local-service marketplaces, Instagram already functions as an informal supply directory. Independent sellers, makers, and small vendors post inventory, prices, and sometimes take orders through DMs before any marketplace exists to organize them. Search the same narrow terms used to size demand as hashtags and location tags, then message the sellers already active in that footprint directly. The demand number from the previous step becomes the pitch: a specific count of people searching for a category in a specific neighborhood recruits a seller faster than a generic invitation to join a platform.
Recruit inside the slice only. A seller two neighborhoods over, or one category adjacent, dilutes the density the entire sequence depends on, even when adding them feels like forward motion.
Instigate the Match
Broker the first transactions personally. Match a specific buyer to a specific seller over DM or WhatsApp, and process the transaction by hand if needed, before automating any of it.
This stage produces proof: a stranger paying a stranger through the platform, more than once, inside this exact slice. Whatever gets built to automate matching later should replace a manual process that already works, rather than guess at one that hasn’t been tested.
Confirm the Loop
A slice is liquid when a transaction completes without a manual push from the team, and it happens more than once, involving more than one buyer-seller pair. Signups, downloads, and total registered users don’t count toward this.
Teams that stall usually keep adding demand-side spend to a slice that looks flat but not obviously broken on an aggregate dashboard. A slice that still needs a personal message every time a listing goes live isn’t liquid, whatever the total user count reports.
Expand the Slice
Once a slice confirms the loop, repeat Size through Instigate in an adjacent slice: the next neighborhood, the next price band, the next sub-category. Opening every slice at once repeats the exact mistake this sequence exists to avoid.
Launching every slice simultaneously multiplies that $150 supply-side acquisition cost across all of them before any single one produces the transaction proof that makes the next slice cheaper to open. A slice that’s already liquid generates reviews, real transaction data, and word of mouth in that specific pocket. Let one slice prove itself before opening the next, and that proof does part of the recruiting work for you.
What Changed Since Airbnb and OpenTable Ran This Playbook by Hand
Airbnb targeted a single city and a single event to get its first bookings. OpenTable sold restaurants’ standalone reservation software years before the diner side of its marketplace existed. Both companies ran a version of this sequence manually, without the tools available today. Two things changed since then: paid acquisition costs more, and organic search sends fewer people straight to a website, which makes skipping the manual diagnostic more expensive today than it was for either of them.
Search data from 2026 shows AI Overviews now answer roughly 26% of desktop searches directly, cutting organic clicks by as much as 58%, while the share of zero-click searches has risen from 56% to 69% ; sites are losing an average of 24% of their organic traffic as a result. A marketplace can’t SEO its way to liquidity if an answer engine resolves the query before a visitor reaches the site. The content that still earns clicks and citations is real, structured transaction data: reviews, completed listings, specific outcomes. That’s exactly what Confirm the Loop produces, and an empty marketplace can’t fake it.
| 2008–2015 marketplace launches | 2026 marketplace launches | |
| Demand sizing | Founder intuition, informal surveys | Localized keyword and intent data shows real search volume at the slice level |
| Supply sourcing | Cold calls, in-person recruiting | Public social inventories (Instagram and similar) make sellers visible before outreach starts |
| Paid acquisition cost | Lower, more forgiving of a wrong guess | B2B acquisition costs up 40–60% since 2023, raising the cost of guessing wrong |
| Organic discovery | Search results sent visitors to the site | AI Overviews resolve roughly a quarter of desktop searches before a click happens |
Founders building marketplaces in 2026 also have tools Airbnb and OpenTable’s teams didn’t. Localized keyword and intent data makes Size the Slice a days-long exercise instead of a guess. Public social platforms already contain visible seller inventories, which turns Land Real Supply into a sourcing problem instead of a cold-outreach problem. Skipping the diagnostic in 2026 costs more time than running it, because the tools that make Size and Land fast now exist and mostly go unused.
Frequently Asked Questions
Which side should I seed first, supply or demand?
Usually supply. It’s the side buyers expect to already exist, and 2026 benchmarks put it at roughly five times the acquisition cost of demand, which is reason enough to prioritize the budget there. Confirm with Size the Slice rather than assuming; a new category can flip this.
How narrow should the first slice actually be?
Narrow enough that a handful of manually matched transactions could plausibly saturate it: a single neighborhood, a single price band, or a single sub-category. If the slice doesn’t feel embarrassingly small, it probably isn’t narrow enough yet.
What if the marketplace is fully digital and there’s no Instagram supply to find?
The sourcing channel changes, the sequence doesn’t. LinkedIn, niche Slack or Discord communities, and existing directories play the role Instagram plays for physical-goods sellers: somewhere the supply side is already visible, just not yet organized into a marketplace.
How long should the team stay in manual concierge mode?
Until Confirm the Loop passes, not on a fixed calendar. Automating before a slice proves repeat, unprompted transactions moves the chicken-and-egg problem into the product instead of solving it.
Closing
Founders who stay stuck in the loop longest usually don’t have a product problem. They spend the first two quarters treating a sequencing problem like a marketing problem, running ad tests and rewriting onboarding copy against a gap that neither one touches. Voxturr’s pre-launch marketing team runs the Size and Land diagnostic before any spend goes out, so the first budget decision is which slice to seed, not which channel to try next.
