Marketplaces
Apex connects sellers, buyers, listings, searches, transactions, messages, experiments, partners, and fees so you can find where the market is thin, fix it, and prove what made it healthier.
No credit card to start.
Find the constraint
Marketplace-wide averages can look healthy while one category has too many sellers and another has buyers leaving empty-handed. Apex measures liquidity, match rate, time to sale, trust, and concentration by the slices you can actually change.
Move both sides
Score new sellers by the gaps they can fill. Help the right ones reach a first listing. Bring quiet supply back before it disappears. Give buyers relevant inventory and stop messaging the moment they act.
The milestone, as a sentence
apex / engagement / adoptionWhen a seller hasn'twithinof, send them.
4,126
Sellers tracked
1,608
Nudged
2,884
Listed
+18.6%
Lift over sellers left alone
Pick the event. Green means Apex has seen it fire.
Protect the economics
GMV isn't revenue. Apex keeps the order, your fee, variable cost, seller payout, and acquisition source connected. You'll see contribution margin, lifetime value, CAC, and payback separately for buyers and sellers, then send the right value back to every channel.
One seller and one buyer, end to end
Marcus sells a camera on Morrow. Nina is looking for one. Morrow is fictional, but every screen below is Apex. In sixteen steps, you'll see how Apex connects both sides from the first ad to the marketplace fee.
Marcus Reyes
Seller
Nina Brandt
Buyer
01 · Before anything happens
A marketplace can easily end up with one definition of a seller in operations, another in marketing, and a third in finance. Then each team brings a different order count to the same meeting.
Apex starts with one list of moments on both sides: a seller signs up, a listing goes live, a buyer searches, an offer is accepted, and an order is paid. The paid-order event names the buyer, seller, order value, fee, category, and city together.
Green means Apex has seen the moment in live traffic. Grey means the name exists but nothing has sent it yet, so you can fix the missing signal before anyone plans against it.
Your schema
mapped to Apex Spec v1.4.0
you Set up Apex in this repo.
apex Scanning 2,418 files for analytics calls and identify traits.
Found 17 events, 22 attributes. 6 names differ from the spec.
cart.addItem() → add_to_cart
firstName → first_name
apex Opened a pull request with the plan. Nothing ships until you merge it.
Point Apex at your code once. Sellers, buyers, and paid orders use the same names everywhere else.
One integration02 · Monday, 7:40am
Marcus has repaired cameras in Portland for nine years. A Morrow ad asks him to open a shop, and he taps it over breakfast.
Apex attaches the campaign before anyone clicks. Every report already knows this was money spent finding a seller, so it never gets mixed with the budget used to find buyers.
That separation matters because the payback is different. A seller can create fees for years. A buyer may make one purchase and leave. Blend them and the cheaper click can win even when it weakens the market.
Morrow
Sponsored
Somebody is searching for the gear sitting on your shelf right now. Listing it takes a few minutes, you set the price and the shipping, and you don't pay a thing until it sells.
morrow.market
Turn the shelf into a shop
Free to list. You keep control of the price.
The link in the ad
apex / linksWhere it sends them
Settled now, not guessed weeks later. Every click here counts as money spent finding sellers, so it never blends into what you spend finding buyers, and the two budgets stay comparable.
The channel is chosen when the link is made, not guessed after the click.
Seller spend stays separate03 · 7:52am
Helping a new seller takes time. Someone may need to verify the shop, answer questions, or get on a call. Doing that for everyone spends the same hour on people who may never list a thing.
Apex scores Marcus from signals your team can inspect: what he sells, how often he returned, how much stock he has, and the campaign that brought him in. He scores 82 because cameras are short on supply while buyers are actively searching.
Your team starts with the sellers most likely to improve the market, not the newest form at the top of an inbox.
What it made of Marcus
scored in 38msEvery other new seller, sorted the same way
2,350 seller sign-upsHe signed up at 7:52 and was scored at 7:52. Nobody waits for an overnight sync.
Scored immediately04 · Day 0 through day 4
The journey starts when a seller registers and watches for the moment buyers can actually find something to buy: the first listing.
Sellers in categories short on stock reach a person. Everyone else gets a measured sequence of waits and messages. An experiment inside the journey compares two reasons to list, then keeps the better one.
The whole plan is visible on one screen, including when Apex stops. Sellers who list leave the journey instead of receiving a reminder that no longer fits.
Every path stops when the first listing goes live. The journey is measured against the same event throughout.
One conversion event05 · Day 2
A registered seller with nothing listed doesn't improve liquidity. Until listing_created fires, searches, selection, and paid orders can't move.
Marcus posts his first camera on day two. The milestone tracks how many sellers did the same, how many needed a nudge, and how many would have listed without one.
The lift is measured against sellers deliberately left alone. That's how you know the journey worked instead of taking credit for behavior that would have happened anyway.
The milestone, as a sentence
apex / engagement / adoptionWhen a seller hasn'twithinof, send them.
4,126
Sellers tracked
1,608
Nudged
2,884
Listed
+18.6%
Lift over sellers left alone
Pick the event. Green means Apex has seen it fire.
You choose what activation means. Apex doesn't invent it for your marketplace.
Your definition06 · Thursday, 9:05pm
Nina searches for a 35mm film camera. Apex keeps the search and how many results came back. An empty search is a buyer naming the stock your market is missing.
A listing view is mild interest. An offer is stronger. A paid order is the Conversion. Apex keeps those moments separate so browsing never gets mistaken for demand.
She joins the Segment of buyers who searched but didn't buy, then leaves it as soon as she pays. The group stays true without a team rebuilding a list each week.
Buyers who searched but didn't buy
Buyer side4,180 peopleSearched in the last 7 days and hasn't bid or made an offer since
Buyers who came back
Buyer side1,764 peopleHas paid for two or more orders and left feedback on the last one
The same Segment can feed a message, an experiment, an Adaptive Journey, or an ad-platform audience.
One living definition07 · Thursday, 9:11pm
Nina offers $475 for Marcus's camera. He accepts in 41 minutes. One paid order now connects the search, listing, offer, buyer, seller, and fee.
13.6% of searchers paid a seller within a week. 47.8% of bids and offers became paid orders. Those rates tell you whether buyers are finding supply and whether sellers are responding.
Trust stays beside growth. Cancellations, refunds, disputes, ratings, and review coverage show whether a higher match rate created a healthier market or simply moved more problems downstream.
Every rate above divides a seller-side event by a buyer-side event. A one-sided tool can't do this math.
Both sides are required08 · Find the constraint
Across the whole marketplace, the ratio looks healthy. Inside it, sneakers has too many sellers and cameras has too few. Those problems require opposite budgets.
Recruiting another sneaker seller would spread the same buyers more thinly. Recruiting a camera seller shortens search, improves choice, and gives existing demand somewhere to go.
Apex shows only the slice-level numbers the data can support. If a field isn't available by category, the metric stays blank instead of becoming a confident-looking guess.
Category, metro, and price band are filters on the same honest read, not separate spreadsheets.
Slice without rebuilding09 · Follow the money
Buyers paid $4,182,940 in thirty days. That's GMV, the value moving through the market. It isn't what Morrow earned.
The marketplace kept $501,953 in fees. After variable costs, contribution margin was $360,733. Sellers received $3,680,987.
Apex keeps those numbers separate on every screen. If the cost field is missing, margin stays blank. A missing number is safer than a guessed profit figure with a budget behind it.
GMV, your fee, variable cost, and seller payout all come from the same order.
One paid-order event10 · Improve the match
Search ranking decides which sellers buyers see. This experiment compares the best seller rating with the fastest ship time.
Ship time wins: 52% of those buyers paid, compared with 43.6% under the old order.
Apex counts paid orders, then values the lift with the fee the marketplace kept. Clicks and offers can explain the path, but they don't get mistaken for revenue.
+$131,706 a quarter
What it is worth
a quarter
Ship time first beat Control · seller rating first. 542 extra orders paid, $81 each, 3 times a quarter.
The value is Morrow's fee, not GMV. Apex checks the result against the control, the previous baseline, a global holdout, and an opt-in industry benchmark.
6,450 buyers got the winning version. That's 8.4 more orders paid per 100 than Control · seller rating first, so 542 more in total. Each one is worth $81, Morrow's fee on an average order. The same 30-day run happens 3 times a quarter.
The order is chosen before the page is sent. Buyers never watch the results jump around.
The winner can become code11 · Set up once
Add three DNS records once and every email carries your name, colors, and sending reputation. Apex never appears in the from line.
A marketplace sends receipts, shipping updates, payout notices, and offers alongside marketing. Transactional messages still arrive when the customer has opted out of marketing because the order depends on them.
Both kinds use the same composer, so the relationship feels like one company instead of a polished campaign beside a disconnected system email.
apex1._domainkey.morrow.marketapex1.dkim.apex.incmorrow.marketv=spf1 include:amazonses.com ~all_dmarc.morrow.marketv=DMARC1; p=quarantine; rua=mailto:dmarc@morrow.marketA full email in their inbox. Authored once, with Nina's attributes resolved from your schema.
Marketing and must-send messages share the same brand. Eligibility rules still match the kind of message.
One editor, two pipelines12 · Write it once
You write the message once. Apex can deliver it as email, mobile push, browser push, or an inbox message inside your product.
Values from the seller record fill in at send time. The store, category, and city stay current, so a seller never receives a message built from last month's export.
The experiment compares two clear ideas: make listing feel easy, or show that buyers are already searching. The second message produces more first listings and becomes the new control.
A full email in their inbox. Authored once, with Marcus's attributes resolved from your schema.
{{first_name}}Marcus{{store_name}}Reyes Vintage Optics{{category}}Cameras & Photo{{geo}}Portland+$161,880 a year
What it is worth
a year
Name the demand beat It only takes three minutes. 71 extra first listings, $190 each, 12 times a year.
The result is checked against the control, previous baseline, global holdout, and opt-in industry benchmark.
804 new sellers got the winning version. That's 8.8 more first listings per 100 than It only takes three minutes, so 71 more in total. Each one is worth $190, Morrow's fee on the orders that listing produces. The same 30-day run happens 12 times a year.
The tabs show the actual shape of every channel. The approved content stays together.
Four channels, one message13 · Six weeks later
Sellers with live listings and no sale in thirty days haven't quit yet. That's the window where the marketplace can still help.
The Segment says who. The Adaptive Journey says when. The experiment compares a general reminder with one real search that found nothing in the seller's category.
The experiment is still running, so Apex doesn't call a winner. A strong week can't silently replace the message everyone receives.
Sellers with nothing listed
Seller side612 peopleRegistered in the last 14 days and hasn't posted a listing
Sellers who've gone quiet
Seller side884 peopleHas at least one live listing and hasn't sold anything in 30 days
The Segment, journey, and experiment refer to the same sellers.
One shared definition14 · Zoom out
Sellers and buyers use a marketplace differently. Sellers may return every day to manage listings. A camera buyer may make one purchase and disappear for months.
Average them together and the result describes nobody. Split them and the charts become useful: seller retention shows whether supply is staying healthy, while buyer cohorts show whether demand returns when it should.
The came-back line matters too. A returning seller already has stock and costs far less than recruiting a replacement.
Switch between sellers and buyers. The same events answer two different retention questions.
Two populations15 · The full record
Marcus has taken in $48,720. Morrow has earned $5,846 in fees from his orders. The record keeps those numbers apart.
Underneath is the whole sequence: the ad click, score, sign-up, first listing, Nina's offer, the paid order, and the seller payout.
Every experiment exposure and message stays attached, including the times Apex deliberately held him out. Someone can understand the relationship without reconstructing it from several tools.
Marcus Reyes
marcus@reyesoptics.com
$5,846
fees Morrow has earned from him
How he got here
every event, in orderArrived from a Meta ad
go.morrow.market/camera-sellers-q3 → /sell/start
page_view
Finished the seller sign-up form
Scored 82, the top bucket, before anyone read it
form_submit
Registered as a seller
Reyes Vintage Optics, Portland, nine years in cameras
seller_signup
Opened the first-listing nudge
Arm: Name the demand
apex_email_opened
Posted his first listing
Serviced 1970s 35mm SLR · $520 Buy It Now · offers on
listing_created
Nina offered $475
Search: 35mm film camera · 3rd result
offer_submitted
Accepted the offer
Answered in 41 minutes
match_created
Order paid
$475 order · $57 fee to Morrow
transaction_completed
Got paid
$418 transferred, two days later
payout_sent
Every experiment he was in
3 experimentsEntered 2mo ago
Marcus, 8,400 searches and 19 sellersv2email · Opened · 2mo ago
Entered 1mo ago
Nina offered $475v1mobile_push · Delivered · 11d ago
The ad, score, messages, experiments, orders, seller payout, and marketplace fee were never separate records.
One continuous story16 · Multiply by the whole market
One seller and one buyer become a market-wide decision. Apex compares what it costs to find each side with the fees each side creates over time.
Seller acquisition pays back on every order that seller ships. Buyer acquisition pays back on the orders that buyer places. Separate tables keep the two clocks honest.
Paid orders can flow back to ad networks at the fee value you kept, so bidding learns from marketplace revenue instead of raw GMV. The Partner Network gives you another way to recruit scarce sellers through the communities they already trust.
Find more camera sellers on Meta
Meta is the only channel bringing in sellers for the categories buyers can't fill. A camera seller found here returns $4.30 in fees for every $1 you spend to get them. A sneaker seller returns $1.60.
Trim Google Display on the buyer side
Display brings buyers who search once and never buy. Its cost per paid order is four times what search costs, and it's feeding the unfulfilled-demand number.
Reconnect the Meta conversions feed
Meta stopped accepting paid orders nine days ago. Its bidding is now chasing offers that may never turn into sales.
Morrow Seller Partner Program
Morrow
Camera shops and collector groups already know every serious seller in their niche. They earn on the fee from every order their sellers complete, for as long as those sellers keep selling.
Experiments, ads, channel economics, and partner payouts all read the same Conversion.
One paid-order eventKeep what worked
The winning seller nudge, search order, channel decision, and supply strategy shouldn't disappear when the quarter ends. Apex keeps each result with the evidence attached, so the next decision starts from what your marketplace already learned.
Ranking listings by how fast the seller ships gets more paid orders than ranking by seller rating.
last moved by Search ranking test
A seller who lists within three days of registering is far likelier to still be selling at 90 days.
last moved by First-listing nudge
Adding sneaker sellers doesn't add paid orders, because that category already has more sellers than buyers.
last moved by Seller-recruiting budget shift
Naming one search that found nothing brings a quiet seller back better than a general reminder.
last moved by Quiet-seller wake-up
Your move
Find the thin side. Help sellers and buyers move. Protect the fee you keep. Prove which change improved the market.