Building Commerce Platforms for Creators

What infrastructure do platforms need to support creators as genuine commerce partners rather than ad placements?

The fastest way to tell whether a company actually thinks of creators as commerce partners is to ask how a creator gets paid. Not the headline rate. The mechanics. When does the commission post? What happens on a return? Can the creator see, before publishing anything, which SKUs are in stock in the markets where their audience actually lives?

Most platforms cannot answer those questions cleanly, because their creator program was built on top of an ads system. Ads systems are designed around impressions and clicks, which settle instantly and never get returned. Commerce settles weeks later, gets partially refunded, and depends on inventory that moves while the content is still being edited. Bolting one onto the other produces a program that looks functional in a deck and frustrates everyone using it.

I ran creator commerce at AliExpress US, onboarding around 400 creators a month and growing the active base past 2,000. That program contributed roughly 40% GMV growth, and almost none of that came from the recruiting motion people assume. It came from fixing infrastructure that made creators willing to post a second time.

Retention Is the Only Creator Metric That Compounds

Every creator program I have seen reports on signups. It is the easiest number to move and the least predictive of anything. You can buy 5,000 signups with a good landing page and a referral bonus. Six weeks later, 400 of them have posted once and 60 are still active.

The metric that actually drives GMV is second-post rate: of the creators who published once, how many published again within 30 days. It is brutal, honest, and almost entirely a function of what happened after the first post. Did the tracking fire? Did the commission appear where they expected it? Did the product page look like the product they filmed?

When we started tracking second-post rate as the primary health metric at AliExpress, the roadmap rewrote itself. Recruiting features got deprioritized. Attribution reliability, payout visibility, and catalog quality moved to the top. The signup number barely changed that quarter. GMV moved a lot.

The Four Systems That Actually Matter

Strip away the community features, the leaderboards, the branded merch, and creator commerce infrastructure comes down to four systems. Get these right and creators will tolerate a mediocre interface. Get them wrong and no amount of community management saves you.

Notice that none of these are growth features. They are all trust infrastructure. Creator commerce has a uniquely high trust requirement because the creator is spending their own audience equity on your catalog. If the product disappoints, the audience punishes the creator, not the platform. That asymmetry is the central design constraint and most programs ignore it.

Stop Treating Creators as a Single Segment

The affiliate playbook I built at AliExpress covered 10,000-plus partners across 40-plus countries, and the thing that made it work at that scale was refusing to treat partners as interchangeable. Tiering by audience size is the obvious cut and it is mostly wrong. Audience size predicts reach. It does not predict commercial fit.

The cuts that actually predicted GMV contribution were intent depth and category coherence. A creator with 40,000 followers who reviews one product category in detail converts at a multiple of a creator with 400,000 followers posting general lifestyle content, because their audience arrives already in a purchase consideration mindset. On a marketplace with 50 million-plus monthly users, we could see this clearly in the data: the recommendation system surfaced different products to traffic from these two creator types, and the conversion gap was not subtle.

This matters for platform design because it changes what you build. If your model is reach, you build tools for scale: bulk link generation, campaign blasts, mass onboarding. If your model is intent, you build tools for depth: category-level performance data, product sampling workflows, early access to launches, direct lines to brand teams. Those are different products. Most platforms build the first and wonder why their top creators leave for the second.

The Commission Rate Is Not the Lever You Think

Every brand I have worked with, including the consumer electronics brands at Next2Market, eventually asks the same question: should we raise the commission rate to attract better creators? Usually the answer is no, and it is worth understanding why.

Raising a rate from 8% to 12% is a 50% increase in your cost per sale and roughly a rounding error in a creator's decision to work with you, because their earnings are dominated by conversion rate and order value, not commission percentage. A creator earning 8% on a product that converts at 4% with a $120 AOV makes far more than one earning 15% on a product converting at 0.8% with a $40 AOV. Rate is the number brands can change in an afternoon, so it is the number they reach for.

Creators do not optimize for commission rate. They optimize for earnings per thousand views, and that number is mostly determined by things the platform controls and the creator cannot see.

The implication is uncomfortable for platform teams: the highest-leverage work in creator commerce is conversion rate optimization on product pages, checkout, and recommendation quality. That work lives in product and growth teams, not partnerships. But it shows up in creator earnings, which shows up in creator retention, which shows up in GMV. At Next2Market, improving conversion rate by 18% did more for partner economics than any rate negotiation we ran that year.

What I Would Build First

If I were standing up a creator commerce program today with limited engineering capacity, I would sequence it deliberately and resist the pressure to launch broad.

First, attribution and payout, built properly, before recruiting a single creator. These are the systems you cannot retrofit without breaking trust with the people already using them. Second, a performance dashboard that shows per-link data in near real time, even if it is ugly. Third, catalog truth: an API and a UI that never shows a creator a product they cannot actually sell. Only then, recruiting.

Start with 50 creators in one category and one market. Run them for a quarter. Measure second-post rate and earnings per thousand views. If those two numbers are healthy, the program will scale almost on its own, because creators talk to each other and a program that pays reliably is the rarest thing in this market. If those numbers are weak, scaling to 5,000 creators just distributes the disappointment more widely.

The creator economy discourse spends enormous energy on discovery, aesthetics, and community. Having built these systems at scale, my honest view is that creators are a distribution channel with unusually high trust requirements and unusually good economics if you meet them. The platforms that win are the ones that treat this as an infrastructure problem and staff it accordingly.