US K-beauty growth experiments: a first-90-days test plan

An existing customer base in Korea is useful evidence about the product, but it does not establish which US customer will buy it, at what price or through which channel. A brand already selling in America has a different starting point: recent orders, returns, acquisition sources and customer questions can show where growth is constrained.

Choose a hero product and a specific customer problem for the first plan. Record the landed product cost, available inventory, margin assumptions, approved claims, fulfillment ownership and purchase destination. If those inputs are unresolved, describe the work as launch preparation rather than pretending that a larger marketing budget will fix them.

Translate the buying decision, not just the packaging language. Explain the product's role in a familiar routine, the amount to use and the evidence behind the benefit. A product celebrated in another market may need a different introduction in the US. Keep the product truth consistent while changing the explanation.

Use the first 90 days to answer a limited set of commercial questions. The worksheet below organizes experiments by dependency rather than prescribing an agency onboarding schedule. Finish readiness work before running a test that depends on it, and allow each measurement window to mature before making a comparison.

Name one decision owner on the brand side. Keep an approval record for the product story, creator brief, budget and destination. Parallel preparation can save time, but work that depends on an unresolved claim or unavailable product should not be rushed into publication.

Use one priority channel for the initial experiment when that gives the team a clear customer, destination and success definition. A direct website, marketplace and TikTok Shop have different buying and operating requirements. Choose the channel your customer can use and your team can support, rather than opening several independent programs at once.

One priority channel does not require silence elsewhere. It means the test has one primary destination and success definition. Announcements and creator content can support that destination without launching several independent programs at once. If the brand already has reliable demand on one channel, investigate the existing bottleneck before opening another.

Write down the uncertainty the content should resolve. For a cleanser it might be how the texture fits an evening routine. For makeup it might be how a shade looks under consistent lighting. Choose creators who can demonstrate that point credibly, then compare clearly recorded approaches rather than asking everyone to repeat a generic launch script.

Separate asset production from audience distribution and paid usage. Contract the work and permissions needed for the test, preserve room for honest opinions and make disclosures clear. Decide how many usable outputs the team can review and deploy before deciding how many people to recruit.

Track the connection between the content and the destination. A video explaining one product concern should lead to a page that answers it. If comments repeatedly ask the same question, update the product explanation and the next brief together. The faster learning cycle comes from closing that loop.

Use an explicit test budget and separate production, creators, samples, logistics, media and management. Keep a reserve for correcting the purchase experience or producing a follow-up explanation. Spending the entire budget on the first wave leaves little room to act on what it teaches you.

An illustrative order with $50 in net revenue and $25 in product, fulfillment, transaction and expected return costs leaves $25 before marketing and overhead. That $25 is not a recommended acquisition cost: spending all of it on acquisition leaves nothing for overhead or profit. Replace every input with the brand's own costs and decide which additional margin must remain.

Do not justify unprofitable first orders with assumed lifetime value. If repeat purchase is part of the plan, use observed cohorts and a realistic time horizon. A replenishable product still needs evidence that customers buy it again at a cost the business can sustain.

At review, write three separate conclusions: what the team delivered, what customers did and what remains uncertain. If conversion improves after a page change and a promotion, acknowledge that both changed. Attributed sales alone cannot establish which action caused the improvement.

Increase a supported activity in a bounded next cycle, with enough stock and people to handle it. If the evidence is weak, simplify the test or change the hypothesis. If demand is real but fulfillment is unreliable, the next investment may belong in operations. Scaling fast means removing the actual delay in the business, not increasing every activity at once.