Commission bands, and the price sweet spot

Start with the going rates, because they tell you where the margin lives. Across the marketplace, the typical bands look like this: electronics and gadgets, 5–10%; fashion, home goods, and accessories, 10–20%; beauty, skincare, supplements, and fragrance, 15–30%. The reported platform average sits around 13%. Launches often run temporarily boosted “open collaboration” rates to seed creator content — a 25% rate on a new beauty product is a marketing budget, not an accident.

Rate alone doesn't pay you. What pays you is rate × price × sales that stick, so run the three-way comparison every time. A $120 gadget at 8% pays $9.60 per sale. A $28 skincare product at 20% pays $5.60 per sale. A $12 trinket at 30% pays $3.60. Now ask the only question that matters: which one will a scrolling stranger actually buy from a 30-second video, and keep? Impulse physics favor the middle — most converting Shop products sit between $15 and $60, cheap enough to buy on a demo, expensive enough that the commission is real money.

Worked example, head to head. Product A: water flosser, $45, 15% — $6.75 per sale, demonstrates beautifully (visible result in seconds), 40,000 units sold on the listing. Product B: posture corrector, $19, 30% — $5.70 per sale, demonstrates awkwardly, and its category is famous for returns because fit and comfort disappoint. B waves the bigger rate. A is the better business: higher pay per sale, proof built into the demo, and a sales history that says buyers keep it. The 30% on B exists precisely because sellers must bribe creators to touch it.

Build your shelf by category logic, not by scrolling mood. Pick one home category where you'll be a repeat face (your audience and the algorithm both learn it), then test adjacent products inside it. A creator known for honest kitchen testing can sell the knife, the pan organizer, and the flosser's cousin the water scaler — the trust transfers. Random-category posting resets that trust with every video.

Reading a product page like a buyer of traffic

Your video is paid media the seller isn't paying for upfront — so read each listing the way a media buyer reads a landing page, because your effort lands on it. Four signals, in order. First, units sold: a listing with thousands sold has proven it converts strangers; a listing with 14 sold is asking you to be its market research. Beginners should ride proven listings and let other people's videos validate new products first.

Second, ratings and the shape of the reviews. Look past the star average to the recent one- and two-star reviews, because they name the return reasons: sizing runs small, battery dies in a week, color differs from photos. Every recurring complaint is a future clawback with your name on it. Third, the shop behind it: the Shop Performance Score and how long the shop has operated. In the pilot you're restricted to 95%+ shops anyway; after graduation, keep the habit — a weak shop's shipping delays become your comment section's problem and your audience's memory of you.

Fourth, the listing's own content: clear photos, a title a human can parse, variants that make sense. A sloppy listing converts your hard-won tap into a bounce, and the algorithm reads that bounce as your video failing. Also check stock depth — a product that sells out mid-video-run strands your best content with a dead link.

Put it together on a real-shaped example. Listing one: 61,000 sold, 4.7 stars, recent reviews praising exactly what the video would claim, shop score 97%, deep stock, clean page. Listing two: 300 sold, 4.9 stars on nineteen reviews, shop score 88%, one recent review saying “arrived broken, seller never replied.” Listing two pays 5 points more commission. Listing one is still the pick, and it isn't close: your scarce resource isn't products, it's posts — spend them where the whole chain under your video is sound.

Samples: free, refundable, and the obligation both carry

You can't honestly demo what you don't have, so samples are the supply line of this business — in two flavors. Free samples: a seller ships you the product at no cost so you'll create with it. As a rule, free-sample access through TikTok's system opens at 5,000 followers with at least one sale in the last 120 days, and sellers approve requests case by case, favoring creators whose niche and numbers fit the product. Refundable samples: you buy the product like a customer, and if your content hits the seller's stated sales goal, the purchase price is refunded. No approval gate, real skin in the game.

Both flavors carry the same fine print, and it's the part new creators blow past: a sample is a commitment. When a seller ships you a free product through the platform, you're expected to post shoppable content with it inside the window TikTok sets — miss it, and your sample privileges get suspended. Refundable samples put your own money on the same clock. So the rule is absolute: never request or accept a sample you aren't ready to film within days of its arrival. A drawer of unposted samples is not inventory; it's a list of broken promises with tracking numbers.

Before 5,000 followers, your supply line is simpler: buy cheap, buy smart. Start with products in your own home you already use (zero cost, genuine history), then buy a small number of proven, low-priced products in your category — the $15–$30 workhorses from Lesson 1 — treating the purchase price as your tuition. A creator who has posted twenty honest videos with self-bought products has a portfolio; that's what makes sellers approve their sample requests later.

When you do request samples, write the request like a professional: your niche in one line, your recent posting rhythm, the specific video you plan to make with the product, and when you'll post it. Sellers wade through hundreds of empty “send me one plz” taps. A two-sentence plan with a date on it is a business proposal, and it gets approved like one.

The product scorecard: your picking system

Intuition is what you use before you have a system; after this lesson you have a system. Score every candidate product on five lines before it earns a post. Per-sale pay (rate × price — under about $3 needs extraordinary volume to matter). Proof (units sold and review shape — is conversion already demonstrated?). Demo power (does it show a visible result, transformation, or satisfying action inside ten seconds?). Return risk (sizing, fragility, exaggerated listing claims — the clawback forecast). Fit (does it belong in your category, for your audience?).

Score each line 1 to 5, and set two hard gates regardless of total: demo power under 3 doesn't get posted (a product you can't show working is a radio ad), and return risk at the maximum kills the pick outright, however pretty the rate. Worked example: the $45 flosser from Lesson 1 scores pay 4, proof 5, demo 5, return risk low (scored as safety 4), fit 5 — 23 of 25, post it this week. The 30% posture corrector: pay 3, proof 2, demo 2, safety 1 — it fails the demo gate before the total even matters.

Keep scorecards in your notes file, including the products you rejected and why. In a month that file is your category's market research: which bands actually converted for your audience, which sellers' listings performed, which demo styles your viewers buy from. Sellers will eventually ask what works on your account, and the creator who answers with a scorecard history gets the target collaboration. The one who answers “idk, vibes” gets the open shelf like everyone else.