Before sourcing, run every candidate product through the same profitability lens so you're comparing ideas on equal footing. The single biggest mistake at this stage isn't bad math — it's incomplete math: modeling landed cost against sale price and calling that "the margin," while leaving out costs that are entirely predictable but easy to forget when you're excited about an idea.

The core question isn't "is there margin" — it's "is there margin after everything"

A product with an apparently healthy 50% gross margin against landed cost can become unprofitable once you add: marketplace referral fees, realistic advertising spend to actually get found in a competitive category, a realistic return rate for the category, and payment processing. Use the Product Profitability Calculator to model all of these together rather than checking margin against cost alone.

The full cost stack, laid out

  1. Landed cost — unit cost plus freight, duties/tariffs, and any inbound handling to get the product into sellable condition at your fulfillment point.
  2. Marketplace referral/commission fees — a percentage of sale price that varies by category (see How to Research a Marketplace Category).
  3. Fulfillment fees — per-unit pick/pack/ship fees if using a marketplace fulfillment program, which scale with size and weight, not just price.
  4. Storage/carrying cost — ongoing cost of holding inventory, which compounds for slower-turning SKUs.
  5. Advertising cost to achieve realistic visibility — in any moderately competitive category, some level of ongoing ad spend is usually required to sustain sales volume, not just to launch.
  6. Returns — a realistic, category-appropriate assumed return rate, including the cost of processing/restocking (or writing off) returned units.
  7. Payment processing — typically a smaller line item but still real, and easy to forget when modeling by hand.

Category-specific cost traps to watch for

  • High-return categories (apparel, anything sized) — build a higher assumed return rate into your model, and account for the fact that some returned units in these categories can't be resold as new.
  • Heavy/bulky items — fulfillment fees can be disproportionately large relative to price; check actual fulfillment fee tiers before assuming a "normal" percentage, since size-based fee tiers can jump sharply at certain weight/dimension thresholds.
  • Highly advertised categories — if the top listings are clearly running heavy ad spend, assume you'll need to as well to get comparable visibility, and model that cost in from the start rather than treating advertising as optional upside.
  • Categories with frequent price competition — if incumbents compete aggressively on price, model your margin at a realistic competitive price point, not at an aspirational price you'd prefer to charge.

A useful gut-check threshold

If a product's modeled net margin (after all realistic costs) is below roughly 15-20%, it usually has little room to absorb the unexpected — a shipping cost increase, a slower sales ramp than planned, or a period of heavier-than-usual advertising to fight off new competition. Products in the mid-30%+ modeled margin range have meaningfully more room to survive a rough patch.

Worked example

A candidate product costs a certain landed amount per unit and is expected to sell at a specific retail price. A naive gross-margin calculation (price minus landed cost, divided by price) might show a healthy margin in the 45-55% range. Once the full cost stack is modeled — a category-typical referral fee percentage, a per-unit fulfillment fee based on the product's actual size/weight tier, an assumed return rate typical for the category, and a realistic ongoing advertising cost per unit sold based on what comparable listings appear to be spending — the modeled net margin often comes in dramatically lower, sometimes into single digits or negative territory for a product that looked strong on the naive calculation alone. Running this full-stack model before committing to a purchase order is the single highest-leverage habit in product research, because it's the step most likely to change a go/no-go decision.

Sensitivity-testing your model

Because several of the inputs (advertising cost, return rate, actual sell-through pace) are estimates rather than known figures before launch, it's worth testing how sensitive your margin conclusion is to each one:

  • Model a "base case" using your best realistic estimates.
  • Model a "stress case" with a somewhat higher return rate and somewhat higher required ad spend than your base case assumes.
  • If the product still clears your minimum acceptable margin threshold in the stress case, you have real confidence in the decision. If the stress case erases the margin entirely, treat the opportunity as marginal even if the base case looked fine.

Mistakes to avoid

  • Calculating margin against landed cost only, without a full cost-stack model.
  • Assuming zero or near-zero ongoing advertising cost in a category where competitors are visibly running ads.
  • Using a generic, marketplace-wide fee assumption instead of the category-specific fee rate.
  • Modeling only a best-case scenario and skipping a stress test of the key assumptions.

FAQ

What net margin should I aim for? There's no universal number, but treat anything below roughly 15-20% modeled net margin as having little cushion for the unexpected, and treat the mid-30%+ range as comfortably resilient. Your own risk tolerance and how much capital you have available to weather a rough stretch should inform where in that range you're comfortable operating.

Should I model margin before or after I have a real supplier quote? Do a rough pass with estimated costs early to screen out obviously weak ideas, then redo the full calculation with a real supplier quote before committing to a purchase order — supplier estimates and quotes for actual production runs can differ meaningfully.

How do I estimate a realistic advertising cost before I've run any campaigns? Look at how aggressively comparable listings in the category appear to be advertising (sponsored placement density is a rough visual proxy), and build a conservative estimate into your stress-case model rather than assuming organic sales alone will be sufficient.