ホーム / Skills / EC運営 / ネットショップの単位経済と価格設計(Unit Economics)
EC運営

ネットショップの単位経済と価格設計(Unit Economics)

AIBarsMITSkill の言語: 英語
ライセンス確認済みスクリプトなし低リスク
デモ実行記録 · claude-sonnet-5-5 · 2026年10月
I sell a stainless steel travel mug on my own web store. Price is 29.00 after my usual discount, I pay 7.50 landed cost per unit, 4.20 shipping out, 0.60 packaging and 1.00 pick and pack. Payment fees are 2.9% plus 0.30 and the store platform takes 2% of each order. About 6% of orders get returned, and a return costs me 4.20 return shipping plus 1.00 handling, and about a quarter of returned mugs can't be resold. My ad dashboard says my ROAS is 3.1 and I'm paying 11 per order in ads. Am I actually making money, and what ROAS do I need to hit a 10% margin after ads?
Skill: ecommerce-unit-economics

Short answer: yes, but barely, if the $11 per order is right. At $11 you keep about $1.26 per order (4.4%). The 3.1 ROAS on your dashboard doesn't match your own numbers, so check that first.

Cost per order (price 29.00)

Line Calculation Amount
Revenue kept after refunds 29.00 × (1 − 6%) 27.26
Product cost, net of resold returns 7.50 × (1 − 0.06 × 0.75) −7.16
Outbound shipping, packaging, pick and pack 4.20 + 0.60 + 1.00 −5.80
Payment fees 29.00 × 2.9% + 0.30 −1.14
Platform fee 29.00 × 2% −0.58
Return shipping and handling 6% × (4.20 + 1.00) −0.31
Contribution before ads 12.26 (42.3% of price)
Ads −11.00
Contribution after ads 1.26 (4.4% of price)

Two assumptions are mine. Payment and platform fees are not refunded on returns. Resellable returned mugs go back into stock at full cost.

Targets

  • Break-even CPA: 12.26 per order.
  • Break-even ROAS: 1 / 0.423 = 2.36.
  • ROAS for a 10% margin after ads: 1 / (0.423 − 0.10) = 3.10. That is a CPA of about 9.36 at a 29.00 order.

Your ROAS figure doesn't match your ad cost

  • At 29.00 per order and 11 in ads, your real ROAS is 29 / 11 = 2.64. That is above break-even (2.36) but well under the 3.10 you need for 10%.
  • A dashboard ROAS of 3.1 with 11 per order would mean an average attributed order of about 34.10. That could be multi-item orders or shipping revenue.
  • It could also mean the "11 per order" is spend divided by all orders, while the platform's ROAS only counts the orders it claims.
  • If the 3.1 is real on a 29.00 order, ads cost about 9.35 per order, and you'd be right at a 10% margin.
  • Use total revenue divided by total ad spend for the period, not the platform's number. That is your blended ROAS, and it's the one to compare to 2.36 and 3.10.

Sensitivity (change in profit per order)

  • Price ±10% (±2.90): about ±2.58. This is the biggest lever. A 10% discount deeper than your usual one would wipe out your profit.
  • Ad cost ±10% (±1.10): ±1.10.
  • Product cost ±10% (±0.75): about ±0.72.
  • Return rate +1 point: about −0.29. The refunded revenue matters far more than the return shipping.

At 4.4% net, a small rise in ad cost or returns pushes you to break-even. The ad cost per order is the number to pin down.

What to verify

  • Blended ROAS and cost per order over a period with a decent number of orders. I don't know how many orders your 11 rests on.
  • Whether your payment processor refunds its fee on returns. If it does, you do slightly better than shown.
  • Whether the 6% return rate and the 25% unsellable share come from real data.
  • Whether you charge shipping to customers. I assumed none.
  • Cash flow, since you buy the stock before you sell it.

If you can get ad cost per order down to about 9.35, or raise the price about 2 to 3 dollars, you reach 10%. I can run either scenario.

Real run in an isolated folder with only this skill installed. Only the Skill and Read tools were enabled, so no command was run and no file was written. The store and all numbers are invented and the model had no access to a real store. The per-order figures were checked by hand afterwards. The currency symbol in the answer is the model's own assumption: the request gave no currency.

できること

販売者自身の数字から1 注文の利益を計算します。返金後に残る売上、商品原価(返品されて再販できる商品は在庫に戻る)、送料、梱包、ピッキングと梱包作業、決済・プラットフォーム手数料、返品コストを扱います。この貢献利益から、損益分岐の CPA と ROAS、望む利益率のための目標 ROAS、目標利益率に必要な価格を導きます。割引は新しい価格でモデル全体を計算し直し、同じ総利益を得るために必要な販売増を報告して判断します。顧客生涯価値は、売上や根拠のないリピート率ではなく、利益と実測したリピート率から作ります。

組み込まれた習慣

数字を作らず、各入力の出どころを示し、プラットフォーム報告の ROAS が注文金額と合わない点を指摘し、感度分析(価格、広告費、原価、返品率)を示し、確認すべき事項を挙げます。参考ファイルに数式と計算例があります。

向いている場面

商品や広告施策が儲かっているかの確認、価格設定、割引をどこまでできるかの判断。

補足とリスク

低リスク:スクリプトのない指示のみのパッケージで、ネットワーク接続やファイル書き込みはありません。これは計算の方法であり、税務・会計・法律上の助言ではありません。手数料率、税の扱い、プラットフォームの規則は国によって異なり変更されるため、ご自身で確認して入力してください。結果は入力した数字次第で、予測ではありません。お店のデータは見えず、仮定した入力は明示します。試用では、通貨が示されていないのにモデルが自分で通貨記号を付けました。AIBars 制作のオリジナル(MIT)。架空のタンブラーのお店で 1 回試用しました。