Every resale decision reduces to one question asked at the wrong time by most people: what is the most I can pay for this and still make money?
It gets asked at the wrong time because people source first and work out the economics later. The number is knowable in advance, and it's not the sale price.
The ceiling
You don't keep the sale price. Platforms take a cut, and shipping comes out of
your side. So for a sale price P, a platform fee rate f, and shipping cost
S, what actually lands is:
net = P × (1 − f) − S
Set profit to zero and you get the maximum you can pay for the item, the sourcing ceiling:
C_max = P × (1 − f) − S
Illustratively: an item that sells for $100, on a platform taking 13%, shipping
for $12, nets 100 × 0.87 − 12 = $75. That is break-even. Pay $75 and you've
worked for free. Pay $40 and you have $35 before anything else goes wrong.
Fee rates vary by platform and category and change over time, so use the current
number for the specific category rather than a remembered one. The structure is
what's stable: the fee is a percentage of the sale price, so it scales with
P, while shipping is closer to fixed. Cheap heavy items get destroyed by S.
Expensive light items are where the arithmetic is kindest.
The number that matters more
Margin alone doesn't tell you how good a deal is, because it says nothing about time. Money tied up in something that isn't selling isn't available for the next thing.
The measure that captures that is return over the holding period, annualized:
ROI = profit / cost
annualized ≈ ROI × (365 / days held)
Which produces the result that surprises people. Compare:
Item A: 50% ROI, sells in 14 days → 50% × (365/14) ≈ 1300%/yr
Item B: 200% ROI, sells in 400 days → 200% × (365/400) ≈ 183%/yr
Item B has four times the margin and is a far worse use of the same dollar. A thin margin that turns over repeatedly beats a fat margin that sits, because the same capital gets used again and again.
This is why "I'll hold out for my price" is often an expensive instinct. Holding for 20% more on an item that then takes six months to move usually loses to selling now and redeploying.
Not everything sells
The last correction is that C_max assumes the sale happens. Some things don't
sell, or sell far below the comp you based the ceiling on. If q is the
probability it sells near your expected price:
EV = q × profit − (1 − q) × (loss on the unsold)
Which means sell-through rate belongs in the sourcing decision alongside margin. A category with reliable demand at a modest spread is frequently a better business than one with huge spreads and thin, unpredictable demand: you get paid more often, and you find out you were wrong faster.
The practical version
Work backward, always. Look up what the thing actually sold for recently, not what it's listed for, since listings are asks and asks include fantasies. Subtract the fee percentage. Subtract real shipping including the box. That's your ceiling. Then discount it for how confident you are it moves at all.
Whatever's left is what you can pay. If that number is below what it costs to acquire, the answer is simply no, and the discipline is in walking away often enough that the deals you do take are worth the handling.