Recovery is defense. This is offense.
Clawing back an overcharge is good. Designing the product so the overcharge never happens is better. We model your packaging against Amazon's size-tier boundaries and tell you exactly what to change — before you place the PO.
Amazon's fee table is a staircase, not a ramp
FBA fees are not proportional to size. They step. Cross a threshold on your longest side, your median side, your unit weight, or your dimensional weight, and the fee jumps to the next tier — the same jump whether you exceeded it by four inches or by one eighth of one.
Which means a large share of catalogs are sitting just barely on the wrong side of a line, paying a full tier more than they need to, permanently, on every unit shipped.
The expensive part: nobody finds out until the product is manufactured, packaged, shipped, and selling. By then the box is tooled, the inventory is inbound, and changing it means eating a production run.
Half an inch, sixty thousand units
A carton comes in at 18.3 inches on its longest side. The tier boundary is 18. Trimming the box by four tenths of an inch — achievable with thinner corner inserts and a tighter fold — moves the SKU down a full tier.
Illustrative figures, not a quote. Real numbers depend on your tier, weight, category, and velocity — which is the entire reason to model them rather than guess.
And unlike a reimbursement, this saving does not need to be claimed, documented, appealed, or filed within sixty days. It just stops happening.
What actually moves a SKU down a tier
Rarely the product itself. Almost always the packaging around it.
Box geometry
Longest side, median side, and shortest side are measured independently, and each has its own thresholds. Reproportioning a box — same volume, different shape — frequently moves a tier without changing a thing inside it.
Compression & fill
Soft goods, textiles, and anything with air in it are usually measured larger than they need to be. Vacuum sealing, roll-versus-fold, and removing void fill are the cheapest wins in the catalog.
Poly bag vs. box
A rigid box locks in its dimensions. A poly bag is measured as packed. For the right products this is a one-line spec change to your manufacturer with a permanent fee reduction attached.
Dimensional vs. unit weight
Your fee is driven by whichever is greater. Light bulky products are billed on air. Knowing which one governs a given SKU tells you whether to attack size or weight — attacking the wrong one is wasted effort.
Multipack maths
A 2-pack looks like an obvious AOV win until it crosses a tier and the per-unit fee rises faster than the price. We model every bundle configuration before you commit to one.
Case pack & inbound
How units are cased affects inbound placement and prep costs, not just fulfillment. The cheapest fulfillment configuration is not always the cheapest all-in configuration, so we model the whole path.
Modelling, then judgment
The software does the search. A human decides what is actually buildable.
Map the catalog
We take your current dimensions, weights, categories, and sales velocity, and locate every SKU relative to its nearest tier boundaries — in every dimension at once.
Rank by opportunity
Distance to the boundary multiplied by fee delta multiplied by annual volume. That ordering is what separates a genuine six-figure packaging change from a rounding error, and it is almost never the SKU you would have guessed.
Search the configurations
For each high-value SKU we model alternative packaging configurations — box proportions, materials, compression, bag versus box, bundle counts — against the fee table, and rank them by projected annual saving.
Sanity-check it
Then a person who has actually shipped product decides whether each option survives contact with reality: will it protect the item, will the factory build it, does it wreck the unboxing, does it change your prep or your inbound cost.
Hand you the spec
You get a per-SKU sheet: current tier, target tier, the exact dimensional change required, the packaging change that achieves it, projected fee delta per unit, and annualized saving at current velocity. Written so you can forward it to your supplier without translating it.
When to use it
- Before a new product is tooled. The cheapest moment to change a box is before it exists. Give us the target dimensions and we will tell you where the walls are.
- Before your next PO. A packaging revision costs almost nothing if it lands with the production run rather than after it.
- After a remeasurement. If we corrected a mismeasured SKU, that is the moment to ask whether the packaging should also change.
- When a fee schedule updates. Amazon moves thresholds. SKUs that were comfortably inside a tier can find themselves on the wrong side of a line without you touching anything.
- Across the whole catalog, once. Most sellers have never had this modeled at all. The first pass is usually where the surprises are.
An honest note on the "AI" part. What is automated is the search: evaluating thousands of packaging configurations against the fee table and your velocity data, and ranking them by annual saving. That is arithmetic at a scale no one wants to do by hand, and it is genuinely good at it. What is not automated is deciding whether a configuration is manufacturable, protective, and worth doing. That part is a person, because it has to be.
What it costs
Free for recovery clients. If we are already auditing your account, catalog modeling is included — we have your dimensions and velocity already, and a client who pays lower fees forever is a client who stays.
Standalone: flat project fee, scoped to catalog size, with no commission attached. We are not taking a percentage of your savings, because measuring "savings" against a counterfactual is exactly the kind of arrangement that gets argued about later. Ask for a quote.
Find out what Amazon owes you.
A free audit takes ten minutes to set up. You get a findings report with real dollar figures, and no obligation to do anything with it.