Amazon’s restock recommendations are reasonable for sellers with a single product, single region and predictable demand. For everyone else, the default algorithm leaves money on the table — either by overstocking aged inventory or running out at peak.

This guide is about what a smarter restock signal looks like, and how to build one.

TL;DR: Amazon’s restock tool optimizes for FBA capacity and category-wide patterns. It does not know your real lead times, your promotional plans, your batch-level COGS or your acceptable stockout rate. A smarter restock signal joins your real velocity, your real lead time and your real working capital constraints to recommend reorder quantities that match how your business actually runs.

How Amazon’s algorithm works

Amazon’s restock recommendations are based on:

• Historical FBA velocity for the SKU.
• Category-wide demand patterns.
• FBA storage capacity assumptions.
• Lead times based on what Amazon estimates, not what you actually see.

It is generic by design. It optimizes for Amazon’s network, not your bottom line.

Where the default falls short

• Lead times you do not control. If your supplier ships in 6 weeks but Amazon assumes 4, every recommendation is too late.
• Promotional plans the algorithm cannot see. A planned ad push in 30 days needs more inventory than baseline velocity suggests.
• Working capital limits. Amazon does not know that ordering 10,000 units puts your cash flow underwater.
• Multiple sourcing strategies. Splitting orders across suppliers, batching for sea vs air freight — invisible to Amazon.
• Cross-marketplace inventory pooling. EU sellers using Pan-European FBA see different optimal reorder logic than the algorithm exposes.

What a smarter signal looks like

The components of a useful restock recommendation:

Real velocity, weighted

Trailing 30-day FBA velocity is the baseline. Weight by recent weeks more heavily than older ones if demand is shifting.

Real lead time

Your supplier lead time, plus your inbound shipping, plus FBA receiving — measured from your last few orders, not estimated.

Target days of cover

How many days of stock you want as buffer past lead time. Conservative sellers run 60–90 days. Aggressive ones run 30.

Promotional adjustments

Planned ad spend or promotions in the next reorder window — adjust velocity assumptions upward.

Cash flow constraints

Cap reorder quantity by available working capital and split the order if needed.

Per-marketplace logic

Each Amazon region has its own velocity, FBA capacity and lead time. The math runs per-region, not globally.

Putting it in practice

Once your data layer has FBA inventory, sales velocity, lead time history and your COGS, building a restock recommender is an afternoon’s work with an AI builder:

“Build me a CLI command that recommends weekly reorder quantities for each SKU based on 30-day velocity, my last three lead times, a 60-day cover target and a $50,000 working capital cap. Include reasoning.”

The AI ships the script. You review the recommendations against your gut. The team runs it weekly.

The bottom line

Amazon’s restock tool is fine as a starting point. For a real operation it is wallpaper over a real planning problem. The smarter signal is not exotic math — it is your real data, joined and queryable.

DataDoe’s Amazon data layer exposes FBA inventory, velocity, lead times and your COGS in one place, ready for AI builders to reason against.