Your first week with Claude connected to your FBA business
1
Day 1: Verify the inputs
Ask which products are missing landed cost or lead time, and fix them together in the chat. All restock and profit math depends on these two fields.
2
Day 2: Store your operating rules
Tell Claude your cover-day floors, supplier shutdown windows, and seasonal patterns so it can apply them whenever a future answer depends on them.
3
Frequently Asked Questions
What should I do first after connecting Claude to my Amazon FBA data?
Verify the inputs before trusting the outputs. Ask Claude which of your products are missing landed cost or lead time, and fix the gaps in the same conversation. Restock timing and profit math both depend on those two fields, so twenty minutes of cleanup on day one makes every answer after it trustworthy.
How long does it take to get value from Claude as an Amazon seller?
With your data already connected and synced, most sellers get their first genuinely useful answer in the first session, usually a profit surprise or a stockout risk they had not spotted. The full routine, from verified inputs to a standing weekly review, takes about a week of 15-to-30-minute sessions.
Matthew Kerns is a software engineer at Inventory Hero, where he builds the MCP server and AI-assistant integrations that let sellers work their live inventory, sales, and supplier data from Claude. He also runs his own Amazon FBA brand, so the workflows he writes about get tested on his own inventory first.
Run the cost-data trust gate, then ask for the profit ranking: which products earn, which quietly lose money after fees.
4
Day 4: Log suppliers and open purchase orders
Dictate in-flight POs with their arrival windows so incoming stock counts in the restock math before you ask for recommendations.
5
Day 5: Run your first restock review
Ask what to reorder and how urgent each item is, then sanity-check the order-by dates against your own calendar.
6
Every Monday after: the standing review
One recurring prompt covering reorders, stockout risk, and margin drift. Fifteen minutes, then decisions.
What if a number Claude gives me looks wrong?
Ask it to show the breakdown rather than accepting or dismissing the answer. Grounded answers are built from auditable components (velocity, lead time, fees, incoming orders), so you can walk down the stack until you find either your bad input or your wrong assumption. Most 'wrong' numbers trace back to a missing cost or an unlogged purchase order.
Read article
Claude for Amazon sellers gets real the week after you connect it. The setup
takes a minute; the habit takes five short sessions. This guide is that first
week, day by day: verify the inputs, store your rules, find the profit leaks,
log what is on the water, and run your first restock review. Each session is 15
to 30 minutes, and the order matters, because every day builds on the one before.
This assumes you are already connected and through onboarding (the
setup guide covers that; Claude on
web and desktop needs a paid plan). If you want the full menu of what a connected
assistant can do before committing a week, skim the
eleven MCP use cases first; this
article is the adoption path through them.
Day 1Verify inputscosts + lead times
Day 2Store rulesfloors, seasons
Day 3Profit scorecardfind the leaks
Day 4Log POswhat's on the water
Day 5Restock reviewthe payoff
MonWeekly review15 min, repeat
Each session is 15 to 30 minutes, and every day feeds the next.
Every useful answer this week (what to reorder, what is profitable, what is
about to stock out) sits on two fields: landed cost and lead time. If
those are missing or stale, the math is confidently wrong. A missing lead time
quietly falls back to a deliberately conservative default (75 days), long enough
that it errs toward reminding you too early rather than too late, which can
still shift a recommended order date by weeks.
So start here:
"Which of my products are missing landed cost or lead time?"
"Show me my product list with costs and lead times so I can sanity-check them."
Then close the gaps in the same conversation:
"Set the landed cost on SKU ABC-123 to $6.40 and the lead time to 45 days."
"Here are the rows from my cost sheet: apply them as one batch."
(Three columns is enough: Seller SKU, landed cost, lead time in days.)
One nuance on lead time: it is really three legs (production, shipping,
receiving). If you track them separately, say which legs your number covers;
Claude will ask rather than guess. These are safe, non-destructive edits to your
own records, made only when you ask. Twenty minutes of this makes the rest of
the week trustworthy.
Your numbers live in your account. How you operate lives in your head: the
supplier who goes dark for three weeks around Chinese New Year, the 45-day cover
floor on bestsellers, the September start of your Q4 ramp. Day 2 moves those
into your account's business memory:
"Remember that my main supplier shuts down for three weeks around Chinese New Year."
"Remember that I never let my top five products drop below 45 days of cover."
How stored rules shape future answers (they change the assistant's read on the
numbers, not the math underneath) is covered in the
use-cases guide. The habit that makes
them pay is simpler: say each rule once, then ask Claude to check plans against
your rules whenever the stakes are high. For why this compounds, see
the AI employee for your FBA business.
"Before I look at profitability, which products have missing or incomplete cost data?"
A profit ranking over half-complete costs looks precise and is not. Once the
gate is clean, hunt:
"Which products actually make money after Amazon fees, and which are quietly losing it?"
"Walk me through revenue to profit on my best seller. What is my biggest fee lever?"
Here is the kind of surprise day 3 tends to produce. Say your best seller is an
oversize, higher-return SKU, and this is what it looks like once real costs and
your account's actual fees are in:
Where a $27.50 sale goesillustrative: oversize, higher-return SKU
Sale price$27.50
Landed cost
−$8.90
COGS (factory price)$6.40
Inbound freight⚡ IH helps here$1.60
Duties$0.55
Prep$0.35
→ Product Profit (CM1)$18.60· 68% of sale
That is a hero product keeping just $2.85 a unit, a
shipped margin (CM2) of about 10 percent, after
landed cost and every Amazon fee. Established operators
commonly target 15 to 20 percent after fees (an operator rule of thumb, not a
law), so a number like this is rarely a crisis but usually a surprise. The
follow-up ask, "what is my biggest fee lever?", tells you whether size
tier, returns, or storage is doing the damage. Sellers rarely find this in a
dashboard because nobody goes looking; in a conversation, it is one question.
The ⚡ marks where Inventory Hero does the finding for you. It cannot repackage
your product or renegotiate a freight rate, but it puts each lever one question
away:
Inbound freight: sharper forecasts mean fewer emergency air shipments,
which run several times the sea rate.
FBA fulfillment: see exactly how fulfillment fees hit your bottom line,
SKU by SKU.
Storage + LTSF: catch aging and overstock before long-term storage fees
pile up.
Placement + removals: the "other" Amazon fees, itemized.
And each lever points at a different move. If it is size tier, a packaging
redesign of a few ounces or a half-inch can drop you into a cheaper band. If it
is returns, the fix is upstream (a clearer listing photo, a sizing note, a
fragile-part tweak), not the fee itself. If it is storage, you are looking at an
aging or overstock problem, and day 5's restock review is where you start
unwinding it. A 10-percent hero is rarely a discontinue candidate; it is usually
one fixable lever away from 15.
If the system cannot see an order you placed three weeks ago, your first restock
review will tell you to buy stock you already own. Day 4 closes that loop:
"Add my supplier Shenzhen Great Wall: 30-day production lead time, 30 percent deposit terms."
"I placed a PO with them Tuesday: 2,000 units of ABC-123 at $4.10, landing in three weeks. Log it."
"Here are my open purchase orders, pasted from my sheet: record them as a batch."
Naive stock vs. inventory position
On hand (1,300)Incoming PO (2,000)
On-hand alone says reorder ABC-123 now; counting the 2,000-unit PO already on the water says wait.
This is the least painful data entry in the stack: read the invoice, dictate the
order with its arrival window, done. The
use-cases guide has a worked example of
exactly how an unlogged PO flips a restock answer from wrong to right.
The payoff, and the session that becomes your weekly habit:
"What should I reorder this week, and how urgent is each one?"
"Which products are about to run out, and how many days do I have?"
The answer comes back as a ranked list. One illustrative row:
SKU
Urgency
Days of cover
Order by
Suggested qty
DEF-456
Critical
19
July 15
1,800 units
That order-by date is not guesswork. For any SKU it is the same back-plan: the
day you would otherwise run dry, minus a full lead time and a safety buffer. The
offset is easiest to see on a longer-lead product, where it stretches to weeks:
Working backwards from the rampWhy the order-by date lands where it does: the stockout date minus your 45-day lead time and a safety buffer.
Each line carries a breakdown you can ask it to expand: velocity, incoming
stock, seasonality. The recommendations account for your lead times and the POs
you logged on day 4, which is exactly why ABC-123, with its 2,000-unit order
already on the water, is not the one topping this list. Sanity-check the
order-by dates against your own calendar this first time, line by line, before
you start trusting the cadence.
For the underlying method, see
when to reorder inventory.
"Run my weekly review: what needs reordering and how urgent, what is at
stockout risk in the next 30 days, and did margin move on any of my top
products? Flag anything unusual."
Fifteen minutes, coffee in hand, then decisions. Because you spent day 1 on
inputs and day 2 on rules, the answers keep getting better: the assistant knows
your suppliers, your floors, your season. Once that weekly rhythm is automatic,
the bigger jobs open up. Start with the
quarterly planning deep dive,
where the same connected data drives a full season's buy plan. And any time you
hit a rough edge, tell Claude to "send feedback"; it goes straight to the
team building this.
Amazon fees
−$15.75
Referral (15%)$4.13
FBA fulfillment⚡ IH helps here$7.20
Storage + LTSF⚡ IH helps here$2.10
Placement + removals⚡ IH helps here$2.32
→ Shipped Profit (CM2)$2.85· 10% of sale
⚡ Inventory Hero helps with this lever.
Shipped Profit (CM2) is what's left after every Amazon fee but before advertising. It is contribution margin, not net profit.