Skip to main content

Prompt library

35 Claude prompts for your Amazon data

You connected the MCP and the chat box is empty. These are the questions worth asking first.

Written for Claude, but nothing here is Claude specific. The same prompts work in ChatGPT, Gemini and any other MCP-capable client, because the numbers come from your account, not from the model.

35
prompts
7
groups
109 ms
median tool call
Try for Free
Rated 4.7/5 onTrustpilotTrusted by 3,000+ Amazon sellers

Try this one first

Which of my products are losing money?

Which of my SKUs had negative net profit over the last 30 days? Show net sales, net profit and net margin, worst first.

The one that replaces a routine

Your Monday morning, in one answer

Give me my Monday briefing: sales and net profit last week versus the week before, anything with under 14 days of stock, listings that broke, and my three worst campaigns by ACOS. Keep it under one page.

Profit, stock, listings and ad waste, pulled by four tools and written as one page. That is a 40 minute routine gone. To send it to a client every week, see Amazon reporting with AI.

P&L, last week vs the week before

  • Net sales $184,200, +6.1%
  • Net profit $21,940, -3.4%
  • Fees took 31.2% of sales, up 1.1pt

Under 14 days of stock

  • 4 SKUs at risk
  • Shortest cover 6 days at 42 units/day
  • 2 shipments inbound, neither lands in time

Listings that broke

  • 1 suppressed on image policy
  • 1 lost the buy box to a 4% undercut
  • Both sit behind trailing sales over $30k

Three worst campaigns by ACOS

  • Auto - Kitchen, ACOS 118%, $2,410 spend
  • Exact - core term, ACOS 84%
  • PT - competitor, ACOS 76%

Answer shape shown with synthetic figures, invented for this example.

Routines that chain several tools

3 prompts

Start here. Each of these pulls from four or five tools and hands back one answer, which is the thing a dashboard cannot do.

Profit and margin

7 prompts

Your P&L reconciled across 40+ fee types before the assistant sees it, so the margin it quotes is the margin you bank.

  • Which of my products are losing money?

    Tested live
    Paste this

    Which of my SKUs had negative net profit over the last 30 days? Show net sales, net profit and net margin, worst first.

    A ranked table of loss-making SKUs and what each one costs you, usually hidden behind healthy looking revenue.

  • Where is my margin actually going?

    Paste this

    Break down last month for me: net sales, total Amazon fees, ad spend, COGS and net profit. Then show the same figures for the month before and tell me what changed.

    A two-period P&L with the deltas called out.

  • Show me a month by month P&L for the year

    Paste this

    Give me net sales, net profit and net margin by month for the last 12 months, in one table.

    All 12 rows from a single query.

  • Which best sellers do not make money?

    Paste this

    Which products are in my top 20 by revenue but sitting below 5% net margin? Show revenue, margin and what is eating it.

    The revenue versus profit mismatch list, the one most sellers only find at year end.

  • Audit my cost setup

    Paste this

    Audit my COGS inputs. Which SKUs have no cost, a zero cost, or a cost that looks like it was entered per carton instead of per unit? Flag anything implausible.

    A setup audit against the raw values you entered. This is the prompt that makes every other profit number trustworthy.

  • Are Amazon's fees creeping up on me?

    Paste this

    How did total Amazon fees change as a share of net sales this month versus last month, and which SKUs drove the change?

    Fee drift attributed down to SKU, usually a size tier change nobody told you about.

  • How much of my December profit will January returns take back?

    Paste this

    Show my December net sales, units sold and net profit, then show refunded units and refund value week by week through January and February against the same SKUs. Tell me what share of the December profit the returns gave back.

    Peak-season profit restated after the return wave, which is the only version of the number worth banking on.

Amazon PPC and wasted spend

7 prompts

Sponsored Products, Sponsored Brands and Sponsored Display at campaign, ad group and keyword grain, plus the customer queries behind the spend.

These read Sponsored Products, Sponsored Brands and Sponsored Display at campaign, ad group and keyword grain. How this differs from Amazon's own Ads MCP.

  • Which campaigns are wasting money?

    Paste this

    Which campaigns spent the most over the last 30 days with the worst ACOS? Show spend, sales, ACOS and ROAS, worst first.

    The cut list, ranked by money at stake rather than by percentage.

  • Find my negative keyword candidates

    Tested live
    Paste this

    Find search terms from the last 30 days with at least 6 clicks and zero orders, ranked by spend. These are my negative keyword candidates.

    The real customer queries behind your spend. A 30 day window on an active account routinely covers over a thousand distinct terms.

  • What should I promote to exact match?

    Paste this

    Which search terms in my auto campaigns converted above my account average over the last 60 days? I want to promote the winners to exact match.

    The discovery to exact pipeline, the most repeated manual task in Amazon PPC.

  • Which keywords should I bid down?

    Paste this

    Which keywords have a CPC above my account average but a conversion rate below it? Show spend, clicks, orders and CPC.

    The overpaying and underconverting list, at keyword level.

  • Show me my TACOS trend

    Paste this

    Show my TACOS by month for the last 6 months, alongside net sales and ad spend.

    Whether your ad dependency is rising, with the two inputs that explain it.

  • Which products soak up my ad budget?

    Paste this

    Which of my ASINs take the most ad spend, and what share of each one's sales is ad driven?

    Spend concentration by product and the organic versus paid dependency behind each.

  • Is my ACoS above or below the market?

    Paste this

    Show my ACoS, ROAS and TACoS for the last 90 days, split by marketplace, and tell me which marketplaces are dragging the blended figure. Include spend and ad-attributed sales for each one.

    Your own efficiency per marketplace, in the shape you can hold against your own margin instead of against a target somebody guessed.

Keywords and market share

5 prompts

Amazon Search Query Performance from Brand Analytics, curated: the whole market's funnel on a query next to your slice of it.

Search Query Performance arrives curated, so you ask for your share of a query instead of parsing report rows. See query_sqp in the tool reference.

  • Which keywords am I losing share on?

    Tested live
    Paste this

    Using the latest complete SQP week, show me core and presence queries where search volume grew week over week but my impression share is under 10%. Sort by search volume.

    Growing demand you are absent from, each query tagged with a verdict and an action such as fix CTR or defend. That verdict layer is ours.

  • Which big queries am I invisible on?

    Paste this

    Which high volume customer queries already generate purchases for me but where I hold under 1% impression share? Flag the untapped ones.

    Proven demand you are barely present on, separated from category noise you should ignore.

  • Do I convert better than the market?

    Paste this

    On my core queries, compare my click and purchase rates to the whole market's on the same query. Where am I below the market?

    Like for like against every other seller on that query. Below-market CTR is a main image problem, below-market purchase rate is a price or reviews problem.

  • Which competitor brands are taking my sales?

    Paste this

    Which competitor brand queries drive purchases for my products, and which of my own brand queries are competitors winning?

    Conquesting in both directions, using the query type classification rather than string matching.

  • Am I bidding on the keywords that matter?

    Paste this

    Take my top 20 SQP queries by search volume for the latest complete week, then check whether I am bidding on them in Sponsored Products over exactly the same days. List the ones I am not targeting.

    The gap between organic demand and paid coverage, on two calendars that are forced to line up.

Inventory and restock

5 prompts

Stock cover computed the way the Nova stock page computes it, so the days of cover in the chat match the dashboard.

  • What am I about to run out of?

    Paste this

    Which SKUs have under 21 days of inventory left? Sort by days of inventory ascending and include sales velocity and units on hand.

    The stockout queue in date order.

  • Where is my dead stock?

    Paste this

    Show me FBA stock that has not sold in the last 30 days, with the units sitting there and what they are worth.

    Capital parked in Amazon's warehouses and the storage fees it keeps generating.

  • Build my restock list

    Paste this

    Build my restock list: every SKU flagged at out of stock risk, with velocity, days of inventory, units available and units inbound.

    A purchase order starting point that already accounts for stock in transit between fulfilment centres.

  • How long until I am truly out, FBM included?

    Paste this

    Counting my FBM stock as backup, how many days before each SKU is completely out of stock?

    The total stockout view. Most tools show FBA only and tell you that you are out when you are not.

  • Am I ready for the Q4 storage jump?

    Paste this

    For every FBA SKU, show units on hand, days of inventory and units sold in the last 30 days, then flag anything holding more than [number] days of cover going into Q4. I want to know what will still be sitting in the warehouse through the expensive months.

    Cover per SKU read against the months when storage costs the most, so you can pull slow units out before the rates step up.

Listing health and buy box

4 prompts

Current state per SKU rather than a report from yesterday, with the enforcement reason attached.

  • What is broken on my listings right now?

    Paste this

    Which of my listings are suppressed, removed, or not buyable right now? Show the worst severity first with the driving issue.

    Live state, with the specific enforcement reason per listing.

  • Who took my buy box?

    Paste this

    Which of my listings lost the buy box, who owns it now, and at what price? Include any hijackers and anyone undercutting me.

    Owner, price, undercut and hijack flags per contested listing.

  • Audit my listing content

    Paste this

    Audit my active listings for missing bullet points, missing description, missing backend keywords, or fewer than 5 images.

    A content gap list from published catalogue data plus the backend keywords you actually submitted.

  • What should I fix first?

    Paste this

    Rank my listing problems by how much revenue sits behind them, then tell me the three to fix today and why.

    Triage weighted by trailing sales, so a broken listing doing 40k outranks a broken listing doing 40.

Diagnostics: why did this move?

4 prompts

Narrow to the products responsible, then pull the signal bundle behind one of them.

  • Which products lost the most sales?

    Paste this

    Which SKUs lost the most net sales this month compared to last month? Show the drop in both currency and percent.

    The biggest losers ranked by money lost, which is rarely the same list as biggest percentage drop.

  • Why did this product drop?

    Paste this

    SKU [your-sku] dropped. Pull traffic, conversion, price, buy box, promotions, ads and rank for the last 14 days against the 14 before, and tell me the most likely cause.

    The whole causal signal bundle in one call, so the assistant reasons over evidence instead of guessing.

  • How is my Best Seller Rank moving?

    Paste this

    How has BSR moved over the last 90 days for my top 10 ASINs, in their main categories?

    Rank history with the category and rank pair kept intact, per marketplace.

  • My sales are down, find the cause

    Paste this

    My sales are down this week. Find the SKUs responsible, then explain the biggest one in detail and tell me what to do about it.

    A two step investigation, narrowing then diagnosing. This is the one that shows why an MCP beats a dashboard.

The flagship prompt

Am I bidding on the keywords that matter?

Search Query Performance weeks run Sunday to Saturday and publish about 10 days late. Advertising weeks start Monday. Nova forces the two periods onto the same calendar days before either table reaches the assistant.

Hand both tables over raw and the assistant compares mismatched weeks, then reports the mismatch as a finding. Confident, tidy, wrong.

Two calendars, one question

SQP week

Sun to Sat, ~10 day lag

Advertising week

Mon to Sun

What Nova sends

Both tables cut to exactly the same days

Alignment handled server side, not in the prompt

Not connected yet?

Authorise Seller Central, paste the endpoint into your assistant, then come back here for the first question.

Three things to know before you point an assistant at your account.

Answers in seconds

Pre-aggregated responses, ranked server side, light on tokens.

Revocable

Per-account credentials you can pull back at any time.

Minutes, not a project

Authorise, paste one endpoint, 24 months of history backfills.

Ask your own account these questions

Connect Seller Central, point your assistant at the Nova endpoint, run the Monday briefing on real rows.

Try for Free

14-day free trial, no card required. MCP reads by default.