# A daily Amazon brief with anomaly alerts

A playbook for an AI assistant connected to Nova. Every morning it reads yesterday against the same day last week, flags what moved more than usual in sales, ads, stock and listings, and says what to look at first. It only reads your data.

Every step calls a Nova tool that reads your live Amazon data, so paste this into Claude, ChatGPT or any assistant connected to Nova's MCP server. Run it each morning, or on a schedule if your assistant supports scheduled tasks.

```text
Use Nova to write my daily Amazon brief for yesterday and flag anything unusual.

Day: yesterday. Baseline: the same weekday last week, so weekday patterns do not read as alerts.

1. Scoreboard. Call Nova's query_metrics for yesterday against the baseline with netSales, units, netProfit, adSpend, sponsoredAcos and sponsoredTacos. Then the same for the last 7 days against the 7 days before. If a figure for yesterday is still incomplete, say so rather than reading it as a drop.

2. What moved. Call Nova's find_movers on netSales for yesterday against the baseline, direction "down", dimension "childAsin", limit 5. Then direction "up". Ignore rows with an empty id or __UNALLOCATED__.

3. Flag anomalies. Flag a product when its net sales moved more than 30% and more than a tenth of yesterday's total change. Flag the account when ad spend rose more than 25% while sponsored sales did not.

4. Explain the flags. For each flagged product, call Nova's explain_movement for yesterday against the baseline. Say whether price, Featured Offer share, promotions, ad spend or clicks moved with it.

5. Stock. Call Nova's query_inventory with oosRiskFlag true, sorted by daysOfInventory ascending, limit 10. Name what runs out within 14 days.

6. Listings. Call Nova's query_listing_health with bands [0, 1], sort "severity", limit 10. Name any suppressed or blocked top seller.

7. What changed. Call Nova's list_events for the last 2 days.

8. Write the brief, under 200 words: one line on yesterday, the scoreboard, the flags with their likely cause, stock and listing risks, and the one thing to do today.

Rules: these are correlations, so never state a cause as proven. Leave out anything that is normal. Say which figures are missing and why, and relay every notice Nova returns. Edit a Seller Central listing only when I ask and an Admin or Workspace owner confirms the preview. Never change prices, stock or advertising.
```

## How to read the result

| What you see | What it usually means | What to check |
| --- | --- | --- |
| Sales down on one product, Featured Offer share down | Another offer took the Buy Box | The Buy Box loss playbook |
| Ad spend up, sponsored sales flat | Bids or budgets rose without return | Campaigns and search terms in Amazon Ads |
| Sales down across the account, no single product | Demand or a site-wide issue | The same day last year, list_events |
| Top seller under 14 days of stock | Lost sales ahead | Inbound shipments and the reorder playbook |
| Suppressed top seller | The listing cannot sell | Seller Central's listing quality page |

## Limits

- Yesterday's figures move for a few days as orders, ads and traffic finish loading. Sessions and conversion refresh once a day.
- The thresholds in step 3 are a starting point: change them to suit your account.
- Stock and listing health are the state today, not a history.
- Nova reads your data and can edit a Seller Central listing when you ask and an Admin or Workspace owner confirms the preview. Prices, stock and advertising never change through the assistant.