Amazon Ads prompts for Claude and ChatGPT, ready to copy

What can you ask your Amazon Ads data with Xnurta MCP?

You can ask for performance, settings and change history in plain language, and for some changes too. Xnurta MCP connects Claude, ChatGPT and other MCP clients to your Xnurta data, so Xnurta MCP answers from your ad accounts instead of an export you pulled last Friday. Copy a prompt, replace the [placeholders], paste. If you have not connected yet, Connect Claude or ChatGPT to Xnurta MCP takes a few steps.

The twelve prompts below sit in four groups, three each. The table shows what each group does and what it needs.

Which of these prompts change your account?

Five of the twelve do, and the other seven only read. Each prompt is labeled. Queries on ad performance, entity settings and the operation log never modify your account, so the seven read-only prompts never change anything.

Writing comes in three kinds, each with its own permission scope: managed group management, ad-entity editing and SP campaign creation. The five write prompts here use the first two. None creates a campaign.

Anything involving money, bids or archiving returns a preview first, showing which objects change and from which value to which value. Nothing is submitted until you confirm, and if you decline, nothing changes. The write prompts below also tell the assistant to list the changes and wait, so the habit holds for every edit. Archiving cannot be undone, so pause a campaign if you only want it to stop.

What makes an Amazon Ads prompt work?

Four things: the ad account, a time window, a comparison and the shape of the answer. Leave one out and the assistant guesses. Name the ad account in the prompt, and if you are unsure what you have, start by asking it to list the ad accounts you are authorized to access. You see only the accounts your own Xnurta login carries, main accounts and sub-accounts included.

Ask follow-ups in the same chat. Start broad, then ask why, then drill into one object. That beats one giant question. To check your connection, ask the assistant to call get_user_authorized_context, which returns your user ID and the profile IDs you can reach.

Which prompts cover reporting and diagnosis?

Three read-only prompts take you from a trend line to a single ASIN. The first puts TACOS beside ad sales and total sales for a product line. The second compares one weekday with the last four. The third diagnoses a child ASIN year to date.

Prompt 3 names the Product Diagnosis skill. It produces ASIN and parent ASIN ranking tiers, declining-product detection, variant comparison, ad eligibility and inventory checks, and related advertising changes. Skills teach the assistant a routine and you install them in a separate step after connecting, so if the skill is missing, ask the assistant to install it from the Xnurta MCP repository.

Show the TACOS trend for product line [product line name] in ad account [account name] over the last 8 weeks, week by week. Put ad sales and total sales next to it. Tell me which weeks moved most and the likely reason. Return a table, then three sentences.

It is [Wednesday] today. Analyze ad account [account name] and compare it against the last four [Wednesdays]. What changed, and why did today's performance drop? Return a table by day, then the two most likely causes.

Use the Product Diagnosis skill on child ASIN [B0XXXXXXXX] in ad account [account name], year to date. Check the campaign structure, the traffic mix and any potential issues. Rank the issues by impact and make no changes.

Which prompts find wasted spend and new keywords?

Harvest the winners first, then deal with the spenders that never convert. Prompt 4 reads your automatic targeting for search terms worth moving into manual campaigns. Prompts 5 and 6 change your account, and both stop for your confirmation.

Prompt 5 lists campaigns that spent with zero orders before it touches anything. You pick, it shows the final list, then it pauses. Pause is the right verb. Archiving is irreversible. Prompt 6 adds negative keywords, which Xnurta MCP supports for SP campaigns, and it works with the ad-entity editing scope rather than campaign creation.

In ad account [account name], which search terms from campaign [campaign name]'s automatic targeting over the last [30] days have CVR above [X]% and more than [two] orders? Which should I add to manual campaigns? Return a table with term, CVR, orders and ACOS. List only.

In ad account [account name], list the campaigns that had spend but zero orders in the last [30] days, biggest spend first, with spend and clicks. Change nothing yet. When I tell you which to pause, show me the final list and wait for my confirmation before you pause anything.

In ad account [account name], add these negative keywords to SP campaign [campaign name] as [match type]: [list of terms]. Show me each term and the campaign it goes to, and wait for my confirmation before you submit.

Which prompts handle bids and budgets?

One reads hourly spend, one raises budgets on capped campaigns, one lowers keyword bids. The hourly prompt works because performance data is daily by default but hourly for SP, SB and SD campaigns and SP keywords. Keep an hourly window to about one week.

Prompts 8 and 9 touch money, so each returns a preview with old and new values before anything is submitted. Check the list before you confirm.

Show hourly spend for campaign [campaign name] in ad account [account name] over the last 7 days, and the hour it usually runs out of budget. Return a table by hour, then one sentence on whether the daily budget is limiting it.

In ad account [account name], which campaigns hit their daily budget cap in the last 7 days? Then raise each one's daily budget by [10]%. Show the current and new budget for every campaign first, and wait for my confirmation.

In ad group [ad group name], lower the bids by [15]% on these high-ACOS keywords: [keyword list]. Show each keyword's current and new bid, and wait for my confirmation before you submit.

Which prompts steer AI managed groups and audit changes?

Find what is outside AI management, create a group with AI switched off, then read what the AI did. Prompt 10 finds campaigns that spent but are not yet in any managed group, grouped by how close their ACOS values are. Prompt 11 builds managed groups from those clusters and leaves AI off, so you can review before the AI acts. Prompt 12 reads the operation log for one group.

The operation log holds human and AI actions on campaigns, ad groups, targets, bids, budgets and managed groups. Through MCP you can write a managed group's name, AI status, personality, goal, target ACOS, total budget, campaign membership, action space on or off, schedule and template. Brand, non-brand and competitor mode, and all word-list features, stay in the Xnurta platform. Deleting a managed group archives it for good, so none of these prompts does that. For the levers themselves, read Read and change Xnurta's AI levers through MCP. For one campaign's history, read Who changed my Amazon campaign?.

In ad account [account name], find campaigns that had spend in the last 30 days but are not yet in any managed group. Group them by how close their ACOS values are. Return a table with campaign, spend, ACOS and proposed group.

Create managed groups from these groups of SP campaigns with similar ACOS: [group 1 campaigns], [group 2 campaigns]. Set target ACOS to [6]% and leave AI switched off for now. List each group's name, target ACOS, AI status and campaigns, and wait for my confirmation before you create anything.

In ad account [account name], list every action AI took on managed group [group name] over the last [14] days, by action type. Count the bid increases, bid decreases, budget increases and budget decreases. Return a table, newest first.

Where do you get the full set of 50 prompts?

The full set of 50 prompts is available as a download, which suits a set this size better than a long page.

Two habits turn prompts into a routine. Save the ones you reuse as templates, or turn one into a weekly report routine that runs every Monday. For a first set built around Monday morning, see Ten questions to ask your Amazon Ads data every Monday. For ten everyday prompts, see Top 10 Xnurta MCP prompts and the data each needs. To change many campaigns in one request, see Edit Amazon campaigns in bulk through an AI assistant.