How-to guide

How to Bulk Edit Amazon Ads Campaigns With an AI Assistant

TL;DR

To edit Amazon Ads campaigns in bulk through an AI assistant, connect Claude or ChatGPT to the ad account with a read-write Amazon Ads MCP, describe which campaigns to change by how they perform, and approve an old to new preview before anything is applied.

  • A weekly spreadsheet job becomes one reviewed request, and every budget change shows its total before anyone confirms.
  • Find campaigns by what they are doing in a read-only step, then write each campaign's exact new value.
  • End every bulk prompt with: show me the old to new preview before applying.
  • Pause, don't archive: Amazon says archived campaigns cannot be reinstated.
  • Bulksheets still win for full rebuilds, very long upload files and fields your MCP does not cover.

To edit Amazon Ads campaigns in bulk through an AI assistant, connect Claude or ChatGPT to the ad account with a read-write Amazon Ads MCP, describe which campaigns to change by how they perform, and approve an old to new preview listing every campaign before anything is applied. MCP (Model Context Protocol) is the open standard Anthropic introduced on November 25, 2024 for connecting AI assistants to outside tools. The edits are worth checking line by line: since January 9, 2024, Amazon has allowed placement adjustments of up to 900% on each Sponsored Products placement, so one wrong number in a batch can multiply a bid many times over.

What can you edit in bulk through an AI assistant?

An AI assistant can edit whatever campaign settings its Amazon Ads MCP exposes for writing, and coverage differs by MCP and by ad type. These are the edits most teams make in bulk each week:

EditTypical bulk useCheck before confirming
Status: enable, pause, archivePausing a portfolio for a stock-outThat nothing is archived by mistake
Daily budgetTopping up capped, efficient campaignsThe total change against the monthly plan
Keyword and product target bidsLowering bids on high-ACOS targetsThe floor and the size of each step
Placement adjustmentsRaising top of search on exact campaignsThe maximum possible bid per placement
Keywords, negatives, product adsAdding harvested terms and their source negativesMatch type, and that no brand negative lands in a brand campaign

Creating campaigns is a separate permission and a separate workflow, covered in how to create an Amazon campaign from a brief.

How does a bulk edit work from prompt to result?

A bulk edit through an Amazon Ads MCP follows five steps. The AI reads, selects and reads back, and nothing applies until you confirm the preview.

  1. Read. The assistant reads current settings and performance. Name the ad account in the prompt.
  2. Select. It applies your rule to find matching campaigns or targets. Set thresholds and exclusions.
  3. Preview. It lists every object with old and new values. Read it, and edit or drop rows.
  4. Apply. It submits the change after your yes. Confirm only what you checked.
  5. Read back. It reads the account again and lists anything that did not apply. Compare it with the preview.

Because the assistant reads live values, the preview reflects today's settings, not last week's export. Keep steps 2 and 3 apart: let the assistant find matches read-only, then write a prompt that states each campaign's exact new value, so the preview checks numbers you chose.

How do you select campaigns by performance?

Selecting by performance means writing the rule into the prompt, with a minimum amount of data so the assistant does not act on noise. A good rule has four parts: a metric and threshold (ACOS above [35]%), a minimum amount of data (at least [20] clicks), a window (14 days ending [3] days ago, because recent days understate attributed sales) and exclusions (skip brand campaigns and AI-managed groups).

Amazon's targeting guide recommends judging a keyword after it has received at least 20 clicks before negating it, and the same minimum is a sensible floor for bid decisions. Judge ACOS against each product's break-even ACOS, set from margin, as shown in the Amazon Ads prompts library.

How do you change budgets across many campaigns?

Budget changes across many campaigns work well as two prompts: a read-only one that finds the capped, efficient campaigns, and a write prompt that names each campaign with its exact new budget. Ask for a total row, because it shows what the change does to the month.

Prompt: find capped, efficient campaigns (read-only)

For account [X], list Sponsored Products campaigns that ran out of budget on at least [4] of the last 7 days with 14-day ACOS below [25]%, with their current daily budget. Don't change anything.

Prompt: set new budgets with a total row

In account [X], set daily budgets: [Product A]_EXACT to $[72], [Product B]_EXACT to $[108], [Product C]_BRAND to $[150]. Show me the old to new preview with a total row before applying.

A budget is a monthly figure in disguise. Amazon says a Sponsored Products campaign can spend up to 25% more than its average daily budget on any given day, averaged over the calendar month, so a $[50] daily increase in the total row is roughly $[1,500] a month.

How do you change bids in bulk without overcorrecting?

Bid changes in bulk stay safe when each step is small, each bid has a floor, and only targets with enough clicks are touched. Moving bids [15] to [20]% at a time and waiting several days before the next step is a common habit. Work out each new bid yourself, then send the values.

Prompt: find high-ACOS targets (read-only)

For account [X], list keywords and product targets with at least [20] clicks in the 14 days ending 3 days ago and ACOS above [40]%, with current bids. Skip anything in an AI-managed group. Don't change anything.

Prompt: set new bids with a floor

In account [X], set bids: [keyword 1] (exact) to $[1.36], [keyword 2] (phrase) to $[0.94], [ASIN C1] to $[0.35]. Show me the old to new preview before applying.

Here each new bid is about [15]% lower, with a $[0.35] floor. Keep campaigns inside an AI-managed group out of manual bid edits: the group's AI adjusts those bids toward its own target, so a manual edit pulls against it.

How do you pause or enable campaigns in bulk, and why not archive?

Pausing or enabling campaigns in bulk takes one prompt naming the set and the new status, and pausing is the right choice whenever you might want a campaign back. Amazon's bulk operations guide states that archived campaigns cannot be reinstated.

Pause all Sponsored Products campaigns in portfolio [Q4 Promo] in account [X]. List each with current status, daily budget and 7-day spend, and show me the old to new preview before applying.

A paused campaign keeps its history and can be enabled again in one step. Archive only campaigns that have been paused for weeks and that you are sure you will not need.

How do you set placement adjustments in bulk?

Placement adjustments in bulk are set by naming each campaign, the placement (top of search, product pages or rest of search) and the percentage, including 0% to remove an adjustment. Xnurta MCP sets placement adjustments on Sponsored Products and Sponsored Brands.

Prompt: placement adjustments with the maximum bid

For exact-match Sponsored Products campaigns in account [X], show 30-day ACOS and conversion rate by placement. Then set top of search to +[25]% on [Campaign A] and [Campaign B], and product pages to 0% on [Campaign C]. Show the maximum possible bid per placement, and show me the old to new preview before applying.

Placement adjustments multiply the bid: +[25]% on a $[1.40] bid means up to $[1.75] at top of search, before any dynamic raise. Per Amazon's dynamic bidding guide, the up and down strategy can raise or lower a bid by up to 100% for all placements, so check each campaign's current strategy in a read-only step. Setting product pages to 0% only removes the uplift; if they still lose money, lower the base bid.

How do you add keywords, negatives and product ads in bulk?

Keywords, negatives and product ads are added in bulk by listing each term or ASIN with its campaign, match type and bid, mostly in Sponsored Products. Xnurta MCP can add and edit all three, with the most options for Sponsored Products.

Prompt: add keywords and source negatives

In account [X], add [term 1] and [term 2] as exact keywords at $[1.20] in [Product A]_EXACT, and add the same two terms as negative exact in [Product A]_AUTO. Add [your brand] as negative phrase in [Product A]_BROAD. Show me the old to new preview before applying.

Four habits keep this clean. Negate a harvested term at its source, so the exact campaign gets the traffic and the auto campaign stops splitting data. Add brand terms as negative phrase in non-brand campaigns, so one negative blocks every search containing the brand. Never add a brand negative to a brand campaign, because it would switch that campaign off. Check stock before adding a product ad, because Amazon stops serving ads for products that cannot be bought. Which terms to harvest or block is a separate job, covered in the Amazon Ads prompts library.

How does an AI assistant compare with Amazon bulksheets?

An AI assistant and Amazon bulksheets both change many Amazon Ads campaigns at once, but they differ in how you pick the rows and how you check the work.

Amazon bulksheetsAI assistant with an Amazon Ads MCP
Choosing what to changeFilter and sort a downloaded fileDescribe a rule in plain words
Data it works fromThe snapshot you downloadedLive values read at request time
Checking the changeYour formulas and a second reviewerOld to new preview, then read-back
ErrorsUpload report; fix rows and re-uploadRead-back lists what did not apply
CoverageSP, SB and SD fields, including creationWhatever the MCP exposes; varies by ad type

Bulksheets, described in Amazon's bulk operations user guide, remain the better tool for full account rebuilds, very long upload files your team already keeps in spreadsheets, and fields your MCP does not cover. The assistant fits the everyday case: a performance rule applied to whichever campaigns meet it this week. Many teams use both.

When should you change an AI-managed group instead of editing campaigns?

An AI-managed group change beats a campaign edit when the edit expresses a lasting goal rather than a one-off fix. If you lower bids in the same category every Monday, the real decision is a lower target ACOS or a narrower bid range for that group. Pausing a portfolio for a stock-out or resetting budgets after a promotion are campaign edits. A category ACOS that keeps drifting above goal is a group target. A sales event window is a date-range schedule on the group, with budget raised separately.

Each AI-managed group also has an AI on/off switch and a per-lever AI personality from 1 (very conservative) to 5, both group settings in the Xnurta platform. When a date-range schedule ends, the group returns to its normal target, goal and levers by itself. See AI-managed group levers and personality for how to set those limits.

What does a safe bulk edit checklist look like?

A safe bulk edit checklist covers scope, size, preview, timing and review, so the same steps run whoever on the team makes the change. Copy this into your team's SOP:

  1. Scope the connection to the ad accounts this person manages, and give write access only to people allowed to change live accounts.
  2. Name the ad account in every prompt, so an edit cannot land in the wrong marketplace.
  3. Set thresholds and caps: minimum clicks, a maximum budget, a bid floor and ceiling.
  4. Exclude what should not move: brand campaigns, AI-managed groups, live promotions.
  5. Ask for the preview with totals: object count, total budget change, largest single change.
  6. Read previews under [30] rows line by line, and split larger jobs into batches.
  7. Pause, don't archive.
  8. Read back and log. Check the new values, then the operation log for the edit's time window; each entry holds the old and new value.
  9. Review the effect after 14 days, once attributed sales have settled, with a read-only before and after prompt.

For agencies, run one MCP connection per client ad account, keep juniors on read permissions, and let a senior account owner hold write permission per client. The audit side is in who changed my Amazon campaign, and the permission split is in read-only vs read-write MCP.

How does Xnurta MCP handle bulk campaign edits?

Xnurta MCP handles campaign edits through the Edit campaigns write permission, granted separately from read access and from campaign creation. Edit coverage varies by ad type, with the most options for Sponsored Products, including keywords, negatives and product ads. Generally available from 27 Oct for Sponsored Ads. Write access is turned on per account by your CSM. To keep one client's account free of AI changes, give its connections read permissions only.

Changes to money, bids or archiving show an old to new preview enforced by the server, so the preview does not depend on how the prompt is worded. They run only after you confirm, and the result is then read back. Each connection can be scoped to selected ad accounts, never exceeds its creator's own Xnurta permissions, and expires after 30, 90 or 180 days or 1 year. The operation log labels every change as human, AI, rule, Amazon console, API connection or a named teammate, with old and new values.

Try this first. On any Amazon Ads MCP, pick the bulk edit your team makes most often, usually budget top-ups, and run it with read permissions only: Which campaigns would this rule select, and what would you change? Don't apply anything. Compare the list with what you would have built in a bulksheet.

Xnurta MCP is for Xnurta customers. Learn more on the Xnurta MCP page, or book a demo to see a live old to new preview, the read-back after a change, and the change log labeled by who made it.

Next question: How do AI-managed group levers and personality work?

Video transcript

Sources & references

FAQ

Frequently asked questions

Can ChatGPT or Claude edit many Amazon campaigns at once?

Yes, through an Amazon Ads MCP connection that holds write permission for campaign edits. You find campaigns by a rule, such as ACOS above a threshold, then state each one's new value, and the assistant lists old and new values. Nothing should apply until you confirm the preview, and the assistant should read the result back.

Can I bulk edit Sponsored Brands and Display campaigns too?

Partly. Edit coverage varies by ad type, with the most options for Sponsored Products, and placement adjustments work on Sponsored Products and Sponsored Brands. For fields your MCP does not expose on Sponsored Brands or Sponsored Display, use Amazon bulksheets or the console. Check what is covered in a read-only step before writing the change.

How do I stop an AI from making a wrong bulk change?

Read the preview before you confirm, and decline it if any row looks wrong. Put a cap, a floor and exclusions in every prompt, and keep previews short enough to read line by line. Scope the connection to the right ad accounts, and revoke it if it is no longer needed.

How many campaigns should one bulk edit include?

Few enough to read every row of the preview, about 30 rows. Split larger jobs by portfolio, and keep one ad account per request so a change cannot land in the wrong marketplace. A short batch is easier to check before you confirm and easier to fix if the read-back shows a row that did not apply.

Can I undo a bulk edit?

Mostly, by editing back. Every change except archiving can be reversed: the preview and the operation log hold each object's old value, so ask the assistant to set those values again and show the old to new preview. A paused campaign can be enabled in one step, but Amazon says archived campaigns cannot be reinstated.

How long should I wait to judge a bulk edit?

Judge a bulk edit on windows that ended at least 14 days ago, because attributed sales keep arriving for up to two weeks. Compare equal windows before and after the edit, leave out the change day, and check a set of campaigns you did not touch, so a slow week for everyone is not credited to your edit.

Xnurta MCP

Ask your Amazon Ads account a question today.

Xnurta MCP is for Xnurta customers only. Create your MCP connection directly in the Xnurta platform. New to Xnurta? Book a demo and we'll walk through your account with you.