Run five read-only Amazon Ads prompts in 30 minutes: last year by day, deal-price break-even, stock cover, budgets at 2x pace and sale end dates.
Less than four weeks out, most Black Friday plans are last year's plan with a bigger number on it. Same targets, same split, same hope that the campaigns that ran dry at 6 p.m. won't do it again. The stakes are real: US shoppers spent $11.8 billion online on Black Friday 2025 and $14.25 billion on Cyber Monday (Adobe).
Xnurta is an AI retail media platform for brands and agencies, and Xnurta MCP connects Claude, ChatGPT and other MCP clients to your Xnurta data. MCP (Model Context Protocol) is the open standard that lets an AI assistant connect to a tool like your ad platform. Five read-only prompts replace the guesswork. If AI near your budgets makes you nervous, start here: all five only read, and nothing in the account changes until you decide, later, with a preview.
Five questions, one decision each. The three read permissions (Performance data, Ads configuration and Operation logs) are all you need. Timings are illustrative, and a large account may take longer.
The 30 minutes: five prompts, five decisions (minutes are illustrative)
Minutes 0 to 6
Last year, day by day
Decision: the pacing shape for Friday, the weekend and Monday
Minutes 6 to 12
Break-even at the deal price
Decision: target ACOS per product group
Minutes 12 to 18
Stock cover at event pace
Decision: which ASINs to flag for a closer look
Minutes 18 to 24
Budgets at 2x pace
Decision: which budgets to raise before Thanksgiving
Minutes 24 to 30
End dates on sale settings
Decision: what switches back by itself, and what goes on the reset list
Last year's event shows the shape of this year's: when demand peaked, when it softened and which campaigns ran out of money too early.
Prompt 1: Last year, day by day
For ad account [X], show daily spend, ad sales, ACOS and total sales from the Monday before Thanksgiving to the Wednesday after Cyber Monday last year. Mark each day a campaign spent [95]%+ of its daily budget.
Xnurta holds about 15 months of daily history, so last year's event is in range by day. Look for the Friday peak, the weekend dip and the Cyber Monday spike. The decision: the pacing shape for each window, and the campaigns that need more headroom this year.
Break-even ACOS falls when you discount, so a target that made money in October can lose it on Black Friday. Break-even ACOS is the ACOS at which ad spend uses up all of a product's margin before ads. Below it, ads are profitable.
Prompt 2: Break-even at the deal price
Here is each ASIN's price, cost and fees at full price, and its planned Black Friday discount [paste]. Recalculate referral and fulfillment fees at the deal price (referral fees scale with price), include deal and coupon fees, and give break-even ACOS at the deal price. List every AI-managed group or campaign whose target ACOS is above it.
Illustrative product: full price vs 20% off
Full price
40%
Break-even ACOS
Price $100, referral fee (15%) $15, margin before ads $40
20% off
About 29%
Break-even ACOS, lower once deal fees are in
Price $80, referral fee (15%) $12, margin before ads $23
Judge each product against its own break-even, and watch TACOS: a falling TACOS means more sales arrive without ads. The decision: the event target ACOS for each group, so optimization chases a number that still earns money.
A product that sells out on Saturday wastes the clicks you buy for it on Sunday, so check cover before you commit budget.
Prompt 3: Stock cover at event pace
Using current inventory plus these inbound units and expected receive dates [paste], estimate days of cover for each advertised ASIN if sales run at [2.5]x the last 30 days from Thursday to Monday. Which run out before Cyber Monday?
Counting inbound units keeps the list honest. Xnurta doesn't hold your inventory plan, so you paste those numbers in. The decision: which ASINs to examine before you split the deal budget.
A common Black Friday miss is a winning campaign running out of budget in the evening, so test every budget against peak pace now.
Prompt 4: Budgets at 2x pace
Using the last 14 days of hourly spend, if spend runs at [2]x our normal rate from Thursday to Monday, which campaigns and AI-managed groups would hit their daily budget, and by roughly what time of day? Rank them by ROAS on data at least 14 days old.
Hourly data covers the most recent 14 days, which is the right window for this week's pace. Rank on settled data, because attributed sales take time to land. The decision: which budgets to raise before Thanksgiving, and which to watch hourly on the day.
Anything you raise for Black Friday should know when to stop, or November's spending carries into December.
Prompt 5: End dates on sale settings
List every budget, target ACOS, bid or status change made since [October 28] in ad account [X] for the event. Group them by teammate, Xnurta AI or rule, and say which sit inside an AI-managed group schedule with an end date and which don't.
Group schedules set target ACOS, daily budget and bid range per event window, and the group returns to its standing settings after the last window. Campaign-level raises don't, so list them now for the Promo Adjustment Reset Skill after the event. Amazon's budget rules and bidding rules for November 26 to 30 live in the console, so note their end dates there. The decision: which settings move into a group schedule, and which go on the reset list.
Weeks out from the event, the real risk is a fast change made on half the picture. This week the connection holds only the three read permissions, so no prompt can move a bid, budget or target. Rankings use data at least 14 days old, so nothing is judged on sales still reporting.
Turn the five decisions into changes later this week, approval-first. Write access is turned on per account by your CSM, and leads then grant Edit campaigns for the ad accounts they work on. Bid and budget changes show old and new values, wait for a yes, apply and read back. For campaigns in AI-managed groups, change the group's target ACOS, budget, bid range or schedule instead of hand-editing keyword bids. Amazon's dynamic bids, budget rules and bidding rules keep running per campaign through the event, and Amazon's Ads Agent works in the console, so keep them. Read-only vs read-write MCP explains each permission.
Book your Black Friday demo and our team will show you Xnurta and Xnurta MCP, and how these five answers come together in one conversation before Thanksgiving week starts on November 23. Xnurta MCP is included in Xnurta plans. Already on Xnurta? Create your connection with the three read permissions and run prompt 1 this morning. For more prompts, see the top 10 Xnurta MCP prompts.
Source: Adobe, Cyber Monday hits record $14.25 billion in online spending (December 2, 2025).
The connection itself. This week it holds only the three read permissions and Edit campaigns stays unticked, so no prompt can move a bid, budget or target, and permissions are checked on every request. When you do make changes, bids and budgets come back as old and new previews that wait for a yes, and group changes apply once you approve them in chat, so ask for the list first.
Amazon's server gives your agent API-level data and the agent builds each answer itself, which suits teams with their own agent. Here each prompt comes back as a finished table with the decision attached, so an agency with 12 clients runs 60 prompts instead of building 60 exports and 60 pivots (illustrative). Xnurta doesn't hold your costs, discount plan or inbound dates, so you paste those in.
Prompt 2. Discounting lowers every product's break-even ACOS, and a target set in October can lose money at the deal price. Prompt 5 comes next, because it makes sure sale-level spending ends with the sale, so December starts from your normal plan.
Divide your margin before ads at the deal price by the deal price, after recalculating referral and fulfillment fees at that price. In the illustrative example, a 20% discount drops break-even ACOS from 40% to about 29%.
No fluff. Just what's working.
