Performance rules are the part of Amazon ad automation that reacts to results. A schedule changes things at set times. A performance rule changes things when a number crosses a line: too much spend with no orders, an ACoS well above target, a keyword that converts far better than its bid suggests. Written well, they do the routine optimization work every week without anyone opening a report.
Written badly, they chase noise. This playbook gives eight rules worth copying, each in plain if-then form with a lookback window, a threshold and an action, and explains how to set the numbers for your own account. All numbers below are illustrative starting points, not benchmarks.
How to read a performance rule
Every rule has four parts. The scope says which campaigns, ad groups or targets it watches. The condition says which metric and which threshold. The lookback window says how many days of data it judges. The action says what changes when the condition is true.
The window matters more than most people expect. A target with ten clicks and no orders over three days tells you almost nothing. The same target with forty clicks and no orders over thirty days tells you a lot. How performance rules work covers the evaluation logic in more detail. Here we go straight to examples.
Rules that stop waste
Example 1: spend with no orders
If a keyword or product target has spent more than one and a half times your target cost per order over the last 30 days with zero orders, pause it. The threshold comes from your own economics: if you are willing to pay a set amount to win an order, a target that has spent well past that without one is unlikely to start converting.
Example 2: clicks with no orders
If a target has received more clicks than roughly twice the number your conversion rate says should produce an order, with no orders in 30 days, lower its bid by a fixed step rather than pausing. If your conversion rate is around 10 percent, that might be 20 clicks. This is gentler than Example 1 and suits targets that are relevant but expensive.
Example 3: ACoS far above target
If a target's ACoS over the last 14 days is more than one and a half times your target ACoS, and it has at least three orders, lower the bid by a small step. The minimum order count keeps the rule from reacting to a single expensive sale.
Rules that find growth
Example 4: strong converters held back
If a target's ACoS over 30 days is well below target, with at least five orders, raise the bid by a small step, up to a ceiling you set. Many accounts have keywords that win sales cheaply but lose impressions because the bid was set once at launch and never revisited.
Example 5: campaigns out of budget
If a campaign is below target ACoS and has run out of budget on most days in the last week, flag it for a budget increase. Running out of budget early in the day means missing the evening hours, which in many categories convert best. Running out of budget midday explains why this costs more than it looks.
Rules that protect the account
Example 6: daily spend spike
If a campaign's spend today is more than double its average daily spend over the last 14 days, alert someone. Spikes usually mean a bid was mistyped, a competitor left the auction, or a broad target started matching something irrelevant. A spend spike rule catches these within hours instead of at the end of the month.
Example 7: ACoS drift across the account
If account ACoS over the last 7 days is more than a set margin above the 30-day figure, alert the account owner. This does not change anything on its own. It tells a person to look before a slow drift becomes a bad month. ACoS creep detection covers how to tell drift from normal variation.
Example 8: low stock
If an advertised product's stock falls below its restock lead time in weeks of cover, lower its bids or budget by a step. Amazon stops serving ads for a product once it is out of stock, but that is too late: full ad pressure in the final weeks speeds up the stockout. Some tools check inventory directly; others need a person to watch it and adjust.
Setting the numbers for your account
Start from your margins, not from someone else's rules. Your break-even ACoS is your profit margin before ad spend. Your target ACoS sits below that, depending on whether the product is in launch, growth or harvest. Thresholds in the rules above are multiples of your target, so they move when your target moves.
Then check your traffic. A target that gets two clicks a week will never meet a 20-click minimum in 30 days, so the rule will never touch it. That is fine for waste rules. For growth rules on low-traffic products, lengthen the window to 60 days rather than lowering the minimum order count.
Finally, size the step. Bid changes of 10 to 20 percent per step are easier to reason about than large swings. A rule that halves a bid in one move tends to overcorrect, and the next rule then pushes it back up.
Common mistakes with performance rules
Rules that fight each other. A rule that raises bids on strong converters and a rule that lowers bids on high ACoS can both fire on the same target if their windows differ. Use the same window for paired rules, or make the conditions exclusive.
Windows too short. Attribution on Amazon settles over several days, so the last two or three days of data understate orders. A rule that judges three days of data will see high ACoS that is not real.
No log. If a rule changes bids and nobody can see what it did, nobody will trust it, and the first bad week will get it switched off. A change log is what lets you keep a rule running long enough to learn whether it works.
Frequently asked questions
What is a performance rule in Amazon Ads?
A performance rule watches a metric such as ACoS, spend, clicks or orders over a lookback window and takes an action when a threshold is crossed. Typical actions are pausing a target, lowering or raising a bid, or flagging something for review. It reacts to results, not to the clock.
What lookback window should a performance rule use?
Long enough that the metric is not noise. For most accounts that means 14 to 30 days for ACoS and order-based rules, and a shorter window only for spend-based rules that protect against runaway costs. Low-traffic targets need longer windows than busy ones.
How many performance rules should I start with?
Two or three. Start with a rule that catches spend with no orders and a rule that catches ACoS far above target. Watch what they do for two weeks in the change log before adding more, so you know each rule behaves the way you expect.
Off Hours runs performance rules like these on a 15-minute cadence and logs every change it makes. Start a free 14-day trial.