Amazon bids at high ACoS are best lowered dynamically and by rule: you tell the PPC-Butler MCP in plain language which target value applies, how much the bid may fall and where the lower limit is, and the MCP delivers the matching formula. If the ACoS of a keyword or target is then above your target value, the rule calculates the right bid from your own figures and moves the bid there in adapted steps.
Expensive traffic gets cheaper, while your keywords stay visible and keep selling, reliably the same way on every run.
What does “high ACoS” actually mean?
The ACoS (Advertising Cost of Sales) is the ratio of advertising spend to advertising revenue: ACoS = advertising spend ÷ advertising revenue. Whether a value is high is decided not by Amazon, but by your goal. Some companies want to be particularly profitable with their advertising, others pursue maximum sales at a break-even ACoS, and still others deliberately run unprofitable campaigns in order to gain market share.
The first step is therefore always a target value: per product, not for the whole account.
Why does a lower bid help?
The bid is the amount you bid for a click. Mathematically, the ACoS depends on three variables: the cost per click, the conversion rate and the order value. If the conversion rate and the order value remain the same, the ACoS falls along with the cost per click.
The downside: a lower bid can cost visibility. What matters is therefore not only that you lower, but by how much.
Why rigid bid jumps choke off keywords and sales
Many tools lower bids in fixed jumps: ACoS too high, so a big flat cut, regardless of whether a keyword is far above target or only just. Mathematical algorithms unfortunately often tend to lower bids too sharply. The keyword slips out of the visible placements, hardly gets any clicks and therefore no new data with which the bid could be corrected again. A keyword that sold too expensively becomes one that no longer sells at all. Your sales drop with it.
In PPC-Butler this cannot happen: your butler does not cut bids across the board, but according to your strategy and your specifications. You decide yourself how big a single step may be at most.
How to calculate the right bid
Instead of a fixed percentage, your rule can work with the keyword’s real figures. The starting point is the current cost per click: if the ACoS is above your target, the bid falls in the same proportion by which the ACoS is too high.
Target bid = CPC ÷ ACoS × target ACoS
A worked example with example values: with a CPC of €1.00, an ACoS of 40% and a target ACoS of 20%, the target bid is €0.50.
The conversion rate and the product price are already contained in the ACoS. If you want to stay profitable, use your own margin instead of the target ACoS and you get the bid at which the advertising just about pays for itself.
Other tools calculate this far too, and this is exactly where the difference lies: they set the bid from €1.00 to €0.50 in one go. Mathematically that is correct, in practice this jump chokes off the keyword. It loses visibility before you can see how it reacts to the lower bid.
In PPC-Butler the target bid is the direction, not the jump. You decide how the bid approaches the target, for example:
- only in steps of 5% downwards,
- and only when 5 new clicks without a new order have come in since the last bid change.
If an order comes in along the way, the bid stays where it is. The bid thus feels its way towards the target step by step, and after every step the keyword gets the chance to show how it reacts to the new bid. This is made possible by the Change KPI, which counts clicks and orders since the last bid change.
Using the MCP to get the right formula in plain language
You do not have to derive such formulas yourself. Via the PPC-Butler MCP you discuss with your AI assistant in plain language what should happen at high ACoS: which target value applies to which product? Should the bid fall cautiously or clearly? Where is the lower limit, and how long should the rule wait after a change?
From your strategy, the MCP delivers the matching conditions and the matching formula. These are suggestions, your butler does not apply them by itself. You check the suggestion, look at it in a test run if you wish, and set it up in your rule with just a few entries.
The rule: set in stone
Once the rule is set up, it is set in stone. The AI assistant helps with the wording, but it is the rule that runs: on every run with the same formula, the same thresholds and the same lower limit. Whether on the first run or the twentieth, across 20 runs it works reliably and the same way 20 times.
Only the keyword’s figures change. The formula reacts to them exactly as you discussed, and you can always trace why a bid was lowered.
Which four building blocks does the rule need?
- Enough data. An ACoS from three clicks is chance. A minimum number of clicks in the period under review ensures that the rule only judges keywords that it is able to assess.
- A calculated target bid. The new bid results from the cost per click, the ACoS and your target value or your margin, not from a flat percentage. A maximum per step leaves you the possibility of seeing the effect before the next step follows.
- A waiting period. After a bid change, new data is needed. A condition such as “since the last bid change, 5 new clicks without a new order have come in” ensures that a keyword is not bid down too quickly: the bid may only fall again once enough new clicks without a sale have come together. The 5 clicks are only an example, you set the value yourself. For this, PPC-Butler has the Change KPI.
- A lower limit. Minimum and maximum bids help to avoid extreme changes. For a lowering rule this means: never below your minimum bid, and never upwards.
How do you stay in control?
In PPC-Butler you see exactly which workflow was applied to which campaign and to which keyword before the bid is set on Amazon. No black box, but your rules, your figures, your decision. And if a rule has gone too far after all: one click and everything is undone.
Example rule: lowering the bid dynamically at high ACoS
- IF ACoS (last 30 days) > 20% AND clicks (last 30 days) ≥ 15 AND since the last bid change, 5 new clicks without a new order have come in → THEN new bid = CPC (last 30 days) ÷ ACoS (last 30 days) × 20%, at most 15% below the previous bid per run, lower limit: your minimum bid, an increase is excluded.
All figures in this rule are example values. You set the ACoS threshold, the number of clicks, the waiting period, the calculation and the largest step size yourself. Instead of the target ACoS, the rule can also calculate with your margin.