How the Ads Optimiser decides when to REDUCE or INCREASE a bid

The Ads Optimiser analyses every keyword's advertising cost of sales (ACoS) against your product's margin and recommends whether to raise or lower the bid — or leave it alone. This article explains the logic behind those REDUCE and INCREASE recommendations so you can understand why a particular suggestion appears and decide whether to accept, adjust, or skip it.

Why it matters

Every keyword in your Amazon campaigns has a cost-per-click bid. If that bid is too high relative to what the keyword earns, you lose money. If it is too low on a keyword that converts well, you leave sales on the table. The Ads Optimiser watches this balance for every keyword across your campaigns and flags the ones that need attention — saving you from manually reviewing hundreds or thousands of bids.

How it works

The reference point — break-even ACoS

Everything starts with one number: your break-even ACoS, which equals your product's margin percentage.

Example: A product with a 30% margin has a break-even ACoS of 30%. At exactly 30% ACoS, every pound or dollar of ad spend is matched by the profit on the resulting sale — you break even. Below 30%, the keyword is profitable. Above 30%, it is loss-making.

This break-even figure is your target ACoS and the dividing line between REDUCE and INCREASE recommendations.

REDUCE logic — when ACoS is above target

When a keyword's ACoS exceeds your target, the Optimiser recommends lowering the bid. The size of the reduction is proportionate to the gap — a keyword that is slightly over target gets a gentle trim, while a keyword that is dramatically over target gets a steep cut.

ACoS relative to target	Approximate bid reduction
10–20% above target	~10%
20–50% above target	15–20%
50–100% above target	25–30%
More than 100% above target	40–50%, or pause

The principle is proportionate response: the bigger the gap between actual ACoS and target ACoS, the bigger the correction. A keyword with very high spend and very high ACoS receives the steepest cut because the Optimiser treats it as urgent.

Note: A keyword with high spend but zero orders is handled differently — it becomes a NEGATE recommendation rather than a REDUCE. See the related article on negate logic for details.

INCREASE logic — when ACoS is below target

Keywords performing below your target ACoS are profitable — and the Optimiser recommends raising their bids to capture more impression volume and, in turn, more orders.

ACoS range (below target)	Approximate bid increase
Under 5% ACoS (with orders)	~20%
5–10% ACoS (with orders)	~15%
10–20% ACoS (with orders)	~10%
20% up to break-even (with orders)	5–10%

The lower the ACoS, the more aggressively the Optimiser suggests scaling up — because the keyword has the most headroom before it approaches break-even.

Important: INCREASE recommendations only apply to keywords that have produced at least one order. A keyword with zero orders is never recommended for an increase, regardless of its ACoS, because there is no evidence it converts.

How the suggested new bid is calculated

The Optimiser reads the keyword's current bid directly from your Amazon account — the live bid value. It then applies the adjustment percentage:

INCREASE: New bid = current bid × (1 + adjustment percentage)
REDUCE: New bid = current bid × (1 − adjustment percentage)

Both values — the current bid and the suggested new bid — are displayed on the recommendation card so you can see the proposed change at a glance.

What the seller controls

The editable bid input

The suggested new bid appears in an editable input field. You are not locked into the Optimiser's number. You can type any bid value you prefer. When you click Accept, the system pushes whatever value is in that input field to Amazon — not the original suggestion. This means you benefit from the Optimiser's analysis while retaining full control over the final bid.

How rules take precedence

If you have configured Bid Rules for a campaign, the Optimiser does not generate REDUCE or INCREASE suggestions for keywords in that campaign. Instead, your rule's criteria and adjustment percentages run on the next daily processing cycle and produce rule-based suggestions (or apply directly if the rule is set to Automate mode). This separation prevents the rule and the engine from issuing conflicting bid adjustments on the same keyword.

What does not trigger a recommendation

Not every keyword receives a suggestion. The Optimiser skips keywords in three situations:

Low data — The keyword has too few clicks for the engine to draw a reliable conclusion. It waits until enough data has accumulated.
Active Bid Rule covers the campaign — The rule takes precedence, and the engine defers to it.
Already negated or paused — There is no active bid to adjust.

Why some recommendations may look counter-intuitive

A few patterns can seem surprising at first glance:

A low-ACoS keyword recommended for INCREASE. This is the Optimiser telling you that you have a winner and you are under-bidding relative to its profitability. Raising the bid can win more impressions and more orders while the keyword remains well within your margin.
A high-ACoS keyword receiving the steepest possible cut rather than a moderate one. When spend is high and ACoS is far above target, the Optimiser treats it as urgent and recommends an aggressive reduction to limit further losses quickly.
A keyword with high spend and no orders appearing as NEGATE instead of REDUCE. With no conversions at all, reducing the bid is unlikely to fix the problem — negating the keyword removes the spend entirely.

When to use this knowledge

Understanding the bid logic helps you in two ways. First, you can review recommendations faster because you know why each suggestion appears. Second, you can use the editable bid field more confidently — for example, accepting the direction of a recommendation but choosing a more conservative or aggressive number based on your own knowledge of the product, season, or competitive landscape.

For step-by-step instructions on reviewing and accepting bid recommendations, or on setting up Bid Rules that automate this process, see the related articles below.

Article #31
Article #34
Article #35
Article #37