What is Bid Adjustment?

Bid adjustment is any change applied to the bid you set, whether you make it, your campaign settings make it, or Amazon makes it at auction time. The number stored on a keyword is a setting. The number that actually enters an auction is that setting after every adjustment layer has been applied.

The three layers

The layers compound, they do not add

A placement modifier multiplies the stored bid, and dynamic bidding then works on the already adjusted figure. Two adjustments that each look modest can produce an effective bid well above the one you typed.

Suppose, as a hypothetical, a keyword is set at 1.00 pound with a top of search modifier of 100 per cent. Auctions in that placement start from 2.00 pounds before Amazon's own dynamic adjustment does anything.

One property is worth stating plainly, because a great deal follows from it. Placement modifiers only ever raise a bid. Amazon accepts an uplift and nothing below the figure you set, so your base bid is the floor for every placement in the campaign. You cannot bid a placement down. You can only lift the others above it.

Dayparting behaves as a further multiplier in practice, although it is not a fourth auction layer. Amazon exposes no native hourly bid control for Sponsored Products, so an hourly schedule works by rewriting the stored bid on a cadence. The arithmetic still compounds with everything above it.

Adjust from cost per click, not from the bid

Your bid is a setting. Your cost per click is a fact about what the auction actually charged you. Because dynamic bidding and placement modifiers sit between the two, the gap can be wide, and it is widest in the days right after a bid change when trailing performance still reflects the old bid. Target maths should always be computed from realised cost per click.

Step discipline matters more than step size

Aggression is step size multiplied by frequency. A flat 25 per cent cut applied three times in one day leaves 42 per cent of the original bid, which is rarely what anyone intended. Shurq caps how far a single run may cut, sized against how often the automation runs, so repeated runs in a day cannot compound into a collapse. Raises are treated as nudges made once per data cycle, because a raise changes the very data that justified it.

Shurq also leaves keywords inside a narrow band around target alone. Restraint is part of the method, not an absence of one.

Where Shurq applies this

The bulk optimisation preview proposes stored bids and placement modifiers in the same run and holds every row in a review grid until you approve it. An applied run is recorded as a batch and can be reverted as a batch.