Amazon PPC bid optimisation

What a click is worth

Start with the profit on one sale before advertising: price, less product cost, Amazon fees and shipping. Multiply that by the share of clicks that become orders for a keyword, and you have roughly the most a click can cost before the sale earns nothing. If you want each sale to keep some profit after advertising, bid below that figure.

This break-even calculation is more useful than any general target, because it reflects your own product. It also shows why the same bid can be sensible for one product and loss-making for another.

A worked calculation

Waiting for enough data

A keyword with a few clicks and no sale has not shown anything yet, and one early sale does not prove a keyword works. Decide how many clicks a target needs before its bid is changed, based on how often your clicks usually convert. Products with low conversion rates need more clicks before a judgement is safe.

Bidding strategies

Amazon's Sponsored Products guide names dynamic bidding up and down and a dynamic down only strategy. These let Amazon raise or lower bids, within limits it describes, depending on how likely a click is to convert. They work alongside your own bid, not instead of it, so the base bid still needs to be sensible. Read Amazon's current documentation for how each strategy behaves before choosing one, and change one thing at a time so you can see its effect.

A bid review routine

1. 01

   Export

   Targets with clicks, spend and sales.

   Output: Target list

2. 02

   Filter

   Only targets past the data threshold.

   Output: Targets ready to judge

3. 03

   Compare

   Cost per sale against break-even.

   Output: Raise, hold or lower

4. 04

   Adjust

   Modest changes, one at a time.

   Output: New bids

5. 05

   Log

   Date, old bid, new bid, reason.

   Output: Change log

Example: adjusting bids on a phone stand

Before changing a bid

How large a change

Large swings make it hard to tell what caused a change in results. Smaller, regular adjustments let you see the effect and correct course. A big cut can be justified where a target is clearly losing money, but most adjustments can be gradual.

How often to review

Reviewing too often means reacting to noise; reviewing too rarely lets losses run. A weekly or fortnightly rhythm suits many accounts, with longer gaps for low-volume products. Allow time after a change for results to settle before changing the same bid again.

When bids are not the problem

A keyword that gets clicks but few sales may be suffering from the listing, the price or the reviews rather than the bid. Lowering the bid reduces the loss, but fixing the listing may turn a weak keyword into a profitable one.