Amazon PPC campaign structure: the auto-to-manual keyword flow
The most expensive mistake in Amazon advertising is launching many campaigns before you know what works. A sound start rests on one logic: first discover cheaply which searches convert, then focus budget on the winners. This guide builds that auto-to-manual flow step by step.
How Amazon PPC actually works
Amazon ads run on cost per click: impressions are free, you pay only when someone clicks. What decides who shows is your bid and relevance — a high bid alone is not enough; the ad and the listing must fit the search.
That two-sided mechanism has a consequence: advertising is expensive on a weak listing. A page that does not convert produces both low conversion and weak relevance signals, which pushes your cost per click up.
What must be ready before you start
- The listing converts: main image, price, bullets and ideally a few reviews
- Stock is sufficient: advertising into a stockout wastes both money and momentum
- You know your break-even ACOS: without a threshold from contribution margin you cannot call anything good or bad
- You hold the Buy Box: otherwise clicks largely benefit a rival
Why start with an auto campaign?
An auto campaign puts Amazon’s own matching engine to work for you: you learn which searches relate to your product without guessing. That is both cheaper and more accurate than building a keyword list from assumption.
In the first weeks the goal is not profit but data: which real searches get clicks, and which convert. Building manual campaigns without that data is shooting in the dark.
The auto-to-manual flow
A simple, manageable starting structure:
| Campaign | Purpose | Match | Budget share (example) |
|---|---|---|---|
| Auto — discovery | Learn which searches convert | Automatic | ~50% at the start |
| Manual — exact | Proven winning keywords | Exact | ~30%, grows with winners |
| Manual — phrase/broad | Explore variations of winners | Phrase/broad | ~20% |
The flow runs: auto discovers → the search term report reveals winners → winners move to manual exact → negate those terms in auto so you stop competing with yourself. Budget then shifts steadily toward proven keywords.
Your first four weeks
| Week | What you do | What you ignore |
|---|---|---|
| 1 | Launch the auto campaign; leave it alone | Daily sales noise |
| 2 | Let data build; negate only obvious waste | Drawing conclusions from ACOS |
| 3 | Read the search term report; split winners and losers | Single-click terms |
| 4 | Move winners to manual exact; adjust bids | Multiple changes in one week |
The most common error is intervening in weeks one and two. Optimising before meaningful clicks accumulate is simply reacting to noise.
Setting budget and bids
Budget: treat the daily budget as learning money you can afford to lose. The first objective is accumulating enough clicks to produce meaningful data.
Bids: start in the lower-middle of Amazon’s suggested range and raise gradually if impressions do not come. Avoid large jumps — they burn budget fast and corrupt learning. Apply the one-variable rule to bid changes too.
What to watch, and what to ignore
Watch: click volume (is there enough data), the search term report, converting terms, ACOS relative to break-even, and the TACOS trend.
Ignore early on: daily ACOS swings, performance of single-click terms, hourly data. These are statistically meaningless and push you into needless intervention.
Common beginner mistakes
- Advertising before the listing is ready
- Opening 5-10 campaigns at once (unmanageable, teaches nothing)
- Changing bids in the first 48 hours
- Judging ACOS without knowing break-even
- Never using negative keywords, so leakage continues
- Moving a winner to manual without negating it in auto
- Leaving campaigns running after a stockout
FAQ
How much budget should I start with?
There is no single right number: enough daily budget to accumulate meaningful clicks and that you can afford to lose. Early on the goal is learning, not profit.
How many campaigns should I start with?
One auto campaign is enough. Add manual exact campaigns as winners emerge. Starting with many campaigns splits budget and complicates management.
When should I make the first optimisation?
Wait until meaningful clicks accumulate — usually one to two weeks. Earlier intervention is a reaction to noise.
ACOS looks high — should I pause?
First compare it with your break-even ACOS. If it is below, there is no problem. If above, cut the leakage rather than pausing: negate non-converting terms and focus on winners.
Should I turn the auto campaign off later?
Usually not. Auto remains a discovery channel that keeps catching new search behaviour. Negate winners there as you move them into manual.
Do negative keywords reduce sales?
Done properly, no: you only stop terms that spend without selling. Be careful with broad negation — work term by term.
Are ads necessary for a new product?
On a new listing organic visibility is low, so ads usually provide the first traffic. But if the page does not convert, ads only accelerate the loss.
Know your margin before setting a target
Break-even ACOS comes from contribution margin. Setting an ad target without it is navigating without a compass.
Sources