I Started by Letting Auto Do the Job It's Actually Good At

The first mistake I stopped making was writing a keyword list before Amazon had told me anything. Automatic targeting isn't technically a keyword match type — it lets Amazon match your product through close match, loose match, substitutes, and complements, based on real shopper behaviour rather than my assumptions about it.

That makes it the right starting point whenever I'm launching a new SKU, entering a new marketplace, testing a category, or trying to learn the local search vocabulary in a market I don't have history in yet.

When I reviewed performance across the account, the pattern was consistent: automatic campaigns were surfacing sales that manually built keyword campaigns simply weren't reaching, because those manual campaigns were built on assumptions rather than evidence.

Then I Used Broad to Explore the Neighbourhood, Not the Whole Map

Once Auto surfaces a promising theme, the temptation is to jump straight to Exact. I stopped doing that too. Locking a single validated query into Exact immediately means capturing that one phrase while missing every reasonable variation around it.

That's the job I built for Broad — the least restrictive of Amazon's match types, designed to explore the space around a theme Auto has already validated, not to catch every keyword loosely connected to the category.

If Auto surfaces something like a "premium packaging" theme, Broad is where I go looking for the different ways shoppers phrase that same intent.

If I'm only watching impressions and spend without pulling actual search-term data back out of them, the campaign isn't doing its job. Every Broad campaign feeds the next decision — it doesn't exist just to spend budget.

Phrase Is Where I Stop Asking "Can This Convert?" and Start Asking "Does It Convert Reliably?"

Broad tells you a theme has demand. It doesn't tell you whether that demand is dependable. That's the distinction I built Phrase around.

Amazon positions Phrase as more restrictive than Broad, and I use it specifically once I start seeing the same commercially meaningful pattern show up more than once — not a single lucky conversion, but a repeated shape.

If Broad keeps surfacing variations around a specific product use case, Phrase is where I isolate that pattern and watch whether it holds up across multiple weeks, not just one good day.

That's a much higher bar, and it's the one that actually matters.

Exact Is a Destination I Make Keywords Earn

I stopped treating Exact as the default home for every keyword someone thought was important. Amazon calls it the most restrictive and precise match type, and I use it exactly that way — as the reward for keywords that have already proven themselves, not the starting point for new ones.

By the time I move something to Exact, I want to already know: shoppers are genuinely using this term, the product is genuinely relevant to it, the clicks have converted more than once, and the resulting order economics are ones I'm actually happy with.

The Step Most Accounts Skip: Actually Reading the Search Term Report

None of this works without someone regularly going through the Search Term Report. This was the single clearest lesson I took from reviewing the account's performance data — let automatic targeting gather evidence, then use that evidence to graduate strong performers into manual campaigns on purpose, rather than by accident.

I built a simple, repeatable review process around it:

I don't force every query through every rung. A search term that's clearly specific and converting reliably from day one can jump straight into Exact.

I Built a Negative List Alongside the Graduation List — Not After It

The other half of harvesting the Search Term Report is deciding what to stop paying for. If a search term keeps consuming budget without ever showing real relevance or commercial value, continuing to bid on it because it generates traffic is a losing habit.

Amazon lets you apply negative targeting inside both automatic and manual campaigns, and I treat every Search Term Report review as producing two lists, not one:

Skip the second list and Auto and Broad campaigns slowly turn into expensive, permanent research projects instead of temporary discovery layers.

Why the Ladder Matters More Once You're Live in Multiple European Markets

The biggest mistake I see in multi-market Amazon advertising is assuming a keyword structure that works in one country will translate cleanly into another.

Search behaviour is local. A shopper in France doesn't necessarily describe the same product the way a shopper in Germany, Italy, or Spain does — and literal translation doesn't reliably reproduce actual purchase intent.

When I reviewed performance across several European marketplaces, the differences between them were meaningful enough that I stopped trying to duplicate one country's winning structure into the next.

Instead of taking a validated French keyword, translating it, and launching it directly into Exact in Germany, I rebuild the ladder from scratch in every new market:

That gives every market its own keyword intelligence, built from its own shoppers, even when the underlying product is identical.

My account structure ends up looking like parallel ladders running side by side — France's Auto through Exact progression, Germany's, Italy's — each one earned independently rather than copied.

I Stopped Optimising Keywords Without Checking the Economics Behind Them

One more lesson changed how I promote keywords up the ladder: a keyword can convert and still be a bad keyword, if the product behind it can't economically support the cost of acquiring that click.

Lower-priced products simply can't carry the same acquisition cost as premium ones, so I stopped forcing one blanket efficiency target across an entire catalogue.

Before I move anything from Broad to Phrase, or Phrase to Exact, I now ask two questions in order:

The Campaign Architecture I Actually Build

For any product I take seriously, the structure I end up with looks like this:

Alongside all four, a running negative list keeps irrelevant or unprofitable searches from quietly draining the discovery budget.

That's a far easier system to manage than a dozen overlapping campaigns with no defined purpose — and it's a system that keeps feeding itself, month after month, market after market.

What I'd Tell Anyone Building an Amazon Strategy Across European Markets