One product. Three stages. Three different ad plans.

The three stages

The ranges below are a guide. Real timelines depend on the category, price and competition.

Launch: buy rank on purpose

Launch advertising is not normal advertising. The older funnel of auto, then broad, then phrase, then exact spends a lot of money learning what a new product could have been told up front.

Instead we start with research, grouping keywords by buyer intent and by how well they fit the product's real features. Then we build:

• Alpha campaigns. One exact keyword per campaign, on mid-tail terms the product can realistically own, such as "omega 3 for small dogs" rather than "omega 3". A small launch might use five to seven. Larger niches use many more.
• Semantic-core campaigns for groups of closely related terms.
• Bids set deliberately. We start slightly above the suggested bid and raise the top-of-search adjuster as needed, then wait for real orders (one to three) before changing a bid, because early data is noisy and attribution can keep updating for up to 14 days.
• Add-ons after a few weeks. Video, remarketing and product targeting follow once there is data to guide them. Auto campaigns come last.

Rank campaigns can run at a high ACoS, even 50% to 90%, because the question is not whether each click is profitable. It is whether the net loss is worth the rank gained. We track that trade-off openly.

Expand: build profit around what works

Once reviews and rank are established, the work shifts to profit and share. We move spend toward the terms and variations that convert, add video ads, product targeting and remarketing, and tighten structure so cost per sale falls. The target ACoS should move below the unit margin. Before that point, the product loses money on each ad sale.

A practical structure for this stage is campaign chaining: discovery campaigns find new terms and audiences, research campaigns test them, and performance campaigns collect the proven winners. Each campaign has one job, so its numbers mean something.

Harvest: hold the position cheaply

When the key terms sit at the top of page one, the aim is to keep that position at the lowest cost. We dial rank-building campaigns back to the minimum that holds rank, stop paying for top-of-search on terms already won, and shift budget to lean campaigns: retargeting, defending your own page and custom creative. Brand protection becomes a main job (see Pillar 4).

When things go wrong

A botched launch is recoverable. We negate clear losers, run only lean campaigns while data rebuilds, and then relaunch on the top-converting terms with single-keyword exact campaigns. In rare cases, such as a product that was out of stock for months, a fresh listing may be the cleaner restart. We will tell you plainly which case you are in.

Common mistakes

• Judging a launch campaign by profit, or a harvest campaign by rank.
• Keeping the launch setup running for years.
• Using broad and auto campaigns at launch with no negatives.
• Changing bids every day on a product with three orders.
• Lowering the price after launch instead of making the thumbnail more clickable.

What we do

• Classify every ASIN by stage and write down what "success" means for each.
• Rebuild or build campaign structure to match, with clear names and single purposes.
• Set the review rhythm: frequent early on, weekly or twice weekly in steady state.
• Re-classify as products move between stages.