Amazon PPC guide for sellers moving past auto campaigns

Group products that share a buyer intent, conversion behavior, margin target, and inventory situation. Avoid placing unrelated products into one ad group because a successful search for one item may be irrelevant to another. Keep the structure compact enough that each campaign receives useful evidence.

Use automatic campaigns to discover queries and product relationships. Within automatic targeting, inspect close match, loose match, substitutes, and complements separately. Their relevance and conversion rates can differ. A single successful group should not conceal waste in another group.

Separate brand defense from generic acquisition when their economics or goals differ. Create a distinct campaign when you need a separate budget, not merely to create a tidy label. Use portfolios or naming conventions to identify product family, marketplace, role, and target metric.

Reserve an explicit discovery allowance and review its output. For an operating starting point, a seller might cap discovery at 20% of an approved product-group budget. That allocation is a choice to test, not a recommended universal ratio. Mature products and new launches can justify different discovery funding.

Export the Sponsored Products search term report with campaign, ad group, targeting, customer search term, clicks, spend, orders, and sales. Keep the original target separate from the customer search term. A broad keyword or automatic target can trigger many distinct queries, each with different intent.

Check the reporting attribution window used for your account and advertiser type. Recent clicks can receive attributed sales later. Assess efficiency using clicks old enough for the relevant window to settle, then keep recent spend visible as provisional exposure. Avoid applying a fixed overnight rule to all products.

• Confirm the product was available and eligible during the review period.
• Check the listing, price, shipping promise, and featured offer status.
• Review returns and cancellations when evaluating product economics.
• Aggregate identical queries within comparable product groups.
• Keep marketplace currencies separate unless you document conversion.

Set operating thresholds before reviewing winners and losers. For example, require at least three mature attributed orders and ACoS at or below the product target before promoting a term. This is a review threshold, not statistical proof. A term with one sale can remain under observation while a clearly irrelevant query can be excluded without waiting for more clicks.

Harvesting means creating a manual keyword from an observed customer query. Start with relevance. The product must satisfy the query, including material, size, use case, and compatibility. Then check mature orders, ACoS, and whether the term already exists in a suitable destination campaign.

Add the qualifying phrase as an exact keyword in the manual campaign for the relevant product group. Verify advertised products, inventory, budget, and eligibility before changing the source. Exact targeting provides tighter control, but matching can include close variations. Continue reviewing the actual search terms after transfer.

Estimate an affordable CPC from economics. Maximum CPC = average order value x target ACoS as a decimal x expected order conversion rate. If order value is $40, target ACoS is 20%, and conversion rate is 10%, the arithmetic gives 80 cents. Use a mature, relevant conversion rate and account for placement modifiers before selecting a base bid.

• Record the source campaign and harvested query.
• Check for an existing equivalent exact target.
• Select the destination product group and starting bid.
• Confirm the manual target receives eligible impressions.
• Decide whether source routing needs an exact negative.
• Review combined source and destination results after attribution settles.

Do not automatically raise bids because the new exact keyword receives few impressions. Check eligibility, relevance, competition, budget, and placement settings first. The Amazon PPC software workflow supports search term harvesting and recurring rule execution on Agency and higher plans.

Use negative exact when a specific query is unsuitable or when routing a harvested query away from discovery. Use negative phrase when the phrase consistently expresses an intent that the product cannot satisfy. Review the platform's matching behavior, including applicable close variations, and inspect results after applying either type.

For a ceramic mug, a query for replacement lids may be irrelevant if no lid fits or ships with the product. A phrase exclusion for replacement lids can be reasonable after checking the offer. A phrase exclusion for gift is much riskier because it can block valuable gift intent. Judge the entire affected query set before broadening the exclusion.

Source routing requires care. Adding a keyword in manual exact does not guarantee delivery. Apply a source exact negative only after confirming the destination is eligible, funded, and receiving traffic. Otherwise, the change can remove an effective path without creating a working replacement. You can also allow overlap temporarily while monitoring combined economics.

A useful cost trigger for review is spend above twice the allowable acquisition cost with no mature orders. With an $8 allowance per order, that trigger is $16. Inspect relevance and listing readiness before choosing an exclusion, lower bid, or more observation. Cost alone does not make a relevant query irrelevant.

Use the negative keyword examples guide to sharpen the distinction between narrow exclusions and reusable intent patterns. Product identifiers need separate product-targeting decisions when supported. Do not assume a negative keyword blocks every product-detail placement associated with an identifier.

Review top of search, rest of search, and product-page performance using the placement data and controls available in your campaign. Calculate mature ACoS and conversion rate for each placement. A high click-through rate at top of search does not justify a higher bid if the resulting orders exceed your margin allowance.

Placement adjustments change the effective bid. In a simplified fixed-bid calculation, an 80 cent base bid with a 50% placement increase becomes $1.20 for that placement. Dynamic bidding and other applicable settings can affect the final auction bid, so evaluate the complete configuration. The displayed base bid is not always the maximum exposure.

Change one meaningful variable at a time when possible. For an initial placement test, a 10 percentage point modifier change can be easier to evaluate than doubling the base bid while adding a large multiplier. Set a daily change cap and an evaluation window. Keep an inventory check in the approval process so a bid increase does not accelerate spend on a product nearing a stockout.

Check daily for runaway spend, missing delivery, product eligibility issues, and stock constraints. Run the deeper search term review on a consistent schedule using mature evidence. Frequent bid changes driven by yesterday's orders can cause unnecessary oscillation when attribution is still arriving.

At the weekly review, inspect product economics first, then targeting groups, search terms, placements, and budget allocation. Expand an efficient manual campaign only when it can absorb additional useful traffic. A campaign exhausting its daily allowance is a delivery observation, not proof that a larger allocation is profitable.

• Recalculate available spend after material margin or price changes.
• Harvest qualifying terms and document their destination.
• Review costly terms with no mature orders.
• Check source negatives against manual eligibility.
• Compare placement efficiency and effective bid exposure.
• Record changes, owners, review dates, and reversal conditions.

Review total product-group spend and sales after restructuring. Moving attributed revenue from auto to exact is a reporting shift unless combined economics improve. Teams managing several stores can use the agency workflow to keep account rules and reporting consistent while preserving product-specific targets.