Amazon PPC Keyword Research: A Step-by-Step Process That Actually Converts

Amazon PPC keyword research means building a list of the search terms shoppers type when they are ready to buy a product like yours, then deciding which of those terms deserve a bid. The fastest reliable process is: write a seed list from your listing, let an auto campaign and Amazon's own suggestions expand it, then keep only the terms your search term report proves can convert.

Most sellers get keyword research backwards. They paste a giant list from a tool into a broad match campaign, spend a few hundred dollars, and wonder why ACoS is 80%. The goal is not volume. The goal is relevance, because relevant ads tend to earn better click-through and conversion rates, while irrelevant ones mostly produce wasted spend.

Step 1: Build a seed list from your own listing

Start with what you already know. Open your product detail page and write down every way a shopper might describe the product. Group the terms into four buckets so you can bid on them differently later.

• Core product terms: the two- or three-word name of the thing ('silicone spatula', 'yoga mat').
• Feature and attribute terms: material, size, color, count, use case ('heat resistant spatula set', 'thick yoga mat for bad knees').
• Audience and occasion terms: who it is for or when they buy it ('yoga mat for kids', 'gift for bakers').
• Branded terms: your own brand name and product line, plus competitor brands you may want to target later.

Aim for 20 to 50 seed terms. Use the words in your title, bullets, and backend search terms, because Amazon is more likely to consider your ad relevant for terms your listing already covers.

Step 2: Expand with Amazon's own data sources

Amazon gives you several free ways to see real shopper language. Use them before you pay for a third-party tool.

Search bar autocomplete

Type each seed term into the Amazon search bar and note the suggestions. Autocomplete reflects popular searches, so it is a quick sanity check on phrasing. Try adding letters after the seed ('yoga mat a', 'yoga mat b') to surface long-tail variations.

Suggested keywords in the Ads console

When you create a manual Sponsored Products campaign and choose keyword targeting, Amazon shows suggested keywords for the ASINs you are advertising, usually with a suggested bid. These suggestions are based on your product, so they are a good expansion source, but review each one; suggestions can drift toward loosely related products.

Brand Analytics (if you have Brand Registry)

Brand-registered sellers can use the Top Search Terms report and the Search Query Performance dashboard to see which queries drive clicks and purchases in their category. This is the closest thing to real search demand data Amazon offers. We cover it in depth in how to use Brand Analytics for PPC.

An auto campaign as a research engine

An automatic campaign lets Amazon match your ad to searches based on your listing content. After one to two weeks, the search term report shows you exactly which queries triggered clicks and sales. This is the single best source of converting keywords because it is your own data. See how to harvest keywords from auto campaigns for the full process.

Step 3: Score every keyword before you bid

A raw list is not a plan. Score each term on three questions, then decide how to target it.

Keywords that score high on all three go into exact match with your strongest bids. Mid-scoring terms go into phrase match to discover variations. Low-relevance terms either get skipped or added as negatives. If you are new to match types, read Amazon match types explained first.

Worked example: from 40 seeds to a launch list

Say you sell a 12-piece silicone baking mat set at $29.99 with a 35% profit margin before ads. Your break-even ACoS is 35%, and you want to launch at a 30% target.

1. You write 40 seed terms from the listing and Amazon autocomplete.
2. You remove 12 that describe a different product ('silicone baking cups', 'macaron mat only') or that are pure research terms ('how to use silicone baking mat').
3. Of the 28 left, 8 are tight, high-intent matches ('silicone baking mat set', 'non stick baking mat half sheet'). These go into an exact match campaign.
4. 15 are relevant but broader ('reusable baking sheet liner'). These go into a phrase match campaign at a lower bid.
5. 5 are competitor brands. You park them for a separate competitor targeting campaign once you have reviews.
6. You also launch an auto campaign at a modest bid to find terms you missed.

To set starting bids, work back from your target ACoS. If your conversion rate is around 10% and your price is $29.99, each click is worth roughly $3.00 in revenue. At a 30% target ACoS, your maximum sustainable cost per click is about $0.90. The PPC bid calculator does this math for you.

Step 4: Let performance data replace guesswork

Keyword research does not end at launch. After two to four weeks, the search term report becomes more valuable than any tool. Use these decision rules on a weekly cadence:

• A search term with 2 or more orders and ACoS at or below target: add it as an exact match keyword in its own ad group or campaign.
• A search term with 15 or more clicks and zero orders: add it as a negative exact in the source campaign, unless it is a core term you know should convert (then check your listing and price first).
• A search term that is clearly irrelevant, regardless of clicks: negate it now. See how to find irrelevant search terms.
• A keyword with impressions but almost no clicks: your bid may be too low for visible placements, or your main image and price are not competitive.

Common keyword research mistakes

• Bidding on every term a tool lists. Third-party search volume estimates are modeled, not exact. Treat them as directional.
• Ignoring long-tail terms. Longer, specific queries often have lower CPCs and higher conversion.
• Putting the same keyword in many campaigns, which splits data and muddies decisions.
• Skipping negatives. Every campaign that uses broad, phrase, or auto targeting needs a negative keyword list from day one.

If reviewing search terms every week is the part that keeps slipping, AdSpark's AI search-term review flags converting terms to harvest and wasteful terms to negate, and logs every change so you can reverse it.

FAQ