Ecommerce PPC: Keywords, Buyer Intent and Negative Keywords

PPC means pay per click: you pay when someone clicks an eligible ad. Search PPC lets your store respond to expressed demand, but a search phrase is evidence of a need, not proof that a person will buy. Plan the query, the message and the destination together.

Start with your actual products

List product names, category terms, materials, dimensions, use cases and objections. Use customer interviews, site searches, support questions and search-term reports. Use a keyword planning tool to estimate demand and cost; estimates are not promised traffic or sales. Avoid building campaigns around products with unreliable stock or inadequate margin.

Classify buyer intent

Words such as buy or best help form hypotheses. Confirm intent through purchase rate, new-customer cost and contribution. A cart addition signals consideration; a completed order is a stronger outcome.

Group tightly related needs

Keep pantry storage separate from lunch boxes if the products and pages differ. Map each group to one relevant page and an RSA asset set. Start with a manageable scope. Google exact match can match searches with the same meaning or intent, not just an identical string. Phrase and broad match expand reach differently. Inspect actual search terms instead of assuming the keyword list fully describes delivery.

Build a negative keyword plan

Create candidate exclusions for products you do not sell, unsupported services and irrelevant needs. A store selling containers may review repair, wholesale or printable as exclusions, depending on its offer. Do not blindly exclude cheap, free or gift: those words can appear in relevant shopping searches such as free shipping or gift set.

Negative match types work differently from positive match types. Review relevant singular/plural variants and synonyms explicitly. Record the reason and scope for each exclusion so a campaign-level decision does not accidentally block useful demand across the account.

Review the economics

Weekly, record query, spend, clicks, purchases, net revenue, new-customer share and contribution after ads. Separate brand from non-brand demand. Pause irrelevant queries promptly; allow conversion lag before judging relevant ones. A high click-through rate with poor sales calls for checking the product promise, destination and price, not simply buying more clicks.

Turn a keyword list into a usable campaign map

For an illustrative travel-backpack store, build this small map before importing hundreds of terms. The query examples are starting hypotheses; validate demand and relevance for your actual catalog.

Keep brand results visible separately so existing demand does not make prospecting appear more efficient. A product comparison may deserve an organic article or a different test budget from a transactional search. Choose the destination that resolves the question.

Use a negative-keyword decision log

Suppose the store sells new retail backpacks only. A query for backpack sewing pattern suggests a different need. A query for backpack repair suggests a service it does not offer. Review whether a phrase negative or a specific exact negative best matches the unwanted traffic, and check what else the exclusion would block. Do not upload this example as an account-wide list without checking the rest of your business.

Log the observed query, spend, reason, chosen match type, scope and date. Revisit exclusions when you add products or services. For a store that later sells repair kits, an old account-level repair exclusion can block useful traffic.

Diagnose a query with numbers

Illustrative example: one theme receives 120 clicks at $1.50 each, creating $180 in spend. Six purchases produce a $30 purchase CPA. If only four customers are new, media CAC is $45. At $40 contribution before ads per first order, the new-customer economics need attention even though the overall purchase CPA looks acceptable. Segment the actual revenue and costs before deciding.

A theme with 20 clicks and no orders is less conclusive. At an assumed 3% conversion rate, the probability of zero purchases in 20 independent clicks is about 54%. This simplified model illustrates uncertainty; actual ad traffic is not a perfectly stable random sample. Use relevance, conversion lag and your loss ceiling alongside the count.

Weekly maintenance you can finish

Review actual search terms, destination relevance, spend and mature conversions. Mark each theme keep, investigate or exclude, with a reason. The search-term report may not expose every query, so compare the available detail with total spend. Keep the original hypothesis and a change log so next week's result has context.

Put this guide to work

Open the advertising campaign workbook, complete the section for this lesson, and record one decision you will make next. Keep your assumptions beside the evidence so you can update the plan when results arrive.