How to Analyze Amazon Ads Branded Search Metrics

Amazon Ads describes branded searches as searches for a brand promoted in a campaign that are attributed to an ad interaction. Its current enhancement notice also defines branded searches from clicks, branded searches from views, branded search rate, and cost per branded search. Copy the exact definitions into the analysis brief and record the date, marketplace, campaign types, attribution settings, and reporting surface used for the export.

Keep this family separate from orders, sales, Store visits, and new-to-brand metrics. A branded search is a discovery behavior, not a completed purchase. New-to-brand is based on whether an ad-attributed purchase came from a customer who had not purchased the brand's products on Amazon during the applicable prior period. Label every chart with the behavior it actually measures so a rising search count is not presented as customer acquisition.

Choose the business question. You might ask whether a launch campaign is followed by more brand-seeking behavior, whether a creative format produces different click-attributed and view-attributed search patterns, or whether cost per branded search is changing. Write the decision threshold and the other evidence required before budget can move. This prevents a newly available metric from becoming a target without an operating purpose.

Pull the available measures with campaign, ad product, creative, placement, targeting, product family, marketplace, and date fields that the reporting surface supports. Amazon says the enhanced metrics are available across several advertising products and surfaces, while Sponsored Products are excluded from this branded-search availability. Confirm eligibility in the current account before creating a cross-campaign comparison that silently mixes measured and unmeasured rows.

Calculate rate and cost only from compatible rows and periods. Amazon defines branded search rate relative to impressions and cost per branded search relative to cost. Recompute those formulas from the exported components when possible, then compare the result with the interface. Exclude incomplete periods and document currency, time zone, attribution window, and whether click-attributed and view-attributed components overlap in the reported total.

Use a stable comparison design. Compare the same marketplace, campaign type, product family, and reporting settings across periods. Annotate launches, major creative changes, pricing events, inventory outages, promotions, and measurement updates. Amazon announced enhanced brand matching for brand names, trademarks, abbreviations, and misspellings, so a trend that crosses the change can reflect measurement as well as shopper behavior.

Start with volume, then add denominators. Branded searches can rise because campaigns reached more shoppers, because a larger share searched, or because the measurement definition changed. Branded search rate helps separate impression growth from the share of impressions followed by the behavior. Cost per branded search adds spend context, but neither measure identifies profit or proves that the advertising caused an incremental search.

Compare click-attributed and view-attributed searches without declaring one superior by default. Click-associated behavior may reflect an immediate path, while view-associated behavior can indicate later brand seeking after exposure. The two groups can differ by format, placement, and audience. Use the platform's reported definitions, preserve attribution context, and avoid adding components unless the documentation or export schema confirms that the totals are mutually exclusive.

Pair the metrics with downstream evidence. Review Store visits, product detail engagement, orders, sales, new-to-brand measures, and contribution evidence available to the account. Look for coherent movement, but describe the observation accurately. A campaign can generate branded search behavior without profitable orders, and an established brand can produce purchases without a large change in branded search rate. The decision should reflect the full operating objective.

Create a review table with the campaign job, branded searches, search rate, cost per branded search, click and view components, downstream evidence, known confounders, and proposed action. Require a written reason for increasing, holding, reducing, or retesting spend. A strong discovery signal can justify a controlled follow-up, but it should not automatically override weak inventory, margin, or purchase evidence.

Use canaries for changes. If one creative or audience appears to improve branded search rate, test it in a bounded campaign or product set with the same reporting contract. Define the evaluation period and stop condition in advance. Preserve the previous setup so the team can reverse a change if cost rises or downstream evidence weakens. Do not backfill a narrative around a result selected after reviewing many segments.

Maintain a metric-change log. Record Amazon announcements, account eligibility changes, attribution setting changes, and reporting-surface differences. Rebaseline after a material definition update rather than forcing the old series and new series into one uninterrupted benchmark. The useful output is a traceable decision that names what the platform measured, what the business observed, and what remains uncertain.

Archive the export or screenshot used for each decision with its date range, filters, currency, marketplace, and attribution settings. If the same chart is refreshed later, preserve the earlier evidence rather than overwriting it. This lets reviewers distinguish a changed recommendation from a changed underlying metric, late platform processing, or an account configuration update.