Conversion Rate (CVR)

Conversion rate (CVR) in Amazon Advertising is the percentage of ad clicks that result in an attributed order within the campaign's attribution window:

CVR = Orders / Clicks × 100

A campaign with 200 clicks and 18 orders has a CVR of 9%. CVR is the single most leveraged variable in the bid formula Bid = ASP × CVR × Target ACOS — and unlike ASP and Target ACOS, which are typically fixed by margin reality, CVR is movable through listing work, pricing, and creative.

Why CVR is the highest-leverage lever in the account

A 10% improvement in CVR permits a 10% increase in your max bid at the same Target ACOS. That bid increase typically unlocks more than 10% additional impression share — Amazon's auction returns non-linear volume gains for moving up the bid curve. The net effect of a 10% CVR lift is often 15–25% more sales at the same blended ACOS.

The reverse is also true: a 10% CVR decline (failed launch, broken main image, stock-out badge appearing) silently forces you to either accept higher ACOS or lower bids and lose impression share. CVR collapse is the second-most-common cause of "Amazon stopped working" complaints after stock-outs (which themselves crash CVR).

What CVR is sensitive to

Ranked by impact on most categories:

1. Price relative to competitors. A 10% price increase typically drops CVR 15–25% on commodity categories, less on differentiated ones.
2. Main image quality. The single most measurable single change. A/B testing the main image via Amazon's Manage Your Experiments tool routinely produces 5–25% CVR swings.
3. Star rating. The 3.9 → 4.1 cliff is real (the rating displays as "4 stars" instead of "3.9"). The 4.3 → 4.5 cliff is also real (most shoppers filter at 4.5+).
4. Review count. Below 50 reviews, every additional 20 reviews lifts CVR meaningfully. Above 500, the marginal effect approaches zero.
5. Prime badge. Non-Prime to Prime is typically a 20–40% CVR lift in categories where shoppers care about delivery speed.
6. A+ content / Premium A+. 5–15% CVR lift on average; higher when the category requires explanation (technical products, supplements).
7. Variations on the listing. Colour/size variations consolidate review count under one parent ASIN and lift CVR vs. fragmenting reviews across child listings.
8. Buy Box ownership. A lost Buy Box can drop CVR to near zero overnight.
9. Stock-out indicators. "Only X left in stock" displays at low inventory levels; depending on category this either lifts CVR (urgency) or kills it (delivery anxiety).

CVR at the keyword vs. campaign level

CVR varies enormously by search term — often by 4–8× across the keywords in the same campaign. This is why the bid formula uses the keyword-level CVR, not the campaign average. A blanket "raise bids 10%" on a campaign with a 7% average CVR over-bids the 4%-CVR keywords and under-bids the 12%-CVR keywords.

For statistical significance, the rule of thumb:

• ≥30 clicks on a keyword → CVR estimate stable enough to bid against.
• 10–30 clicks → use Bayesian shrinkage toward the campaign average.
• <10 clicks → use the campaign or ad-group average; don't bid against the noisy point estimate.

Diagnostic: where is CVR being lost?

Use Search Query Performance to break down the funnel by query: impressions → clicks → adds-to-cart → purchases. The step where your brand-share collapses is the step where you're losing CVR.

• Impression-share leak = bidding or organic-rank problem (not CVR).
• Click-share leak = main image, title, price, badge problem (pre-CVR — fix the SERP creative).
• Cart-share leak = A+ content, review, pricing, variation problem (PDP creative — this is the classic CVR fix).
• Purchase-share leak = checkout friction, stock, shipping, Buy Box (post-cart — fix operations).

Common mistakes

• Treating CVR as a fixed property of the SKU. CVR is a function of price, listing quality, reviews, badges, and traffic source. All of those are movable.
• Optimising bids without monitoring CVR. A CVR drift of -15% over two weeks silently destroys the entire bid model. CVR should be a weekly dashboard metric, not an afterthought.
• Reading campaign-average CVR as the bidding input. Use keyword-level CVR; the campaign average hides 4–8× variance.
• Confusing PDP CVR with ad CVR. PDP CVR (organic + paid combined) is reported in Brand Analytics. Ad CVR is reported in the Ads Console. They are different numbers and they move for different reasons.