Did conversion actually fall, or did your traffic change?
Answer this before anything else. If you recently added auto or broad campaigns, widened keywords, raised bids or launched on a new term, you bought different traffic — traffic that matches your product less well. Session-based conversion falls because the mix changed, not because the page got worse.
This is the single most common misdiagnosis in the category, and it is expensive: it sends people off rewriting a listing that was converting fine, which then makes the next month impossible to read.
How to separate them:
- Line the conversion drop up against your campaign change dates. If the drop starts the week you widened targeting, you have your answer.
- Look at organic sessions on their own. If organic conversion held steady and the blended number fell, the listing is not the problem.
- Check whether ad share of total sessions rose. A bigger paid share on looser match types drags the average down by arithmetic.
- Note that Business Reports gives you a unit-based and an order-based session figure. A multi-unit promotion or a pack change moves one and not the other, which reads as a conversion change that never happened.
What should you rule out before anything subjective?
Hard breakages. They are binary, fast to check, and they produce step-changes rather than drifts — so if one of these is live, nothing about your images or copy is worth discussing yet.
- Buy box. If you lost it, shoppers are converting for someone else on your own detail page.
- Stock and buyability. Not just zero stock — intermittently unbuyable, a regional gap, or an ASIN that keeps flicking in and out counts too.
- A variation or parentage change. A child detached from its parent loses the reviews and the traffic the family was carrying.
- Price. Yours, and the comparison set around you. A competitor discount moves your conversion without you changing anything.
- A suppressed element or a suppressed listing — see the suppressed listing guide for the causes and the order of the fix.
- Delivery speed and shipping. A change in the promised date, or losing Prime eligibility, moves conversion quickly and silently.
Pause ads on anything that turns out to be unbuyable. You are paying for clicks that cannot become orders, and the data is teaching the algorithm the wrong thing.
What does the shape of the curve tell you?
The shape is the fastest triage available, and most people skip it. Plot conversion daily over a long enough window to see both, then read it:
- A cliff on one date. Something broke or changed. Go straight to the breakage list above, and to your own change log for that week — including changes made by someone else on the team.
- A slow drift downward. Nothing broke. This is competitive pressure, seasonality, review erosion, or a page that stayed the same while the results around it improved.
- A step down that never recovers. Usually a permanent change in the comparison set: a new competitor, a badge someone else won, or your own price now sitting on the wrong side of a threshold.
- Noise on a low-volume ASIN. Sometimes there is no drop. On thin traffic, a normal week looks like a trend if you squint at it.
Cliffs and drifts have almost no causes in common. Deciding which one you are looking at removes most of the search space in about a minute.
What should you segment before you conclude anything?
An account-level conversion number is an average, and an average hides the thing that actually happened. One collapsing SKU inside a healthy catalogue looks like a mild decline everywhere.
- By ASIN. Always first. If one ASIN moved and the rest held, you have a product problem, not an account problem — and a much smaller thing to fix.
- Ads versus organic. The traffic-mix test from the top of this page, applied per ASIN.
- By device. Most shoppers are on a phone. A layout or image change that reads fine on your desktop can be unreadable at thumb size.
- New-to-brand versus repeat. If repeat buyers still convert and new shoppers do not, the problem lives in first-impression material — image, price, reviews — rather than in the product.
Segment until the drop stops being an average and becomes a specific thing on a specific page. Then, and only then, look at the page.
When is the listing itself the problem?
Once breakage is ruled out and the segments point at one ASIN, work down the page in the order a shopper meets it:
- Main image, at mobile size. Look at it as a thumbnail next to the competitors it appears beside, not full-screen in isolation. Most image failures are legibility failures.
- Price relative to the comparison set on the page. Not to your history — to what is sitting next to you today.
- Review count and rating trend. Recent negatives move conversion faster than old ones, because they are what shoppers read first. Read them for a cause, not just a count.
- A+ content. Missing, or present but never revisited since launch — see A+ content for what it should be carrying.
- In beauty: selection friction. Shade and size choice is where beauty conversion is won or lost. If shoppers cannot tell which shade is theirs, they leave — and the ones who guess come back as returns. Broken variation families make it worse. See shade and size variations.
- The returns signal. In beauty, shade mismatch and irritation show up in returns and reviews before they show up anywhere you were looking. Treat the return reasons as conversion research.
Then look off your own page. A competitor launched, cut price, won a badge, or fixed the thing you were beating them on. Seasonality moves whole categories, and a category-wide shift is not something your listing did. If everything on your page is unchanged and the drift is real, the change happened around you — and the response is competitive, not cosmetic.
How do you fix it without losing the diagnosis?
One change at a time, dated, written down. Change the image, the price and the title in the same week and you will never know which one moved the number — and if it gets worse, you have nothing to roll back to.
- Pick the single most likely cause from the segments, not the easiest one to edit.
- Change it, note the date and the reason.
- Give it long enough on that ASIN to be a signal rather than a week of noise.
- Log the outcome, including the changes that did nothing. Those are the ones that stop you repeating the work next quarter.
A change log with dates is what makes the next diagnosis fast. It turns "conversion is down" from a week of guessing into a comparison against a list.
Judge the result on profit, not on conversion alone. A price cut will usually lift conversion and can still leave you worse off. Work out what the product needs to clear after ad cost — our break-even calculator does the arithmetic — and check the fix against that number.
What this cannot tell you: whether your conversion is objectively good. Nobody can, and any page quoting a benchmark percentage is quoting an average across products that have nothing to do with yours. Your trend and your break-even are the only two comparisons that mean anything. It also cannot fix a product whose price, cost or fee structure does not work — that is an economics problem wearing a conversion problem's clothes.
Where to go next
Questions people actually ask
Want the drop traced on your own account?
Read-only access, three days, a real report with real numbers — no obligation. If we can't find enough to justify our fee, we'll be the ones to say so.
No spam. No 12-field forms. Just a real reply from our team — usually the same day.
Prefer to talk first? Book a 30-minute call