After managing hundreds of brands on Amazon, there is one takeover call we can predict word for word. The brand runs a separate Amazon PPC agency and a separate creative agency. Conversion rate on the hero SKU has slipped two points over a quarter. The PPC agency says the listing is not converting the traffic. The creative agency says the ads are buying the wrong traffic. Both have a deck. Both are partly right. And nobody in the arrangement is responsible for the number that actually fell.
This is not a personality problem with either vendor. It is a structural gap, and it costs more than either retainer.
The metric nobody owns
Look at what each vendor is graded on.
The PPC agency is graded on ACOS, ROAS, maybe TACoS. Every one of those is a ratio with ad spend on top. The agency controls the top of the ratio and about half of the bottom. It can decide which shoppers arrive and what you paid for them. It cannot decide whether they buy.
The creative agency is graded on CTR when a hero changes and, loosely, on “the listing looking better.” It controls what a shopper sees after the click. It does not control who the shopper is, what they searched, or whether they came from a discovery campaign at 2am or a branded term at noon.
Conversion rate sits exactly between those two scopes. CVR is a function of who arrived multiplied by what they saw. Split the two inputs across two companies and you have built a metric with two half-owners and no whole one.
The reporting reflects it. The PPC deck shows campaign CVR by targeting type. The creative deck shows unit session percentage before and after a stack change. Neither report contains the other’s variable, so each vendor’s explanation is internally consistent and externally unfalsifiable.
How the blame loop actually runs
We have watched this sequence enough times to lay it out by month.
Month one. CVR on a $90K/mo SKU drifts from 12.1% to 10.8%. Nobody flags it because it is inside normal variance.
Month two. ACOS rises with it, because the same clicks are buying fewer orders. The PPC agency reports “listing conversion has softened” and recommends the brand “review creative.” Reasonable.
Month three. The creative agency reviews the stack, finds nothing changed on the listing since March, and reports “creative is performing consistently, we would look at the traffic mix.” Also reasonable.
Month four. The brand, now down four points of CVR and three of ACOS, asks both vendors for a fix. The PPC agency lowers bids on the worst-converting campaigns, which reduces spend and makes ACOS look better while reducing sessions. The creative agency proposes a hero refresh, which will take ten weeks and clear an A+ queue.
Month five. Sessions are down, the hero refresh is in production, and the brand cannot tell whether the original decline was ever diagnosed. It was not. In two of the last three cases we took over with this shape, the actual cause was a competitor entering the category at a lower price point, which changed the comparison set every paid shopper saw. Neither vendor’s report contained that variable either.
Total cost: a quarter of drift, a bid reduction that gave up rank, a creative rebuild with no baseline to grade it against, and two agencies each able to show they did their job.
The four questions a single owner asks first
A team that owns CVR end to end asks these questions in the first week, in this order. Split vendors rarely ask any of them, because each question crosses the scope line.
Did the traffic mix change? Pull branded versus non-branded sessions. A shift of five points toward non-branded traffic will drop blended CVR on a healthy listing with nothing wrong on the page. This is a PPC-side fact that explains a creative-side metric.
Did the comparison set change? Screenshot the search grid for the top three queries and compare against last quarter. New entrants, price moves, and competitors’ image changes all move your CVR while your listing sits untouched.
Did the page change without anyone deciding? Check listing change history. Amazon’s own tools have been auto-publishing title and attribute changes on review clocks shorter than most brands’ catalog cadence. A stripped phrase or a changed variation theme moves conversion and generates no alert.
Did the split by device move? Business Reports can split sessions and unit session percentage by mobile app versus browser. A gap that opens on one SKU and not the others is a rendering problem, not a traffic one.
None of these takes more than thirty minutes. All four require someone whose job includes both the ads console and the detail page.
What it costs in dollars
On a $200K/mo brand with $30K/mo ad spend, a one-point CVR decline is not a rounding error.
At 12% CVR and roughly 140,000 monthly sessions, the brand converts about 16,800 orders. At 11%, it converts 15,400. That is 1,400 orders a month at, say, $34 and $19 of contribution before ads: about $26,600 in contribution gone, with ad spend unchanged. ACOS rises about two points on the same spend for the same reason.
Now add the blame loop’s costs. A defensive bid cut of 20% on non-branded campaigns gives up paid velocity, which feeds organic rank; the position is bought back in the following quarter at an ACOS the brand would have refused. A hero rebuild ordered without a diagnosis is a five-figure creative spend graded against a moving baseline. And the diagnosis itself is delayed by a quarter because each vendor’s monthly report closes the question inside its own scope.
The retainers are the smallest line on that list.
Why brands split in the first place
Fair to say: brands do not do this out of carelessness.
Some inherited it. The PPC agency was hired in 2023, the creative shop in 2025, and nobody revisited the structure. Some did it deliberately, because a specialist PPC shop pitched deeper bid management and a specialist creative shop pitched better images, and both pitches were true. And some did it because a single full-service agency underdelivered on one half, so the brand carved that half out and gave it to someone else.
All defensible. The problem is not the choice to specialize. The problem is that the brand became the integration layer without anyone saying so. When two vendors each own half of a metric, the only party who sees both halves is the founder or the marketing lead, and that person is now doing diagnostic work they hired two agencies to do.
How to fix it without firing anyone
You have three options, in ascending order of disruption.
Name a CVR owner internally and give them both consoles. One person, usually the marketing lead, is the owner of unit session percentage on the top 20 ASINs. Both agencies report to that person on CVR, in the same document, against the same baseline. This person runs the four questions above monthly. It is the cheapest fix and it works if the person actually has time, which in Q4 they usually do not.
Make one agency the lead and put it in the contract. Either the PPC agency or the creative agency is designated the diagnosis owner for conversion. When CVR moves, that agency runs the four questions and produces a written finding before any bid change or creative brief is approved. The other vendor executes against the finding. This removes the deck-versus-deck dynamic without changing who does the work. The agency that pushes back on being the lead has told you something.
Consolidate. Bring advertising, creative, and catalog under one team with one report and one owner of CVR. This is the model we run, and we would be lying if we said it was free of trade-offs: a full-service team is not always the deepest specialist at any single lever. What it does have is a single party that cannot explain a CVR drop by pointing at a vendor it does not manage.
Whichever you choose, do it before Q4. A CVR decline discovered in November inside a split structure will not be diagnosed until January, and it will be diagnosed against peak-contaminated data by two vendors each protecting their renewal.
Three things to do this week
Ask each agency, in writing, who owns conversion rate. If both answer with a paragraph about collaboration, nobody does. If one answers “we do, and here is the baseline we track it against,” you have found your lead.
Pull one report both vendors have never seen. Unit session percentage by ASIN, split by mobile app and browser, for the trailing 90 days, alongside the branded versus non-branded session split. Send it to both. The first meeting where both vendors look at the same data is usually the meeting where the actual cause surfaces.
Freeze reactive changes until the diagnosis is written. No bid cuts and no creative rebuilds on the hero SKU until someone has answered the four questions in a document with a date on it. A quarter of drift is expensive. A quarter of drift plus an undiagnosed bid cut plus an unmeasured rebuild is how a brand loses a year.
FAQ
Is a full-service agency always better than specialists?
No. Specialists can be deeper at a single lever. The question is not depth, it is whether anyone in your structure owns the metric that sits between the levers. If you run specialists, name an integrator, in writing, with both consoles.
Our PPC agency says the listing is the problem. How do we check?
Split CVR by branded and non-branded traffic first. If branded CVR is flat and non-branded fell, the traffic mix or the comparison set moved. If both fell together, look at the page: change history, device split, and the search grid your listing now sits in.
Our creative agency says the ads are the problem. How do we check?
Look at whether the decline started on a specific campaign or across all traffic including organic. An organic-and-paid decline at the same time is almost never an ads problem.
Can we make the two agencies talk to each other?
You can put them on the same call. What you cannot do is make either accountable for a number outside its contract. Shared meetings without a named owner produce shared explanations, not decisions.
What is the fastest sign that the split structure is costing us?
Two monthly reports that each explain a CVR change by referring to the other vendor’s area. Once you see that pattern twice, the diagnosis is not going to come from either report.
If you’re looking for a team that manages every lever — creative, advertising, and operations — Velocity Sellers works with brands doing $100K+/month on Amazon. Contact us for a free account audit.
