There is no industry ROI table worth trusting, because nobody audits the inputs. Build your own from four numbers: incremental profit, total cost of service, time to first measurable change, and the share of your catalog that improved. Anything an agency quotes as a benchmark is a marketing figure until it is reproduced in your account.
The short version
- Published benchmarks are self-reported. No independent body verifies agency performance data.
- Four numbers replace the whole table: incremental profit, total cost, time to first change, catalog coverage.
- Thirty days is the honest first checkpoint, and it measures direction rather than return.
- Cost of service is more than the fee. Add ad spend, creative, and your own hours.
- A benchmark you cannot reproduce is a claim, not a benchmark.
The number that actually decides it
Your break-even ACoS. Everything else in this conversation is decoration until that number exists.
Break-even ACoS is your contribution margin expressed as a percentage of sale price, after Amazon fees, landed cost, and returns. If a product sells at $30 and contributes $8 after all of that, break-even sits near 27 percent. Every campaign, every agency claim, and every ROI benchmark you read gets judged against your figure, not against a category average from someone else's catalog.
Once you have it, industry benchmark tables stop being useful, which is the point. A "3.5x average return" means nothing without knowing the margin structure underneath it, and margin structure is exactly what those tables never disclose.
The four numbers to track instead
| Number | How to calculate it | Read it at |
|---|---|---|
| Incremental profit | Profit in the period minus the pre-engagement baseline, net of returns | 90 days |
| Total cost of service | Management fee plus ad spend plus creative plus your own hours | Monthly |
| Time to first measurable change | Days from onboarding to a metric moving outside noise | 30 days |
| Catalog coverage | Share of ASINs that improved, not just the hero product | 90 days |
The fourth is the one that exposes weak management. An agency can lift a single strong ASIN and show a good chart while the rest of the catalog drifts. If eighteen of your twenty products look the same after six months, the engagement is a campaign service with a management invoice attached.
Where an ROI number goes wrong, worst first
- Judging on revenue instead of profit. The most expensive and most common. Revenue rises when you discount, spend, and expand into unprofitable keywords. All three feel like progress on a dashboard.
- Buying traffic into a page that does not convert. If conversion is the constraint, no amount of ad spend fixes it. You multiply the leak and pay for the privilege. Fix the images, the price, and the reviews first, then scale.
- Comparing to a benchmark from a different margin structure. A supplement brand at 60 percent contribution and a home goods brand at 18 percent cannot share a target. Borrowed benchmarks make good accounts look bad and bad accounts look fine.
- Measuring too early. Thirty days shows whether efficiency is moving. It does not show return, because returns have not landed and rank has not settled. Calling ROI at day 30 produces decisions you reverse at day 90.
- Ignoring who is doing the work. The least visible and the slowest to hurt. Results depend on the hands on the account.
Ask where the people are
That last failure mode is worth its own section, because it is the input that no benchmark captures. Ask any agency who performs each function and where they sit. Listing copy, images and video, PPC, sourcing, catalog health, translation. Then ask which of those are subcontracted.
Our answer is that none of it is. About 50 operators do the work in-house, with a creative studio in Dubai, a sourcing studio in Guangzhou, and an internal technology team building the ads, marketing, and brand valuation tools we run on. I am not claiming a subcontracted model cannot work. I am claiming you should know, because a chain of vendors is a chain of handoffs, and every handoff is a place where your account waits.
The reason this belongs in an ROI discussion is speed. Time to first measurable change is one of your four numbers, and it is decided almost entirely by whether the person who spots the problem can also fix it.
What most agencies will not tell you
Benchmark figures are produced by the people they flatter. There is no audit, no standard definition of return, and no requirement to disclose which accounts were excluded. A single successful launch can be presented as a house average, and there is no mechanism that would catch it.
The second thing: some of the best work produces a temporary drop. Cutting wasteful spend, killing a weak variation, or repricing to protect margin all reduce revenue in the month they happen. Judged against a growth benchmark, the right decision looks like failure. If you hire on benchmarks alone, you will fire the people doing the honest work.
Related answers
- ROI calculator for Amazon PPC and listing optimization
- Common pitfalls that delay Amazon agency payback
- KPIs an Amazon agency should report weekly
- Amazon agency red flags to watch out for
- Amazon agency pricing and economics: the complete guide
If you want a written diagnosis of your account before any of this is theoretical, request the free audit at Flapen.
