Score five things: whether their ACoS target changes by product stage, whether they measure cost of customer acquisition rather than ACoS alone, whether they can diagnose a conversion problem, how many accounts one specialist runs, and what they commit to in writing. Everything else is presentation.
The short version
- Two ACoS numbers, launch and mature. One number for both is disqualifying.
- Cost of customer acquisition, not ACoS alone. ACoS improves when you stop spending.
- Give them a broken account and ask what they check first. A good answer does not start at bids.
- Accounts per specialist, in writing. We run about 1.4 brands per operator.
- Written commitments on reporting contents and response time.
A scoring sheet you can actually use
I run Flapen with 50 operators managing about 70 brands. Score each candidate out of two on these five, and stop interviewing anyone below seven.
| # | Criterion | 0 points | 2 points |
|---|---|---|---|
| 1 | ACoS by stage | One target for everything | Two numbers, each reasoned |
| 2 | Profitability metric | Reports ACoS and spend | Reports cost of customer acquisition by channel |
| 3 | Diagnosis | Starts at bids | Starts at conversion and click-through rate |
| 4 | Caseload | Will not say | A number, under eight |
| 5 | Written commitments | "We are responsive" | Cadence, contents, response times |
1. ACoS by stage
Paid traffic is not one thing. Text, image, and video ads have different economics, and the right target changes as a product matures.
A new product needs aggressive ACoS to build velocity and ranking. A mature product needs efficient ACoS to protect margin. Most sellers set one target and wonder why their ads stop working, and most agencies inherit that habit.
Give a candidate a real product and ask for the number at launch and the number twelve months in. You are grading whether the numbers move and whether they can explain the transition.
2. The profitability metric
ACoS is an advertising metric. Cost of customer acquisition is a business metric, and it is the one that says whether the growth is worth having.
There is a trap in ACoS worth naming: it improves when you spend less. An agency optimizing to an ACoS target can hit it by shrinking your business, and the report will look like progress.
3. The diagnosis test
The most predictive question. Describe an underperforming product and ask what they would check first.
A strong answer walks through listing quality, primary image click-through rate, conversion rate, ad performance, traffic channel activation, pricing, and return rate before arriving at bids. A weak answer starts at bids, because bids are the thing they are selling.
If your conversion rate is low, no amount of ad spend fixes it. An agency that cannot say that sentence will optimize around a broken listing for a year.
4. Caseload
Ask how many accounts one specialist carries and get it in writing. It is the single biggest driver of the attention your account receives and it never appears in a proposal.
5. Written commitments
Not guarantees on results, which are always hedged into meaninglessness. Commitments on process: what the weekly report contains, how fast they respond to a suppressed listing, who is named on your account, and when targets get reset by stage.
What to ask for before you sign
- A sample weekly report, redacted. Does cost of customer acquisition appear in it?
- The name of the person on your account, and their current caseload.
- Their ACoS target logic for your specific product at two stages.
- What they would check before touching a bid.
- What they would need from you to set a target that means anything. The answer should include landed cost.
Point five is a quiet filter. An agency that sets ACoS targets without asking your landed cost per unit is optimizing toward a number with no connection to your profit.
What most agencies will not tell you
Case studies in this category are close to useless. Percentage improvements are quoted from whatever baseline flatters most, over whatever window flatters most, on accounts you cannot verify, in categories unlike yours.
A 60 percent ACoS improvement can mean an account was badly mismanaged before, which says more about the previous agency than the current one.
Ignore the case studies and test the method live. Give every candidate the same real product and the same question, and compare the answers side by side. Fifteen minutes of that separates candidates better than a deck ever will.
One more: most PPC agencies run one part of one traffic channel. There are five, and image and video ads, promotions, the influencer and creator program, and off-channel are where the highest return on ad spend currently sits. Ask what happens to those, and who owns them if not the agency you are hiring.
Related answers
- How to vet Amazon PPC agencies
- KPI benchmarks for Amazon PPC agencies
- Compare Amazon PPC vs full-service agencies
- What to include in an Amazon PPC agency SLA
- Hiring an Amazon agency: the complete guide
Give us the same product and the same question you give everyone else, at Flapen.
