A PPC manager on Amazon is either the person who runs your campaigns, in-house or at an agency, or the software that automates bidding. The two solve different problems, so choose by market size first. Below a $2 million market nobody you hire earns their fee, and above it you pick the option whose failure mode you can afford.
The short version
- The role and the tool are two separate purchases. A person owns judgment. Software owns the speed of execution.
- The most expensive mistake lands before the first campaign. A market turning over less than $2 million a year cannot repay anyone you hire.
- Each option breaks in a predictable place. An in-house hire, an agency, a freelancer, and a software seat carry one failure each.
- About $1,000 a month in media is the working floor. Under that budget, whoever you hire reads noise and invoices you for it.
- Every failure below announces itself early. Catch the signal in month one and the mistake costs a month instead of a year.
The failures ranked by what they cost
Rank the ways this decision goes wrong by the money each destroys, then hire against the list. We refuse markets under $2 million a year, because once the cost of customer acquisition is paid there is nothing left to capture. Four signals decide whether a live product is scaled or stopped: rating trend, return rate, conversion rate, and cost of customer acquisition trajectory.
| Failure | What it costs | Early signal |
|---|---|---|
| Advertising into a market that cannot repay the spend | The whole budget, because no bid optimization reaches a market under $2 million a year | The candidate quotes a fee before asking what the market is worth |
| Paying someone to scale a product that should be stopped | Every month of spend past the 60 to 90 day window the four signals define | No report has ever contained the word stop |
| A fee that grows with the number it exists to control | Restraint, because your budget increase becomes their raise | The first strategic recommendation in month two is more budget |
| Buying automation and calling it management | A wrong rule run thousands of times before anyone reads the result | The account holds its ACoS target while unit volume falls |
| One person with no bench | An account nobody watches from the day that person leaves | A vacation week passes with nothing changed |
Ranked costliest first. Flapen figures as of September 2026.
The four options and the failure each one carries
Before Flapen I ran data and technology at BRANDED and Moonshot Brands, two large Amazon aggregators, and audited 60+ acquired brands doing $5M to $10M each. One pattern held across them. The structure of the arrangement predicted the result better than the skill of the person inside it.
| Option | What you pay | The failure it carries | The signal that shows it early |
|---|---|---|---|
| In-house hire | Salary, tooling, training time | One person's judgment, nobody to check it | Campaign structure stops changing after month three |
| Agency | A flat fee or a share of ad spend | Your account inherits the last client's template | Nobody names one thing they would handle differently |
| Freelancer | A small fee plus your media | Ads tuned in isolation from price, stock, and images | A bid change is proposed for an image problem |
| Software seat | A subscription per marketplace | Rules run faster than anyone verifies them | The dashboard counts optimizations instead of profit |
The rule I give sellers: buy the option whose failure you can detect and survive. One product rarely justifies a salary, and five rarely survive a freelancer's calendar.
At Flapen the work is priced flat, from $800 a month at one product to $2,400 at five, no commission on your spend, and 30 days' notice to leave. Budget about $1,000 a month in media beside it.
With one product and a thin budget, run campaigns yourself and buy a written audit instead of a retainer. That answer costs us the sale and remains correct.
The failures no hire removes
Two failures survive a good manager, and both are yours to fix before anyone starts.
- Skipping validation, then hiring someone to explain the numbers. Phase 1 is 200 units and $5,000 to $10,000, with up to four products tested at once, so a manager hired before that run has no baseline to attribute anything to. The early signal is that nobody can state the conversion rate the product needs to break even.
- Nobody being allowed to say kill. The person paid to run advertising rarely holds the authority to end the product, which is how a bad launch survives on a healthy ACoS. The tell is the pause after you ask what would make them tell you to stop.
What PPC managers will not tell you
Four things stay out of the pitch, ranked by cost, and on a careless day that includes us.
- A percentage fee makes restraint expensive for whoever advises you. The manager who tells you to cut spend cuts their own invoice, and no amount of integrity should have to carry that structure. You see it early when the proposal explains the percentage before the work.
- Most accounts receive the same campaign architecture. Match types, negatives, and bid rules travel from client to client, because that keeps a low fee low. It shows early in an answer about what would differ for you that arrives in adjectives.
- Software vendors report activity, not outcomes. Bids changed, keywords harvested, and campaigns created all rise whether or not the account earns more. The signal is a report that leads on the volume of changes.
- Free audits exist to start sales conversations, ours included. Judge the document rather than the offer, and ask whether a freelancer could execute it without the author. Watch for an audit with no ASIN, no number, and no fix you could run this week.
Hold us to the same list. If we quote a fee before sizing your market, do not hire us.
Related answers
- What ACoS means on Amazon
- How PPC works on Amazon
- Amazon PPC benchmarks for ACoS, ROAS, and TACoS
- Best Amazon agency for DSP optimization
- Amazon account measurement and audits: the complete guide
This week, at no cost, take your three highest-spend products and write the category's annual revenue beside each one, from your own sales data and the sellers you compete with. Any product inside a market under $2 million a year is a spending decision, not a management problem.
To have that sizing and your campaigns read against it in writing inside 48 hours at no charge, request the free audit at Flapen.
