312: Should You Hire An Agency To Run Your Amazon PPC Ads With Edward Ruffin And Jeremy Crowe

Podcast: Download (Duration: 51:54 — 59.7MB)

An Amazon PPC agency makes sense above roughly $2,000 a month in ad spend, because below that the management fee consumes whatever improvement they produce. I ran this experiment on my own account for a year, and the result was about $5,800 in additional ad spend producing roughly $26,000 in additional revenue.

In this episode I spoke with Edward Ruffin and Jeremy Crowe of Seller Labs, who actually managed my Amazon ads during that period. This is a candid account from both sides rather than a pitch.

Below is everything: the spend threshold, what they restructured first, how long the transition takes, why listings are usually the real problem, and what the fees look like.

Key takeaways

• Roughly $2,000 a month in ad spend is the threshold below which an agency cannot justify its fee.
• My results: $5,800 more ad spend produced about $26,000 more revenue, roughly a 4.5x return.
• Account restructure is the first move, with one ad group per SKU inside category campaigns.
• Expect three months to transition fully from your old campaigns to theirs.
• Listings cause most ad failures. In my experience it is over 80% of the time.
• Expect higher spend and ACOS initially while a new structure gathers data.
• Fees are flat tiers by ad spend rather than a percentage, which leaves room to grow.
• Ask where the team actually sits. Plenty of agencies route work to cheap offshore labor.

When an Amazon PPC agency makes sense

The two arguments are time and accumulated pattern recognition.

Edward’s framing on time is direct. You may fully understand PPC and still not have multiple hours weekly to spend on it, and that time may be worth more spent on inventory, new products, or brand development.

The second argument is exposure. An agency has likely worked with something similar to your product, and testing across many accounts produces patterns a single seller cannot see.

The platform changes constantly, which compounds that advantage. Amazon adds targeting options and placements regularly enough that staying current is itself a job.

The ad spend threshold below which it does not work

Seller Labs’ rough minimum is $2,000 a month in ad spend.

The reason is arithmetic. Someone spending $100 a month cannot generate enough incremental sales for a management fee to make sense.

Jeremy’s version of the same point is about outcomes. His job is producing clients who are still happy three months in, and a client without the budget to act on recommendations will not be.

They will work below that threshold when the seller genuinely intends to grow, and otherwise recommend self-service tools instead.

What actually gets fixed first

Account structure is the first change on nearly every account they take on.

Edward restructures roughly 99% of the accounts he works with, and describes the resulting structure as the thing clients keep even if they later take the account back.

The structure is one campaign per product category, with one ad group per SKU inside it.

The reason is data clarity. All napkin search terms flow through a napkin campaign, and each SKU’s ad group shows which specific search terms work for that specific product.

The problem it solves is overlap. When a product is advertised in multiple places, impressions split across them, which slows down how long it takes to learn anything useful.

The Amazon auto campaign settings most sellers ignore

Auto targeting campaigns have four separate targeting groups with independent bids.

Two are keyword based. Close match covers tightly related keywords, and loose match covers keywords Amazon associates more broadly with your product.

Two are product based. Substitutes are products people bought instead of yours, and complements are products people bought alongside yours.

Before those existed, an auto campaign had a single default bid controlling everything.

What Jeremy did on my account was split them across two campaigns. One running the two keyword-based options and a mirrored one running the two product-based options.

The benefit was cost control. That separation let them pay appropriately for the product-targeting side rather than applying one bid across fundamentally different traffic.

Keywords ended up working better for my account, which the split made visible.

How long the transition takes

Three months is the benchmark for moving fully onto new campaigns.

The method is gradual rather than abrupt. Existing campaigns get decreases on poorly performing elements while new campaigns build up alongside them.

Anything still performing well after three months stays running. Jeremy’s position is that killing a profitable campaign purely because it is not theirs would be leaving money behind.

His reasoning involves how Amazon accumulates data. If an auto campaign has gathered the right signals and performs well, that history has value that a new campaign does not have.

Why your ads are probably failing on the listing

From teaching my course, my estimate is that over 80% of the time a seller cannot make Amazon ads work, the listing is the problem rather than the ads.

Edward’s framing is that the listing is the foundation. You can advertise anything, and a product that does not present well will not sell.

The metrics they watch are sessions and unit session percentage. Jeremy’s rough thresholds are around a thousand sessions and roughly 10% unit session percentage before he can make confident decisions.

Falling below that is not automatically a red flag. Category competitiveness matters, and a conservative starting bid in a category with $5 clicks will naturally produce low session counts.

Back end keywords are the other common cause. An unfilled back end means you are not indexing for terms you should be, which damages traffic broadly.

Their handling of bad listings varies. Minor issues get feedback, and substantial problems get a conversation about whether to take the work on as a separate service.

Why listing problems often surface late

A large catalog makes it impractical to audit every listing before starting.

Jeremy’s account is that with thousands of products, reviewing each listing individually would consume the time meant for advertising.

So the pattern is that ads run, conversion underperforms, and the investigation into why leads back to the listings.

The complicating factor is that advertising sometimes fixes it. A reasonable listing that never got attention can start indexing better once traffic arrives, which lifts organic and paid performance together.

What results to actually expect

My own goal was more total profit and sales, accepting a higher ACOS to get there.

The benchmark we settled on was year over year comparison, since cost per click rising was a given.

Jeremy’s approach was finding secondary metrics that would demonstrate progress regardless. Click-through rate, which measures whether people engage with the ad, and TACOS, which compares ad spend against total revenue rather than just attributed revenue.

The numbers after the three month settling period: ad spend rose by roughly $5,800 and revenue rose by roughly $26,000, which is about a 4.5x return on the incremental spend.

The honest caveat is my margins. They are high enough to justify spend that many sellers could not.

The largest gains came from product targeting, which my account had barely used. Most of my sales came through brand searches from people who already knew us.

What agency fees look like

Seller Labs charges flat monthly tiers based on ad spend rather than a percentage.

Edward’s argument for flat tiers is that they leave room to grow. You can scale within a tier without the fee rising, and renewals carry a discount.

Additional services like store builds and listing optimization are priced separately rather than bundled.

Jeremy’s framing on low-margin clients is worth noting. He had a client with roughly a dollar of margin per product, and the conversation there has to include how to improve that margin rather than only how to advertise into it.

What to ask before hiring any agency

The horror stories are real and worth screening for.

The pattern I have heard repeatedly is a polished salesperson closing the deal and the actual work routed to inexperienced cheap offshore labor learning on live accounts.

My own decision to try this came entirely from knowing these people for over five years, and my first question to Jeremy was literally whether he was based in the US.

Their stated training approach is roughly four weeks of shadowing, joining calls, and running optimization processes before touching a client account independently.

The other question worth asking is what happens before you sign. Their process includes a campaign analysis over a 60 day period, which shows where the opportunities are and occasionally concludes that an account is already running well enough not to need them.

What the working relationship looks like

Every account gets three people: an account director, an account manager, and an account coordinator.

Accounts are reviewed roughly four to five times a week, on a rotating schedule so the same task recurs weekly with fresh data in between.

One coordinator’s week runs bid adjustments Monday and Tuesday, keyword transfers, budget checks, and negations across the remaining days.

Calls default to biweekly, adjustable by client preference. Some want weekly, some want monthly, and some prefer email updates.

My own reaction to biweekly calls was that little changes in two weeks, which is a fair criticism of my own engagement rather than of their cadence.

Why a low ACOS target can cost you money

My ACOS was around 12% when I started, which meant I was leaving money on the table.

The reason I ran it that way was availability rather than strategy. I did not check my account often, so a low ACOS target meant it could not do much damage unattended.

Pushing for growth requires closer monitoring, because more aggressive bidding produces data faster and needs faster optimization to prevent waste.

Edward’s advice for anyone managing their own account is to build a repeatable schedule. Even one hour a week at a fixed time creates a process you can hold yourself to, and the absence of that is precisely why agencies exist.

Frequently asked questions

When should you hire an Amazon PPC agency?

Roughly above $2,000 a month in ad spend, and when your time is worth more elsewhere. Below that threshold the management fee consumes whatever incremental sales the agency produces.

What results should you expect from an Amazon PPC agency?

My own account saw roughly $5,800 in additional ad spend produce about $26,000 in additional revenue after a three month settling period, which is roughly a 4.5x return on the incremental spend.

What does an agency change first?

Account structure. One campaign per product category with one ad group per SKU, which prevents impressions splitting across duplicate placements and makes it clear which search terms work for which product.

How long before an agency’s changes take effect?

About three months to transition fully. Existing campaigns get wound down gradually while new ones build, and anything still performing well after three months often stays running.

Why are my Amazon ads not converting?

Usually the listing rather than the ads. In my experience teaching sellers, over 80% of the time it is the listing. Empty back end keywords are another frequent cause, since they prevent indexing for terms you should rank for.

What are the Amazon auto campaign targeting options?

Four groups with independent bids. Close match and loose match are keyword based, while substitutes and complements target products people bought instead of or alongside yours.

How do Amazon PPC agencies charge?

Seller Labs uses flat monthly tiers based on ad spend, starting around $999 for up to $5,000 in monthly spend. Flat tiers leave room to grow within a tier without the fee rising.

How do you avoid a bad agency?

Ask where the team actually sits and what their training process is. The common failure pattern is a polished salesperson closing the deal with the work routed to inexperienced staff learning on live accounts.