Amazon agency vs freelancer vs in-house: how should sellers choose?

How do the three options actually compare?

When is a freelancer the right answer?

When the work is narrow, defined and finite. A freelance copywriter rewriting twelve listings is often better value than an agency doing the same job, because you are paying for one skill rather than a team.

Choose a freelancer if: you are under about £20k a month on Amazon, you know specifically what needs doing, you have someone internally who can brief and review, and you can tolerate the work stopping if that person becomes unavailable.

Avoid if: you need several disciplines at once, or if nobody in your business can tell good Amazon work from bad. A freelancer needs a competent client to be effective.

When does an agency make sense?

When you need four or five skills and cannot justify four or five hires. A retainer buys keyword research, copywriting, advertising, design direction and compliance monitoring from one contract.

Choose an agency if: you are between roughly £20k and £150k a month, advertising is running and needs weekly attention, you want one accountable party, and you value not carrying the recruitment risk.

Avoid if: your budget only stretches to the bottom of the market. Under about £500 a month you are usually buying one person managing many accounts to a template, and a good freelancer at the same price will do more.

When is in-house cheaper?

Later than most people think. A £40,000 salary is roughly £48,000 once employer National Insurance and pension are included — about £4,000 a month for one person's knowledge.

That is more than most agency retainers, and it buys one skill set rather than several. The advantage is that the knowledge stays in the business, which compounds.

Choose in-house if: Amazon is above roughly £150k a month, it is strategic rather than a channel, and you can give that person enough work to fill a week. Most in-house Amazon hires under that threshold end up doing something else half the time.

Is a hybrid better than any of the three?

Frequently, and it is the arrangement we see working best above about £100k a month: one in-house person who owns the account and the relationships, with specialist support bought in for advertising, listing rebuilds or launches.

The in-house person keeps the knowledge and the context. The specialists bring the pattern recognition that only comes from seeing many accounts. It costs more than either alone and usually returns more than both.

The split that works most often puts advertising and listing rebuilds outside and everything else inside, because those two are the most technical and the most time-hungry. Our PPC management and listing optimisation are sold separately for exactly this arrangement, and the pricing page shows what each costs on its own rather than only as part of a full retainer.

What are the failure modes of each option?

Each of the three fails in a predictable way, and knowing which failure you can least afford is often a better basis for the decision than the cost comparison.

Two of these are cheap to insure against and worth doing in every case. Keep Seller Central, Amazon Ads and Brand Registry in your own company's name whoever runs them — there is more on why in what an Amazon agency actually does. And keep a written record of what changed and when, because that log is the only thing that survives a handover intact.

The third risk is the one nobody quotes for: the cost of a bad month going unnoticed. A freelancer on holiday and an agency mid-handover both look the same from the outside, which is a suppressed listing or a policy warning sitting unattended. Whatever you choose, make sure someone is named as responsible for account health notifications — that is the failure that turns expensive fastest, as the suspension guide sets out.

A decision guide by monthly revenue

These are starting points, not rules. A complex catalogue at £30k a month can need more help than a simple one at £100k. If you want a read on your own account before deciding, the free audit will tell you which of the three we would actually recommend — including when the answer is none of them.

Questions people also ask

Last reviewed 13 June 2026. We update articles in place rather than republishing at a new URL, so the link you have keeps working.