Four things: a named owner with protected weekly hours, enough budget for the data to mean something, a reporting rhythm you can audit, and written stop rules. Tools matter less than the person. Around $1,000 a month in spend is the practical floor for meaningful optimization.
The short version
- Name the owner before you name the tools. Software does not decide what to stop.
- Protect the hours. Ads run in-house fail on calendar pressure far more often than on skill.
- Below about $1,000 a month in spend, there is not enough data for optimization to mean much.
- Write the stop rules while things are calm. You will not write them fairly in the middle of a bad month.
- Decide what you are keeping in-house. Execution, strategy, or both, because they are different commitments.
Start by writing the stop rules
Before you open a campaign manager, write down what would make you stop. Rating trend, return rate, conversion rate, and cost of customer acquisition, each measured over a defined window, with an agreed action attached. Pin it somewhere you will see it in month three.
I am starting here because I learned it the expensive way. Early on I poured money into a failing product for three months, convinced that the next round of ad changes would turn it around. They did not. The product was not being let down by its advertising, and every week I spent proving otherwise was a week of budget and attention taken from something that worked. Our scale, fix, and kill criteria came directly out of that.
In-house teams are more exposed to this than agencies, because the person running the ads is usually the person who chose the product. Written criteria are how you protect a good decision maker from their own investment.
The checklist
- A named owner. Done properly: one person, named, with this in their objectives. Not "marketing" and not the founder in the evenings once the account grows past a couple of products.
- Protected hours. Done properly: a recurring block, typically a few hours a week for a small catalog, more during a launch. Ours is 4 to 6 hours a week of client time during a launch even with us doing the work, which tells you what an unsupported in-house launch demands.
- Enough budget to learn from. Done properly: around $1,000 a month or more. There is no hard minimum to run ads, but below that level the clicks arrive too slowly for any change to be readable.
- Access and permissions in order. Done properly: user permissions granted per person from your own account, reviewed when someone leaves, never a shared login.
- A keyword and audience map. Done properly: the terms you intend to own, the terms you have conceded, and the reason for each. Without it, campaign structure becomes personal preference.
- A campaign structure standard. Done properly: naming conventions, match type logic, separation of discovery from harvest, and a negative keyword routine on a fixed schedule.
- A reporting rhythm. Done properly: a written weekly summary that a non-specialist can read, plus a deeper review every two weeks. Write it even if the audience is only you, because the act of writing catches drift.
- Contribution margin per product. Done properly: advertising judged against profit, not against advertising cost of sale alone.
- The stop rules from the section above, agreed by whoever owns the profit and loss statement.
- An escalation path. Done properly: a named outside opinion you can call when something breaks, agreed before the emergency rather than during it.
Nine and ten are the ones in-house teams skip, and they are the two that decide whether the first bad quarter is survivable.
What you do not need
You do not need enterprise software on day one. A spreadsheet, the advertising reports, and a disciplined weekly routine will outperform an expensive tool used inconsistently. You do not need an agency-grade dashboard either, because the metrics that decide anything fit on one page.
What you do need is honesty about capacity. Running ads properly for a growing catalog is a real job, and treating it as a task that fits between other tasks is how accounts drift. If the hours cannot be protected, the choice is not in-house versus agency. It is supported versus neglected.
The hybrid most sellers land on
Plenty of brands keep strategy in-house and buy execution, or the reverse. Both work if the boundary is written down. The version that fails is the one where nobody knows who changes bids, so two people change them in the same week and the account teaches you nothing.
If you go hybrid, put the keyword map, the stop rules, and the reporting rhythm in one shared document and give one person final say on structure. At Flapen, clients typically spend around two hours a month with us once onboarding is complete, and that number only works because the boundary is explicit.
What most agencies will not tell you
Running ads in-house is the right answer for a lot of sellers, particularly single product brands with steady demand and an owner who enjoys the work. An agency has no incentive to say that, and the honest version of the pitch is not that you cannot do this. It is that doing it well takes hours you may prefer to spend elsewhere, and that experience across many accounts is mostly a shortcut through mistakes you would otherwise make one at a time.
The other thing: in-house or outsourced, the hardest part is never the bidding. It is stopping. Whoever runs your ads will eventually have to recommend that a product you love should be wound down, and if that person reports to the person who chose the product, the recommendation needs written criteria behind it to survive.
Related answers
- How to scale Amazon ads without overspending
- Who to hire to run my Amazon FBA
- Amazon agency vs in-house team pros and cons
- KPIs an Amazon agency should report weekly
- Amazon agency pricing and economics: the complete guide
If you want your stop rules pressure tested by people who wrote their own the hard way, talk to Flapen.
