Choose on what they will tell you to stop doing. A partner who only owns a scale plan will keep you inside a bad product until the money runs out. Ask for written stop criteria, the research behind the product choice, and the name of whoever physically does the work.
The short version
- Written stop criteria are the single most valuable clause in a launch agreement. They are also the rarest.
- The first purchase order should be small. About 200 units and $5,000 to $10,000 buys the answer before the risk.
- Ask who does the work, by name and function. Launches fail on execution gaps far more often than on strategy.
- A launch is a seven-month project, not a campaign. Plan cash and attention across the whole arc.
- Judge the partner on the decisions they are willing to lose money on. Everything else is presentation.
The mistake that taught me this
Early on I kept a product alive for three months that was clearly not working. Every week I told myself the advertising would turn it around. I raised budgets, restructured campaigns, rewrote the copy twice, and watched the same conversion rate come back. The ads were never going to fix it. The product was wrong for the segment, and every dollar after week two bought me a slightly more expensive version of the same answer.
That episode is where our stop criteria came from, and it is why I now treat a launch partner's willingness to call a stop as the primary selection criterion rather than a footnote. Anybody can write a plan for the good case. The money is made and lost in what happens when the numbers say no and everyone in the room still wants a yes.
The checklist, with what done properly means
- Written stop criteria before launch. Done properly means named metrics with thresholds and a window: rating trend, return rate, conversion rate, and the trajectory of acquisition cost. Vague language such as we will monitor closely does not count.
- Research you can inspect. Done properly means they can show market size, growth, return rate by segment, and where the differentiation came from. The strongest source is the pattern in competitor negative reviews, not an invented feature.
- A small validation run first. Done properly means around 200 units and $5,000 to $10,000, with the option to test several products at once rather than betting everything on one.
- Named execution owners. Done properly means you know who writes the copy, who shoots the images, who builds the campaigns, and who talks to the factory. A single account manager who coordinates strangers is a different product.
- Sourcing capability, not sourcing opinions. Done properly means somebody has inspected a factory, negotiated a landed cost, and can read a specification sheet.
- A defined phase two. Done properly means scale begins only once rating, conversion rate, and acquisition cost are proven, and everyone agreed the thresholds in advance.
- Advertising targets tied to stage. Done properly means a different efficiency expectation at launch than at maturity, stated as two numbers.
- Transparent cost separation. Done properly means the fee is one line and inventory, freight, seller fees, trademark, and ad spend are separate lines you own.
- An exit you can use. Done properly means month-to-month terms, deliverables that become yours on payment, account access you can revoke, and a written handover.
The stop criteria conversation, verbatim
Bring these three questions to every candidate and write down the answers.
What would make you tell me to kill this product? You want thresholds and a window, not a philosophy.
Who makes that call, and what happens to your fee? If their revenue falls when you cut a product, say so out loud and see how they handle it. Ours is tiered by product count, so I would rather have that discussed openly than pretended away.
Show me a product you killed. Any partner who has run real launches has stopped one. The details of that decision tell you more than any success story, because success has many parents and a stop has exactly one.
The economics of stopping early
| Decision point | Typical commitment so far | What stopping costs | What continuing costs |
|---|---|---|---|
| After validation run | 200 units, $5,000 to $10,000 | The run, minus liquidation recovery | Full production and freight |
| After first reorder | Inventory plus three months of fees and spend | The remaining stock and the fees paid | Another quarter of both |
| After a year | Full launch capital, $8,000 to $15,000 for one product | The capital, plus the opportunity cost | Continued monthly losses |
The point of the table is not that stopping is good. It is that the cost of a bad decision compounds, and the only cheap moment to be wrong is the first one. A partner who front-loads risk into a full container has removed your cheap moment.
What most agencies will not tell you
Fee structures quietly discourage stopping. Most launch services are priced per product or as a percentage of what you spend, which means the honest recommendation to shut something down reduces their revenue. Nobody says this in a sales call. You should raise it yourself and watch the reaction, because the reaction is the information.
The second thing: a large share of launch failures are decided before any marketing happens, at the moment the product was chosen. If the research was thin, no amount of execution recovers it. Ask to see the research before you ask to see the marketing plan.
Related answers
- Amazon product launch timeline template
- Best launch strategy for Amazon in the US vs EU
- How to launch first product on Amazon
- Questions to ask before hiring an Amazon agency
- Done-for-you Amazon management: the complete guide
Ask us for our stop criteria in writing before you sign anything at Flapen.
