What a defensible Amazon decision looks like

There is an old saying in the corporate world: nobody ever got fired for hiring IBM. It is usually said as a joke, but it captures something real about how decisions get made inside organizations. When you are responsible for a channel, a budget, or a team, the upside of a bold decision is nice, but the downside of a bad one is personal. The person who made the call is the person who owns the outcome. And in most organizations, avoiding a visible loss carries more weight than generating a moderate gain.

You want to cap your downside without capping the potential of the upside. The language is different from what you hear in a CPG brand's quarterly review, but the underlying principle is identical. You want to minimize or hedge against what can go wrong while continuing to do the things that are right.

Amazon decisions tend to resist easy defensibility. The platform is complex, the data is overwhelming, the rules change, and the person making the call often has to defend it to someone who does not have Amazon-specific expertise. Which means the quality of the decision and the ability to explain the decision are two different skills, and both matter.

 Why defensibility matters more than it sounds

Here is why it deserves more attention than it usually gets.

• Person asking, your boss, board, CEO, or investor, is usually not an Amazon expert. They are not going to evaluate your decision based on whether the ACOS threshold was technically correct or whether the keyword strategy was sound. They are just trying to figure out whether the channel is being run well, opportunities are maximized, things are taken care of. Defensibility here is the proxy for thoughtfulness. When you can explain why you made a decision, what alternatives you considered, and how you will know if it worked, the person across the table can see the logic, which increases their level of confidence (good for you!). Not because they understood every detail, but because they can see that someone is thinking clearly about it.

• Defensibility forces you to actually have a reason. This sounds obvious, but surprising number of Amazon decisions are made on instinct, copied from a competitor, or recommended by an agency without much context of YOUR business. "We should increase ad spend to drive sales" is not a reason. "We should increase ad spend on these five SKUs because their organic presence is strong, conversion is above category average, and unit economics can support a higher TACOS" is a reason. The act of making a decision defensible often improves the decision itself, because it requires you to articulate the logic before you commit the resources

• Third, and this one is easy to overlook: the person making the decision is often not the person executing it. If you are a VP of ecommerce or a director responsible for Amazon, you are probably not the one setting bids, configuring deals, or resolving catalog issues. You are defending the decisions of an agency, a contractor, or an internal team to someone above you. Defensibility is how you bridge that gap. It is the difference between "our agency says we should do this" and "here is the business logic behind what we are doing, and here is how we will measure whether it is working”

Three questions to make a defensible decision

I have found that most Amazon decisions become defensible when you can answer three questions clearly:

1. What business problem are we solving? Not what Amazon metric are we trying to improve, but what business problem. "We need to lower ACOS" is a metric goal. "We need to improve profit margin on Amazon" is a business problem, and ACOS can be a metric to measure success of solving for more profit margins.The distinction matters because a business problem opens up the full range of possible solutions, including ones that have nothing to do with advertising. I have written before about how ongoing challenges with high ACOS often have just as much to do with the product offerings and pricing as with the quality of PPC management. When you start with the business problem, you are less likely to lock yourself into a narrow solution that sounds right but does not actually address the constraint.

2. What are the alternatives, including doing nothing? Every decision exists in comparison to other options, and one of those options is always to not act. Sometimes not acting is the right call. But it’s better to be explicit about it rather than letting it happen by default. Let’s say you have a catalog of 50 SKUs on Amazon and the portfolio is underperforming. You could continue as is, accepting that a long tail of low-volume products is consuming operational attention and advertising dollars without contributing meaningfully to the bottom line. You could try to optimize all 50 and give them equal attention, which sounds thorough but spreads resources thin. You could re-allocate ad spend toward the SKUs that are already converting well and pull budget from the ones that are not. Or you could reduce the active catalog entirely, cutting the products that don’t do much in sales, and focus on the few that do well. Each of those is a different path to improving performance, with different tradeoffs in cost, complexity, and risk. Laying them out, even briefly, demonstrates that the decision was considered rather than reactive

 What to do when you do not have a defensible decision yet

Sometimes you do not have enough information to make a fully defensible call. In fact, that happens often, both on Amazon and in business: needing to make a call without complete data. So in these scenarios a mistake would be to avoid the decision entirely until you feel certain (full certainty and complete data don’t always come), OR to make a decision of ‘just pick one’ style.

Here there can be two honest options. The first is to make a smaller version of the decision that you can learn from. Instead of switching your entire catalog to FBA, you move your top three SKUs first, run it for 60 days, and evaluate the fee impact, sales lift, any operational friction before committing the rest. That is a pilot, and it is inherently defensible because it is structured as a test with defined learning objectives, not a bet.

The second option is to go forward with defined goalposts. These can be time-based: we give this 30 days to show results before we reassess. Or they can be performance-based: we continue scaling this campaign as long as TACOS stays below a defined threshold and new-to-brand rate stays above a defined floor. If either crosses, we pause. Goalposts turn an uncertain decision into a managed experiment, and a managed experiment is always defensible even if the result is not what you hoped for.

But be careful with the third option, which is not making a decision at all. I occasionally work with business owners who take risk hedging too far, who genuinely want to nail down all the uncertainty before committing to anything. The problem is that on Amazon full certainty does not exist. If you are considering switching to FBA, you can estimate fees and model the sales lift, but you cannot anticipate every ad hoc issue that will come up or know for certain how well Amazon will handle your product in their fulfillment centers. At some point you have to move forward with imperfect information, and the framework above, defining the business problem, naming the alternatives, setting the goalposts, is what makes that possible without being reckless.
Not making a decision leaves your outcomes to circumstances or inertia. And when someone above you asks why the channel has not moved, "we were still evaluating" is a harder answer to defend than "we ran a structured pilot, here is what we learned, and here is what we are doing next."

Defensibility as a practice 

Here I want to comment on defensibility potnetially sounding like a bureaucratic exercise, like something you do to cover yourself (‘CYA’). It is not. Defensibility is the discipline of thinking clearly before you act, and being honest about what you know and what you do not. It is how you stay focused on what matters on a platform that will happily drown you in metrics, notifications, and things that feel urgent. And it is how you build credibility with the people you report to, by always being able to explain your reasoning, even when things do not go right.

Minimize what can go wrong. Continue doing what is right. And be able to explain both.

If you are making Amazon decisions that need to hold up in front of leadership, and you want a thought partner who can help you think through the logic before you commit, that is the work I do. Reach out if a conversation would help.

Saludos,

Irina