Track four leading indicators from the first week: conversion rate, click-through on the primary image, advertising cost of sale, and return rate. Revenue is the lagging number. Set each target as a direction of travel across a defined window, and decide in advance what result would make you stop.
The short version
- Leading indicators first. Revenue tells you what already happened, and by then the decision window has closed.
- Targets are trends, not absolutes. Your category norm is the baseline, and improvement against it is the target.
- Each thirty-day block has one primary question. Answering three questions at once produces three unclear answers.
- Write the stopping rule on day one. A criterion decided after you are emotionally invested is not a criterion.
- Return rate is the metric sellers ignore until it is expensive. It is the earliest signal of a product problem.
The mistake that made me write kill criteria down
I once poured money into a failing product for three months, convinced the advertising would turn it around. It did not. Every week there was a reason to give it one more week, and every week the reason was slightly weaker than the last. The money is gone and the lesson is the only return I got: define what failure looks like before you are the person who has to admit it.
That is what these ninety days are for. Not to hit an arbitrary revenue figure, but to answer a specific question honestly, on a schedule, with numbers you agreed to in advance.
The three blocks and what each one is for
| Window | The question it answers | Primary metrics | What good looks like |
|---|---|---|---|
| Day 1 to 30 | Is the page capable of converting? | Sessions, conversion rate, primary image click-through | Conversion rate stabilizes at or above the category norm |
| Day 31 to 60 | Can we acquire customers at a sane cost? | Advertising cost of sale by keyword, orders per day, review velocity | Efficiency improving week on week on the terms that convert |
| Day 61 to 90 | Is any of this durable without paid support? | Organic share of orders, rank on target phrases, return rate, repeat purchases | Organic share rising while efficiency holds |
Judge each block on its own question. A poor advertising cost of sale in the first thirty days is not necessarily a failure, because you are buying data. The same number at day eighty, with organic share flat, is a serious signal.
The failure modes, in order of what they cost
- Measuring revenue and nothing else. The most expensive habit on this list. Revenue can be bought with spend right up until the cash runs out, and it hides every underlying problem until the last possible moment.
- No baseline. If you did not record conversion rate, click-through, and return rate in week one, you cannot tell improvement from noise for the whole quarter.
- Reading data windows too short to mean anything. Daily checking of a low-volume product produces panic and reversals. Fix the window length in advance and hold to it.
- Changing several variables at once. A new image, a price cut, and a campaign restructure in the same week means the quarter teaches you nothing.
- Ignoring the return rate. Returns rise before the reviews arrive. A climbing return rate at day forty is a product conversation, not a marketing one.
- Moving the criteria to protect the decision. The one that cost me three months. Written criteria, agreed early, reviewed on a date.
The criteria that decide scale, fix, or kill
At the end of each window, a product gets one of three verdicts. Scale means the numbers are proven and the constraint is inventory or budget. Fix means one identified thing is wrong and there is a specific plan with a date. Kill means the evidence says stop.
Four inputs drive that verdict: rating trend, return rate, conversion rate, and the trajectory of customer acquisition cost, each measured over a defined window rather than at a single point. Note the word trajectory. A high acquisition cost that is falling steadily is a different situation from a moderate one that has stopped improving.
When you interview any agency, ask what would make them tell you to stop. A partner who has never recommended killing a product either has not been doing this long, or has an incentive not to. At Flapen the fee does not change when we recommend killing a product, which is the only reason that recommendation is worth anything.
What most agencies will not tell you
The first ninety days of reporting are the easiest place in this business to look competent without being useful. Impressions, clicks, and spend all rise when you turn on a budget, and a dashboard full of rising lines feels like progress.
Ask for the four leading indicators instead, weekly, in writing, against the baseline recorded in week one. We send a written update every week and hold a live review every second week, and the written part matters because it creates a record you can hold anyone to, including us. If a candidate agency only reports in calls, you will end month three with no way to reconstruct what was actually claimed in month one.
Related answers
- KPIs an Amazon agency should report weekly
- First 90 days Amazon launch plan
- How to test product-market fit on Amazon
- Top strategies for Amazon PPC in first 90 days
- Amazon launch services: the complete guide
If you want your first ninety days mapped to these four indicators before you start, that is what the free audit at Flapen produces.
