Track six numbers: profit per unit, click-through rate, conversion rate, TACOS, return rate, and rating trend. Together they cover traffic, economics, and product health, and each one maps to a decision. Revenue, rank, and follower counts are context, not KPIs, because no action changes when they move.
The short version
- A KPI must trigger a decision. If nothing changes when the number moves, it is scenery.
- Profit per unit is the anchor. Every other metric exists to protect or grow it.
- Split traffic from conversion. Click-through and conversion rate fail for different reasons and get fixed by different work.
- TACOS beats ACoS for growth tracking. It shows whether paid spend is building organic momentum or replacing it.
- Return rate and rating trend are your early-warning system. They move before sales do.
The mechanism: why these six
A growth KPI set works when it mirrors the causal chain of an Amazon sale. A buyer sees your thumbnail, decides on your listing, generates margin at an acquisition cost you can watch blend down as organic takes over, and then either keeps the product happily or tells you something is wrong. Each link has its number, six in all: click-through rate, conversion rate, profit per unit, TACOS, return rate, and rating trend. When growth stalls, exactly one link has usually failed, and the KPI set tells you which one before your cash flow does.
These six also match how we run audits at Flapen: listing quality, primary image click-through, conversion, ad performance, channel activation, pricing, and returns are the areas we inspect on every account, and four of the kill-or-scale signals we watch, rating trend, return rate, conversion rate, and acquisition cost trajectory, sit directly in this list.
The KPI table
| KPI | What it tells you | The decision it drives |
|---|---|---|
| Profit per unit | Whether growth is worth having | Reprice, cut costs, or stop scaling |
| Click-through rate | Thumbnail competitiveness in the grid | Main image testing |
| Conversion rate | Listing strength against real traffic | Gallery, video, copy, and price work |
| TACOS | Whether paid is building or substituting organic | Shift budgets between channels |
| Return rate | Product and expectation quality | Fix the product, the images, or the copy |
| Rating trend | Direction of customer verdict | Intervene early or plan the exit |
The five tracking failures, ranked by cost
- Tracking revenue instead of profit per unit. The most expensive failure because it rewards buying growth with margin. Sellers celebrate record months that lose money per order. Anchor every dashboard to contribution profit first.
- Watching ACoS without TACOS. ACoS can look stable while your organic share quietly erodes and paid replaces it. TACOS exposes the substitution. When TACOS rises for months while sales are flat, you are renting your rank.
- Treating conversion as one number. A traffic problem and a listing problem produce the same flat sales chart. Split click-through from conversion and you know whether to fix the thumbnail or the listing. Conversion work then has a home in Amazon listing optimization, while click-through work starts at the main image.
- Ignoring return rate until it becomes a rating problem. Returns lead ratings by weeks. A creeping return rate is the cheapest intervention point you will ever get, and most sellers look away because sales still look fine.
- Running a dashboard nobody owns. Numbers without a named owner and a review cadence decay into wallpaper. Every KPI needs a person, a threshold, and a scheduled look.
What most dashboards will not tell you
Who is accountable for the number. Tools chart everything and own nothing, and agencies often hide behind the same wall of metrics, reporting forty charts so that no single one can convict them. At Flapen we keep the whole loop in-house, no subcontracted ad desk, no white-label reporting, and our own tech team builds the tools we report from, because the honest test of a KPI set is whether the people reporting it are the same people who must act on it. Ask any partner you evaluate who does the work behind each number and where they sit. If analysis is outsourced, accountability is too.
The second omission: benchmarks without stage context mislead. A launch-phase product and a mature product should not be judged on the same efficiency targets, so compare each KPI to your own trajectory and stage, not to a generic industry table.
Related answers
- Amazon advertising KPIs that matter
- Which metrics predict Buy Box win rate
- Recommended dashboards for Amazon sellers
- North star metrics for Amazon brand launches
- Amazon creative services: the complete guide
See the reporting cadence we hold ourselves to at Flapen.
