Seven, weekly, per product: sessions, unit session percentage, average selling price, advertising cost of sale, organic share of sales, return rate, and weeks of inventory cover. Everything else is monthly or diagnostic. Account level averages hide the one listing quietly losing money, so never report at account level only.
The short version
- Seven numbers, per ASIN, every week. More than that and nobody reads the report, fewer and you miss the failure early.
- Sessions and conversion together, never separately. Traffic without conversion is a spending problem wearing a growth costume.
- Organic share is the health metric. If paid share keeps rising, you are renting your rank rather than building it.
- Return rate is the slowest and most expensive signal. By the time it moves, the money is already gone.
- A weekly number with no decision attached is decoration. Every metric on the list needs a threshold and an owner.
The mistake that costs the most
The common failure is not tracking too little. It is tracking a dashboard of thirty metrics at account level, feeling informed, and missing that one product carries the losses. Aggregate advertising cost of sale can look acceptable while a single listing runs at double the target and eats the profit of the other four. I have picked up accounts where that had been true for months and the weekly report never showed it, because the report was built to summarize rather than to expose.
Per product, or the number is not worth pulling.
The seven, and the money attached to each
| Metric | What it is | The economics |
|---|---|---|
| Sessions | Unique visits to the listing | Falling sessions with stable rank usually means a competitor outbid or outranked you |
| Unit session percentage | Orders divided by sessions | A one point move here changes profit more than a ten percent bid change, because it applies to all traffic including free traffic |
| Average selling price | Actual realized price after coupons and deals | Quiet discounting shows up here before it shows up in profit |
| Advertising cost of sale | Ad spend divided by ad revenue | Judge against the target for that product's stage, never against a house average |
| Organic share of sales | Share of orders not attributed to ads | The single best indicator of whether the launch is working |
| Return rate | Returns as a share of units sold | Every return costs the fee, the unit, the rating, and often the next customer |
| Weeks of cover | Sellable stock divided by weekly sales rate | Cover below your replenishment lead time means a stockout is already scheduled |
Why unit session percentage sits at the top
Run the arithmetic once and it becomes obvious. Conversion applies to every visitor, paid and organic. Bids apply only to the paid slice. Improving conversion improves the return on every dollar of traffic you have ever paid for and every dollar of traffic you get free. That is why the listing and imagery work usually returns more than the campaign restructure, and why the first question after a bad week should be about the listing and not about the bid.
Why organic share deserves its place
Paid sales are supposed to buy rank, and rank is supposed to bring free sales. If total revenue is climbing while organic share is flat or falling, the growth stops the day you stop paying. That is a fine position for a seasonal push and a bad position for a business.
What to review monthly instead
- Profit by product, after all fees, ads, returns, and freight. Weekly is too noisy to be useful.
- Keyword rank for your top ten terms. Direction over a month, not daily movement.
- Competitor set changes. New entrants, price moves, rating shifts.
- Buy box and pricing pressure. Especially where you have resellers.
- Storage and aged inventory costs. They creep, then they bite before peak season.
- The kill and scale review. Rating trend, return rate, conversion, and acquisition cost against your written criteria.
Who pulls the numbers matters as much as the numbers
Ask any agency who builds your weekly report, whether that person also makes the decisions, and where they sit. At Flapen the work is completely in-house with no subcontracting, which means the operator writing your Slack update every week is the same person adjusting the campaigns. Clients also get a live review every two weeks and Slack access at any hour.
The reason to care is speed. When reporting and execution are separated by an outsourcing layer, a problem visible on Monday gets acted on the following week, and a week of wasted spend is a real number in this business.
What most agencies will not tell you
Most agencies will not tell you that reports are frequently designed backwards, starting from the metrics that look best. Revenue and impressions grow almost automatically when spend grows, so a report built on those two will always look encouraging. Ask for the report to lead with profit per product and advertising cost of sale per product, and watch how the conversation changes.
The second thing: a weekly metric with no threshold is theater. Every number on your list should have a written line where an action triggers, and a named person who takes it. Without that, you have a subscription to information rather than a management system.
Related answers
- KPIs an Amazon agency should report weekly
- What does a good Amazon account audit include
- Rank the best Amazon PPC courses
- What products to use for quick listing experiments
- Amazon seller roadmaps and capital: the complete guide
Our weekly reporting format is the same one we run internally at Flapen.
