Amazon Sales Analytics: Four Trend Shapes and What Each One Means

A sales line that drops tells you something is wrong, and not much else. What narrows it down is the shape of the move, because a hard dated drop, a slow weekly decline, a spike that gives back its gains, and a dip that resembles a normal season all have different causes behind them and send you to different reports. If you name the shape first, you skip the part where you open five reports while the line keeps moving. These four cover most of what a Vendor Central account throws off, and each one comes with a likely cause and a place to start.

The overnight cliff

You can put a date on this one: ordered revenue on a product, or on a small group of them, runs normally through Tuesday, sits near zero from Wednesday on, and stays there. Demand rarely moves like that on its own, since customer interest fades over weeks and competitors take share gradually, so a vertical line almost always means something switched off. On Amazon those switches are mechanical: your offer stopped being buyable, the item ran out, or the listing stopped being served at all.

Start by pulling glance views for the affected ASINs over the same window as the sales line. Vendor glance views are recorded only while your retail offer is the featured offer, so when a third-party seller takes that spot, shoppers keep opening the detail page and none of those visits land in your numbers. That makes a glance-view drop on a single ASIN your first signal and a lost featured offer the first suspect behind it, and the detail page tells you which: if another seller is holding the buy box, that's your cause, and if your offer is still featured but views fell anyway, the listing stopped being served, so go looking for a suppression, a broken variation family, or content that was overwritten. Glance views that held steady while ordered units fell tell you the opposite, that shoppers arrived and your offer was in front of them, which puts availability ahead of buy box ownership on the list.

Date the drop precisely before you commit to an explanation: if the cliff starts on the same day a catalog change published, or the same day stock hit zero, you can stop generating theories and go verify that one.

The slow bleed

This shape has no bad day in it anywhere. Each week lands a little under the week before, every individual day looks defensible, and then two months in the cumulative total is far enough down that somebody asks what happened. There's no incident to point at, which is part of why these run so long before anyone treats them as a problem rather than as a soft market.

Price is the usual place to start. Put the product's price history beside the sales line and check whether your offer drifted above competing offers on the same ASIN. That history also tells you how long a pricing problem has been running and which seller is behind it. Watch units and revenue as separate series while you do it, because they disagree in a useful way: units sliding while your price holds flat usually means share is going to another offer, whereas revenue sliding while units hold flat means the price came down and nobody told you.

Then look at margin across the same weeks, since a bleed can eat the account without ever touching the top line. Allowances, damages, and price concessions all land below revenue, and product-level Net Pure Profit Margin is where that erosion becomes visible.

The spike that fully reverts

A short pop, then the line settles back to roughly where it started. The trap here is reading the peak as the new baseline, which turns the return to normal into a decline and quietly builds a forecast on top of a promotion.

Line the spike up against advertising spend for the identical window, then against purchase order history. Most of these resolve into one of three things. If a campaign budget or bid change starts and ends with the pop, it's an ad-spend artifact. If there's a deal or promotional period with a matching tail, that's your answer. And if a single large purchase order sits under the spike and hasn't repeated, treat it as a one-time buy and leave your demand assumptions where they were. Amazon does the same thing, since its next forecast follows what actually sold through rather than the quantity it last ordered.

The inverse is worth checking too. If spend, promotions, and PO history were all flat through a spike that later gave back its gains, you may have been picking up a competitor's stockout, and the height of that pop is a decent read on your ceiling when their offer isn't there.

The dip that looks seasonal

Every category has a couple of months that everyone on the team already has an opinion about, and that shared opinion is what makes this shape dangerous. A real decline sitting inside an expected one gets explained in a single sentence in a status meeting, and nobody opens the report.

So test the assumption rather than trusting it. Compare the same weeks against last year at the product level, since a portfolio view can look perfectly seasonal while two ASINs inside it are in real trouble. Then compare Amazon's demand forecast for those products against what you actually shipped. A genuine seasonal dip shows up in both places, in the prior-year shape and in the forecast, because the forecast reflects real customer demand patterns and Amazon saw the same pattern you did. When the prior year has no dip in it and the forecast never stepped down, you're looking at something else, and seasonality was only the first explanation that fit.

Shape, cause, and first report

Two habits make this hold up. The first is dating the change before naming the cause, since a precise start date rules out most of the candidates on its own. The second is reading shapes at the product level. An account-wide sales line is several of these averaged together, and an average doesn't have a single cause behind it, so the shape you think you're looking at may not exist anywhere in the catalog. Once you know which shape you actually have, your weekly sales review can carry a specific question into the reports instead of a general one.

Where InsightLeap fits

Each of these shapes lines up with something InsightLeap already tracks. For the cliff, the daily catalog audit flags lost buy boxes, replenishment code changes, broken variations, and content changes, with third-party offers, buy box ownership, MAP and price violations, and Best Seller Rank monitored alongside it. For the bleed, historic price changes are recorded per product, so you can see how long a pricing problem has been running and which seller is behind it, and product-level Net PPM shows margin erosion from allowances, damages, and price moves. For the spike, advertising campaign performance is correlated with product and inventory data over the same window. And for the seasonal question, demand forecast history is kept per product and compared against what actually shipped. The vendor analytics page covers the rest.

Name the shape before you open anything

When the number moves, put the line on screen and name the shape first: cliff, bleed, spike, or seasonal-looking dip. Each name cuts the search down to one or two reports and one question to ask of them. If you'd rather have those four checks already running the next time a line moves, book a demo and bring a week you had to explain after the fact.