A price history that looks like a saw blade — up eleven cents, down eleven cents, up again, every few minutes, for hours — is not a volatile market. It is a feedback loop, and loops have specific causes that you can identify from the shape of the oscillation.
They are also worth fixing properly rather than suppressing, because a bouncing price costs you more than it appears to: it burns through price-change capacity, makes your reporting meaningless, and on some listings it can trip the guardrails that remove your offer entirely.
Loop Type 1: Two Systems with Authority Over One SKU
Check this first, every time. It is the most common cause and the easiest to rule out.
If two systems can write prices to the same SKU, each will submit its preferred price, see the other’s price, and correct. The oscillation is perfectly regular because it is two machines disagreeing on a fixed cycle.
Candidates: Amazon’s own Automate Pricing still enabled on SKUs you migrated to a third-party tool; a second repricing tool left connected after a switch; a channel manager or listing tool that writes price as part of inventory sync; a scheduled bulk upload running against a SKU under automated control.
Shape: metronomic. Two prices, alternating, at a near-constant interval. The two values barely change over hours.
Fix: one system per SKU. Detach the others rather than pausing them, because paused rules get re-enabled by someone who does not know why they were paused.
Loop Type 2: Mutual Undercutting with a Competitor
Two sellers each configured to undercut the other, both automated, both fast.
You go a cent below them. They go a cent below you. Your rule reads their new price and responds. The cycle repeats until one of you hits a floor — and then the price sits at that floor, which is how a bouncing price becomes a stuck price.
Shape: a downward staircase rather than a flat oscillation. Prices trend lower over the sequence, with each step small and the interval short.
Fix: stop referencing what they reference. If both of you are pricing against the lowest offer on the listing, you have built a machine with no stable state. Reference the Featured Offer price instead — which is the price Amazon’s ranking has concluded is working — and position relative to it based on your delivery and performance rather than relative to one competitor’s move.
Also check your floor. A staircase that ends in a stuck price has told you something about whether that floor reflects contribution or gross margin.
Loop Type 3: Conflicting Rules Inside One Tool
A single SKU falling under two of your own rules with incompatible targets, or inside two overlapping SKU groups.
This arises from accumulation: rules get added over months, group definitions overlap, and nobody audits the whole set. One rule wants $28.90 because it is competing against the Featured Offer. Another wants $27.40 because it is a clearance rule applied to a group that still includes this SKU.
Shape: oscillation between two specific prices that correspond to two of your own rule targets. The giveaway is that both prices are explicable from your own configuration.
Fix: audit rule membership. Most tools can report which rules govern a given SKU — run it for any bouncing SKU. Remove the overlap rather than adjusting the rules, because adjusting leaves the structural conflict in place for the next SKU that lands in both groups.
Loop Type 4: Oscillation Around a Boundary
Your price is sitting at the edge of a constraint — a floor, a ceiling, or a threshold that changes the economics.
The rule wants to go lower, the floor blocks it, something in the evaluation moves slightly, and the price steps across the boundary and back. On Amazon there is a specific version of this worth knowing about: fulfilment fees for standard-size items are banded by price, so a price crossing a band boundary changes the fee per unit. A rule optimising on price alone, oscillating across a band edge, is producing genuinely different unit economics on each side without knowing it.
Shape: oscillation around a single known value — your minimum, your maximum, or a round price point.
Fix: add a deadband. Most tools support a minimum change threshold, so the price will not move unless the required change exceeds some amount. Setting it to a meaningful figure rather than a cent eliminates boundary jitter and costs you nothing, because sub-cent competitive positioning is not winning you anything.
Why to Fix it Rather Than Tolerate it
Three real costs.
Price-change capacity. Amazon applies a daily allotment on price updates. Worth noting that changes made by Amazon’s own Automate Pricing tool do not count toward it, but a loop running through a third-party integration can consume capacity you would rather spend on actual decisions.
Reporting becomes noise. Any analysis of how price affects your sales is destroyed by a price that was never stable long enough to measure.
Guardrail exposure. Erratic pricing is more likely to clip a threshold — a potential pricing error, a suppressed buy button — than a stable price in the same range. A loop with a wide amplitude is sampling the edges of what Amazon will accept.
The Diagnostic Order
1. List every system with write access to your prices. Resolve any duplicates before investigating anything else.
2. Look at the shape of the oscillation. Metronomic between two values, downward staircase, two explicable rule targets, or jitter around a known boundary. The shape names the loop type.
3. Check which rules govern the SKU in your own tool.
4. Check your floor and ceiling against the oscillation range.
5. Add a deadband as the general-purpose damper once the structural cause is addressed.
The Setting Worth having Regardless
A minimum change threshold is the cheapest insurance in repricing configuration. It prevents boundary jitter, reduces mutual undercutting velocity, preserves price-change capacity, and makes your price history readable.
The reason sellers do not set it is a belief that cent-level precision wins placement. On most listings it does not — a cent is not what separates offers, and the sellers holding position at higher prices are doing it on delivery and performance rather than on the last cent.
The research team at PricingMag has documented all four loop types for sellers seeing a price bouncing up and down every few minutes, with the oscillation shapes mapped to causes.
A saw-blade price chart is a system talking to itself. Find the second voice.
