The worst thing about a disconnected repricer is how quiet it is. Nothing breaks. No listing goes down, no order fails, no alert fires. Your prices simply stop moving, and because a static price looks exactly like a stable market, you can run for days or weeks without noticing.
What you lose during that window is invisible too: position on contested ASINs as competitors move around you, and margin on the listings where competition thinned and your price never recovered.
Here is how to detect it, recover without creating a second problem, and make sure the next one announces itself.
How a Disconnection Actually Presents
Four symptoms, none of which looks like a failure:
- Prices identical for days on SKUs that normally move
- A connection status that still shows as connected — because the integration exists; it is the authorisation or the submission queue that has stopped
- Your tool displaying prices that do not match the live offer
- Featured Offer share declining gradually with no competitive explanation in your own records
The second is the one that misleads people. Connected and authorised are different states, and most interfaces show only the first.
The Five Causes, in Order of Likelihood
1. Authorisation lapsed. Marketplace integrations use tokens that expire or get revoked. A password change, a security event, or a routine expiry can end the authorisation while leaving the connection object in place.
2. A permission was removed. If your tool connects through a user account with specific permissions, a change to those permissions — often made for an unrelated reason, like tidying up team access — silently removes its ability to submit prices.
3. The submission queue stalled. The connection is live, the tool is evaluating, and submissions are not landing. This is the version most likely to show a healthy status indicator.
4. Plan or account state changed. Automated pricing depends on an active Professional selling plan. Amazon’s own Automate Pricing is included with that plan, and third-party tools rely on the same account standing. A billing lapse or an account issue takes pricing automation down alongside everything else.
5. A platform-side fault. Real, and least common. Check your tool’s status page before assuming it, but do not assume it first.
Detecting it: The One Number to Watch
Time Since Last Successful Price Submission
Not connection status. Not last login. Not price changes made today. The timestamp of the last submission that Amazon accepted.
Every repricing tool exposes this somewhere, and almost nobody looks at it. If your normal cycle is every fifteen minutes and that timestamp is eleven hours old, you have been frozen for eleven hours regardless of what any status indicator says.
Add it to whatever you check daily. It is the single most valuable number in repricing operations and it takes four seconds to read.
Recovery, in the Right Order
1. Confirm the freeze before fixing anything. Compare your tool’s price against the live offer price for five SKUs, viewed as a logged-out shopper. If they match, your prices are not frozen and you have a different problem — possibly a stable market.
2. Re-authorise rather than reconnect. Where the tool offers it, refreshing the authorisation preserves your configuration. A full disconnect-and-reconnect can reset rule assignments, and rebuilding those is how a thirty-minute recovery becomes a day.
3. Check your floors before letting automation resume. This is the step that prevents the second incident.
Your prices have been frozen while the market moved. When automation resumes, it will evaluate the current competitive picture and act immediately — possibly with a large move. If your floors are stale, that move goes somewhere you did not authorise. Floors built before Amazon’s 2026 fee changes, which took effect for most sellers on 15 January 2026, or before the 3.5% fuel and logistics surcharge added to FBA fulfilment fees in the US and Canada from 17 April 2026, are understated.
Verify them, particularly on SKUs where the market may have moved down during the freeze.
4. Resume in batches. Ten to twenty SKUs first, watched through one full cycle, then wider. A configuration problem revealed on resumption is much cheaper discovered on twenty SKUs than on two thousand.
5. Reconcile what the freeze cost. For each affected SKU, compare your frozen price against where the competitive price went. Two populations emerge: SKUs where you were left too high and lost sales, and SKUs where you were left too low and sold at a discount you no longer needed. The second group is the one worth noting, because it usually surprises people.
Preventing the Silent Version
Three controls, in order of value:
A sync-age alert. If the last successful submission exceeds a threshold — two cycles, or an hour, whatever suits your cadence — someone should be told. If your tool cannot do this, a daily manual check of that one timestamp is a reasonable substitute.
A live-price reconciliation. Periodically compare your tool’s prices against live offer prices for a sample of SKUs. This catches the stalled-queue case that status indicators miss.
A dedicated integration user with documented permissions. Connect through an account whose permissions nobody will tidy up by accident, and write down which permissions it needs. A surprising share of these incidents are caused by routine team access housekeeping.
Why This Matters More Than it Appears
A frozen price is not neutral. On a contested ASIN you are drifting out of the rotation as competitors move, and on a thinning ASIN you are sitting at a discount the market no longer requires. Both are losses, and neither appears in any report as a problem — they appear as slightly lower sales and slightly lower margin, which is indistinguishable from a quiet week.
The sellers who catch these quickly are not better at troubleshooting. They are watching one timestamp.
The research team at PricingMag has documented the full recovery in PricingMag’s diagnosis path for a disconnected repricer, including the floor verification that should happen before automation resumes and how to reconcile the cost of the freeze.
Automation fails loudly when it breaks and quietly when it stops. The quiet version is the expensive one.
I am a proud University of Wisconsin-Madison (BA, Political Science) alumni and “pop” culture addict. My interests include music, films, books (including comics) and vintage stereo equipment. My passion for all forms of entertainment from mainstream to obscure takes up too much room. In addition to skateboarding and Great Lakes surfing, I love good food and micro brews of all types.