The problem / 01
Reimbursements are one job. They are not the job.
The money Amazon owes you is the easy half, and it is the half every tool sells. The expensive half is the listing that went down on Tuesday and is still down on Friday, the variation family somebody else rebuilt, the Buy Box you keep not holding, and the Brand Registry account that is no longer entirely yours. All of it costs money every single day it lasts, and all of it is fixed the same way — by opening the right case and arguing it properly.
Money Amazon owes you
Inventory lost inbound, lost or damaged in the warehouse, returns never returned, and fees charged on the wrong dimensions.
Listings going down
Search-suppressed, attribute-suppressed, closed, deleted, restricted or under review — and the stock stranded behind them.
Who is allowed to edit your listing
Split reviews, detached children, and copy that keeps reverting for a reason almost no brand ever discovers.
Competitor attacks and Brand Registry
Hijacked listings, malicious reporting, and Brand Registry itself taken over — detected early and defended.
Review integrity
Reviews that breach Amazon's community guidelines, reported with the evidence Amazon acts on. Nothing beyond that.
Compliance, ungating and account health
Documentation demands, policy violations, category gates and anything quietly counting down against your account.
Every number on this page is either from our own database, labelled with the date it was queried, or from our own operating notes, labelled as such. We do not name clients, we do not guarantee outcomes, and where we have used something we did not discover ourselves we say whose it is.
Class 01 · what you are owed / 02
Six months, 53 accounts, $803,457 — after the reversals.
Everyone in reimbursement recovery quotes gross. Gross is the number before Amazon takes some of it back, which makes it the number you cannot bank. Here is ours both ways, with the subtraction shown, because a seller who has run this work before will trust the page more for it.
Why the reversals matter. Amazon grants a reimbursement and then, sometimes weeks later, reverses it — because the unit turned up, because the claim was re-assessed, because a duplicate was caught. Roughly 17.5% of what was granted came back off. Netting it out is the only honest way to state the figure, and it is why we track reversals separately rather than banking the announcement.
Where the gross came from — the five largest sources
| Source | Recovered (gross) | Items | Relative size |
|---|---|---|---|
| Customer returns never returned | $508,269 | 19,200 | |
| Inventory lost inbound | $215,676 | 1,271 | |
| Customer service issues | $102,858 | 4,202 | |
| Damaged in the warehouse | $83,426 | 5,594 | |
| Lost in the warehouse | $62,818 | 5,233 |
The box sets the fee. Not the product.
This is the clearest illustration of "fees charged on the wrong dimensions" we have. A unit from our own operating notes weighs just over a pound. It ships in a box that is very nearly a cube. Dimensional weight is calculated from the box, and dimensional weight is what gets billed — rounded up.
A product that weighs 1.119 lb is billed as though it weighs 2.00 lb, on every unit, forever — and the packaging is the only reason. Sometimes the right answer is a fee case, because the dimensions Amazon holds are simply wrong. Sometimes the right answer is a smaller box, because the dimensions are correct and the packaging is the problem. Dr. Shield tells you which, with the arithmetic attached, rather than disputing a measurement that turns out to be accurate.
The same arithmetic applies inbound. Splitting a shipment across the placements Amazon actually wants, rather than accepting the default, has taken an inbound placement fee of $5,000–7,000 down to roughly $1,000 in our own operating experience — a decision made before the boxes move, not a case argued afterwards.
Class 02 · listings going down / 03
A suppressed listing does not send you a bill. It just stops earning.
This is the expensive one. A reimbursement claim is money you did not have and will get back eventually. A listing out of search is revenue you are losing right now, every hour, while the storage fees on the stock behind it keep running. Here is what that looks like across the accounts we watch — not a projection, a snapshot.
And every reason is a listing problem. Across 13 accounts, this is inventory stranded behind a listing that was closed, a listing error, a restricted ASIN, inventory under review, or a listing that was deleted. None of it is an inventory problem. You already paid for these units, they are already in the building, and Amazon is charging you storage on them for as long as the listing stays broken. Fixing the listing is the whole intervention.
How it usually goes
- Nobody notices for days, because nothing announces it — sales just soften
- The seller checks the detail page, which looks completely fine, because the cause is behind it
- A case is opened, gets a templated reply, and goes quiet
- A second case is opened because the first is slow, and both get closed as duplicates
- The stock accrues storage the entire time
What Dr. Shield does
- Watches suppression, stranding and attribute status on every ASIN daily, so the clock starts on day one
- Diagnoses the actual trigger — the field, the phrase, the classification — rather than guessing at the page
- Corrects what needs correcting, then opens one case with the before-and-after attached
- Runs it on the three-day call cadence, every call documented, until somebody answers it
- Puts a number on what the suppression is costing per day, so it stops competing with a spreadsheet for attention
One phrase can take a listing out of search
A product in one of our accounts was miscategorised as adult. Its organic rank cratered and stayed down — not because of anything on the detail page a seller would think to check, but because of a classification the seller could not see. Finding the trigger was the entire job; the case was straightforward once we could name it.
A month-long reinstatement fight, won
One listing took a month of continuous case work to get back — parallel tracks, a call every three days, every call documented, one case ID held the whole way, and internal review rather than a fresh appeal when it was refused. It came back. We cannot promise the next one will, because Amazon decides. We can promise it gets fought like that one.
Class 03 · variation control / 04
If your copy keeps reverting, here is why.
Split reviews. Children detached from their parent. A variation family quietly rebuilt by somebody who is not you. And underneath most of it, the single most useful thing a brand can learn about how Amazon listings actually work — and one almost nobody is ever told.
Amazon ranks everyone who may contribute data to a listing on a contribution score from 0 to 100. When two sources disagree about a title, a bullet or an attribute, the higher score wins — automatically, silently, and without telling you.
52 versus 52.1
Brand Registry is beaten by Amazon's own automated data augmentor by one tenth of a point. That is the whole explanation for the most maddening experience in this business: you fix the bullet, it is correct for two days, then it is not, and support tells you the listing looks fine — because to them it does. Nobody attacked you. Something with a marginally higher score simply overwrote you, silently, again.
Diagnose before you argue
Watches your listing content for reversion, identifies whether the change came from automation, a vendor, another contributor or Amazon itself, and picks the route that actually works for that source. An argument aimed at the wrong contributor cannot be won, no matter how well it is written — which is why most brands lose this one for months.
What breaks in a variation family
- Split reviews — a child detaches and takes its review count with it, and the survivor looks brand new
- Children detached from the parent, so the family stops sharing traffic and social proof
- A family rebuilt by someone else, with your ASINs arranged the way they wanted them
- Variation themes and attributes overwritten, so the family no longer displays as one product
What he does about it
- Monitors the structure of every variation family and flags detachment the day it happens
- Reassembles the family through Amazon's own process, with the correct parentage evidenced
- Argues review reattachment where the merge is legitimate and documented
- Tracks the contribution source afterwards, so the same reversion does not quietly happen again
Class 04 · attacks and Brand Registry / 05
Some of it is not Amazon's fault. It is somebody's plan.
Hijacked listings, malicious reporting, and Brand Registry accounts taken over by people who are not the brand. This section is about defending — spotting it early, evidencing it properly, and getting control back. We do not describe how any of it is done and we do not do it to anyone, which is not a marketing position; it is the difference between a partner and a liability.
Across 393 ASINs and 19 brands over 30 days. Not every lost Buy Box is an attack — pricing, stock and eligibility all cost you it honestly. But you cannot tell the difference between a pricing problem and a hijack if nobody is looking, and a quarter of the time somebody else was in the box on products these brands own. The first job is simply knowing, continuously, rather than discovering it when the sales report arrives.
What brands come to us with
- Someone else in the Buy Box on a product they own, and no idea how long it has been going on
- A sudden violation or complaint against a listing that has been clean for years
- A Brand Registry account they no longer fully control, with an unfamiliar name holding the keys
- Reviews that read like they were written to trigger something rather than to describe a product
- The strong feeling that this is deliberate, and no evidence anyone at Amazon will act on
What defending it looks like
- Continuous Buy Box and offer monitoring per ASIN, so the timeline exists before you need it
- Every anomaly evidenced and dated as it happens — the timeline is the case
- Brand Registry rescue: proving ownership, removing the wrong hands, restoring the right ones
- Violations and complaints answered with documentation rather than protest
- Rule-breaking reviews reported through the proper channel, with the guideline cited — and nothing beyond that
$3,500 just to get Brand Registry back
A brand doing millions on Amazon engaged us on a flat $3,500 fee for one thing only: rescuing a hijacked Brand Registry. Not advertising, not listings — recovering control of the account that gates every protection tool the brand is entitled to. If you are wondering whether this problem is real and whether it warrants a specialist, that engagement is the answer.
“Predatory… trying to get our listing down, maybe leaving these reviews to trigger Amazon.”A Full Circle client, anonymised
We quote this because the feeling is real and most vendors treat it as paranoia. It is not. What we do about it is evidence and process — a documented timeline, the right case, the right channel. We never retaliate, and we never do to anyone else what was done to them.
We defend. We do not attack.
Dr. Shield will not report a competitor to get them removed, will not file complaints we do not believe are true, and will not touch anybody's reviews but the rule-breaking ones on your own listings. We will not explain to you how these attacks are carried out either — including on this page, and including where it would make better marketing.
The reason is self-interested as well as principled. Every one of those tactics is against Amazon's terms, and the account that pays for them is yours, not ours. A partner who will run them for you is a partner who will eventually cost you the business they were hired to protect.
Class 05 · review integrity / 06
We take down the reviews that break the rules. Not the ones that hurt.
This is the part of the category built on an implied promise nobody can keep. So here is ours, stated precisely enough that you can hold us to it: some reviews breach Amazon's community guidelines and can be reported with the right evidence through the right channel. A review that is simply a bad opinion of your product is not one of them, and never will be.
Reportable — breaches the guidelines
- Profanity and abuse, including harassment aimed at a person
- Content about shipping or the seller rather than the product itself
- Promotional spam, including reviews pointing readers at a competitor
- Disclosed personal data — names, addresses, contact details, order information
- Incentivised reviews and undisclosed conflicts of interest
- False factual claims about the product — statements of fact that are demonstrably untrue, not opinions you disagree with
Not reportable — and we will say so
- A negative review that is accurate
- A negative review that is unfair but honestly held
- A one-star rating with no text
- A review from someone who simply did not like the product
- A review that is costing you money and breaks no rule at all
A negative review is not removable. We will never imply otherwise.
We do not suppress, gate, bury, buy off or "handle" honest criticism, and no page of ours will ever suggest we can. Anything beyond reporting genuine guideline breaches is review manipulation — banned outright by Amazon, and a risk to your account long before it is a risk to ours.
“We get the reviews that break Amazon's rules taken down, and we tell you straight when one doesn't.”
That claim is smaller than the one our competitors imply. It is also the only one either of us can actually keep, which makes it worth considerably more.
The channel decides the reception
The same report lands differently depending on how it arrives. Reports made through the Brand Registry portal get priority handling, and Contact the Buyer is Brand-Registry-only — so a brand that is not enrolled is arguing the same case from a worse position with fewer tools. Getting you enrolled, and keeping the enrolment in your own hands, is part of the job.
An answer, not a shrug
Every review you flag is assessed against the actual guidelines and comes back with a verdict: which rule it breaches and what we will file, or why it breaches nothing and will not come down. A straight no on day one is worth more than three weeks of a vendor being encouraging about a review that was never going anywhere.
Class 06 · compliance and account health / 07
The cases where the clock is the opponent.
A documentation demand that appears overnight. A policy violation attached to a listing that has been clean for years. A category you are suddenly not allowed to sell in. None of these resolve themselves, all of them are winnable, and every one is decided by whether somebody kept working it after the first templated refusal.
Thirty-five days, and then it was gone
A violation that took 35 days of continuous case work to remove. Not 35 days of waiting — parallel tracks, a call every three days, each one documented, one case ID held throughout, and internal review instead of a fresh appeal when it came back refused. Most sellers stop somewhere around day nine, which is roughly when the queue expects you to.
An ungating that won back a client
A client had paused with us. The thing that brought them back was getting them ungated in a category they had been locked out of — documentation assembled properly, the case argued rather than merely submitted, and followed until somebody with the authority to open the gate actually looked at it. Compliance work is unglamorous right up until it is the reason the business can sell something again.
The case craft / 08
This is what "thousands of cases" actually means.
Every section above ends the same way: somebody has to open a case and win it. Dr. Shield is built by Full Circle, an Amazon agency with $500M+ in managed spend across 100+ brands, and this is the part of that experience worth having. Six habits, learned by losing cases that should have been won. They are not clever and they are not secret — they are simply what a support queue built to close tickets responds to.
Run three tracks at once
An account-health case, a support case, and an email to the specialist team — opened in parallel, on the same issue, carrying the same evidence.
WhyEach queue is staffed and escalated differently, and you cannot tell in advance which one will land on somebody who reads the attachment. Running them in sequence costs you weeks per attempt. Running them together costs you nothing.
Call every three days
Not when you remember. Not when the client asks. Every three days, on a schedule, until the case is answered or closed with a reason.
WhyA case with no contact on it ages into the background and is closed for inactivity. Three days is short enough that it never goes quiet and long enough that the case has actually moved between calls.
Make the rep document the call
Before the call ends, the representative writes what was discussed and what was agreed into the case notes. Then you read it back to them.
WhyThe next person to open the case sees the notes, not the conversation. An undocumented call did not happen, and a case with five undocumented calls looks identical to a case nobody has touched.
Keep ONE case ID
One issue, one case, followed the whole way. Never a fresh case because the old one is slow, and never three cases on the same problem hoping one lands.
WhyDuplicates get merged or closed as duplicates, and the history — every note, every attachment, every escalation you earned — goes with them. The accumulated record on one case ID is the entire asset.
Ask for internal review, not a re-appeal
When a case is refused, the instinct is to appeal again. Ask for the decision to be reviewed internally instead, on the record already filed.
WhyRe-appealing restarts the clock. It re-enters the queue at the back, with a new reviewer and none of the momentum. Internal review keeps the case where it is and puts it in front of somebody senior to the person who refused it.
Dig until you find the actual cause
Most refusals are not policy decisions. They are somebody's mistake — an agent who attached the wrong ASIN, a document filed against the wrong case, a note describing a different issue entirely.
WhyYou cannot argue a refusal you have not diagnosed. Once you can point at the specific error, the case stops being an opinion about your evidence and becomes a correction, which is a far easier thing to win.
A person can do this for a handful of cases. Nobody can do it for a hundred, every three days, for months, without something slipping — which is the entire argument for having an agent do it. The craft is ours. The stamina is his.
What he watches, every day
Refunds against physical returns
Inbound receipt discrepancies
Warehouse dispositions
Reimbursement reversals
Fee and size-tier accuracy
Suppression and stranding
Listing content reversion
Variation family structure
Buy Box and offer changes
Review guideline breaches
Compliance and account health
Open case age
The console / 09
This is what he actually hands you.
Not a count of open tickets. A written case update and a stack of specific filings — each with the evidence behind it, what it is claiming, and what happens if Amazon disagrees. In human-approval mode nothing is filed until you click, and anything touching a claim, a compliance requirement or a legal assertion comes to you at every level.
Correct the classification behind the suppression on SKU-7702, then argue the reinstatement
- Diagnosis
- The listing is out of search because of a back-end classification, not a detail-page problem. Nothing visible on the page is wrong, which is why the seller-side checks came back clean and the drop looked inexplicable.
- Cost of delay
- Units behind this listing are unsellable while it is suppressed and are accruing storage the whole time. This is the most expensive item in the queue, which is why it is at the top of it.
- Plan
- Correct the field, then open one case with the correction and the before-and-after attached — run alongside an account-health case and a specialist email, on the same evidence.
- If refused
- Internal review on the same case ID, three-day call cadence, every call documented. Amazon decides reinstatement — we decide how well the case is argued.
Your bullet copy on PARENT-3390 has reverted twice — the source outranks Brand Registry
- Diagnosis
- The copy is being overwritten by a contributor with a higher contribution score than Brand Registry. Nobody attacked the listing and nobody on your team changed it — it is losing a ranking contest it is never told it is in.
- Route
- Because of the source, a standard listing edit will simply revert again. The case argues the correct attribute values at the level that actually sets them, with the reversion history attached as evidence.
- If refused
- The copy stays as it currently is and we tell you it is not going to hold, rather than re-editing it every fortnight and invoicing you for the activity.
Open one grouped case on the refunded orders where the unit never came back
- Evidence
- Refunds issued and closed with no matching physical return receipt against any of them, clustered in one facility over a two-week span. Order IDs, refund dates and return-tracking status attached.
- Claiming
- Reimbursement for units refunded and not returned, under Amazon's own policy, inside Amazon's stated window. No amount is asserted — the claim is for the units, and the value is whatever Amazon's schedule returns.
- If refused
- If Amazon shows the units were received, the case closes and we stop flagging them. If it is refused without an explanation, internal review on the same case ID — not a re-appeal.
Two reviews flagged. One breaches the guidelines. One does not, and will not come down.
- Reportable
- A review whose text is entirely about a courier delay, with no assessment of the product at all. That sits squarely inside the guideline on content about shipping rather than the item, and the report cites it.
- Not reportable
- The two-star review you asked about breaks no rule. It is a genuine opinion, honestly held, and unflattering. We are not going to file on it, and we would rather tell you today than be encouraging about it for three weeks.
- Note
- Filed through the Brand Registry portal, where these get priority handling. Amazon decides — a correct report is not a removal, it is a report that deserves one.
Your call / 10
You choose how much he does on his own.
"Autonomous" is not one setting, and it should never be the vendor's decision — least of all when the decisions are filings made in your name to a platform that can suspend you. Dr. Shield runs at whichever of these three levels you pick, and you can change it at any time. One thing does not move with the dial: anything touching a claim, a compliance requirement or a legal assertion comes to a human at every level, including Level 3.
Full human approval
- Every case, appeal, report, re-measure request and listing correction arrives as a card
- Each card carries the evidence, what it claims, and what happens if Amazon disagrees
- Approve, reject, or approve in bulk — one click each
- Rejections are read back, so he stops proposing what you keep refusing
Supervised autonomy
- Runs on its own: chasing open cases on the three-day cadence, documenting calls, re-measure requests, internal-review requests on existing case IDs, variation-structure repairs
- Queues for you: every new claim, every appeal, every review report, anything compliance-adjacent or legal
- You set the band — which case types, which ASINs, which value thresholds
- Daily digest of what ran on its own, with the same evidence attached
Fully autonomous
- Every case is still written down with its evidence and its reasoning
- Claims, compliance and legal assertions still come to a human — that is not a permission you can switch on
- Guardrails are yours: which case types, which ASINs are untouchable, which value ceilings
- Weekly written case update; drop back to Level 1 at any moment
What does not change between levels
The same evidence trail
The same standard for filing
The same human gate
The same switch, both ways
Honesty / 11
What he does alone — and what he never does.
Everyone in this category says "autonomous" and nobody defines it. So here is ours, in public. The right-hand column applies at every approval level, including Level 3 — these are not permissions you can switch on.
Runs unattended
- Reconciles refunds against physical returns, and shipped against received, every day
- Reads Amazon's own disposition reports line by line and matches them to reimbursements already granted
- Tracks reversals separately, so the net figure is always the one on the screen
- Reconciles charged fees against the dimensions and size tiers on file
- Watches suppression, stranding, attribute validity and listing-content reversion on every ASIN
- Monitors variation structure and Buy Box ownership, building the timeline before anyone needs it
- Screens new reviews against the actual community guidelines
- Chases every open case on the three-day cadence and gets each call documented
- Keeps one case ID per issue and requests internal review rather than re-appealing
- Writes the weekly case update with the reasoning attached
Always needs a human
- Filing any claim — at every level, including Level 3
- Anything touching compliance, policy or account health
- Any report about another party, including a review report — an assertion about a person is not an automation decision
- Appeals against a suspension or an enforcement action
- Anything constituting a legal assertion, including brand and IP claims
- Filing anything we do not believe is accurate and supportable — which is not a setting, it is a refusal
- Anything depending on information he cannot see — a supplier problem, a product issue you already know about, a conversation had off-platform
- Deciding what to say to Amazon about your business
And the ones we will not dress up: we cannot guarantee a recovery, a reinstatement, a removal or an ungating. Amazon decides its own cases. We control which case is opened, how well it is evidenced, and whether anybody is still working it in week five. If a vendor tells you otherwise, ask them to put the guarantee in the contract.
Orbit, included / 12
The whole software suite comes with him. Free.
Dr. Shield lives inside Orbit — Full Circle's full Amazon software suite. Cases, reimbursements, listing health, inventory, finance, sales, advertising, ASIN-level profitability, and a set of trackers that watch your BSR, buy box, price and fee changes for you. This is software other companies charge a monthly subscription for. With Dr. Shield it is included at no extra cost.
What's in the suite
- Case ledgerEvery case ID, its age, its last contact and its next call
- ReimbursementsGranted, pending and reversed — shown net, not gross
- Returns reconciliationRefunds matched against physical returns, per order
- Fee reconciliationCharged fees against the dimensions on file, SKU by SKU
- Dr Shield™ ConsoleThe diagnosis, the filings and the decision log
- Listing healthSuppression, stranding, invalid attributes and missing images
- Content reversionWhat changed back, and which contributor outranked you
- Variation monitorFamily structure, detached children and split reviews
- BuyBox TrackerWho is in the box on your ASINs, and for how long
- Review screeningNew reviews assessed against the community guidelines
- Compliance & account healthRequirements per ASIN, gates, and anything counting down
- Finance & ASINsFees, reimbursements and per-product contribution profit
- Inventory & salesOn hand, inbound, reserved, revenue and trend by marketplace
- BSR, Price & Fee trackersRank movement, pricing, and Amazon's fee changes
- Alerts & notificationsSuppressions, stranded units and anomalies
The case ledger — the view every morning
| Case | Type | Status | Last contact | Costing | Next action |
|---|---|---|---|---|---|
| CASE-1148 | Suppression | Awaiting reply | 2 days ago | Daily | Three-day call due tomorrow |
| CASE-1152 | Lost inbound | Reimbursed | Closed | — | Watch for reversal |
| CASE-1160 | Variation conflict | Filed | Today | Daily | Await contributor response |
| CASE-1163 | Violation removal | Refused | 4 days ago | Daily | Internal review — same case ID |
| CASE-1171 | Brand Registry | Escalation appeal | Today | — | Documentation with a human |
Included, not upsold.
The third category / 13
You've had two bad options. This is the third.
Reimbursement software finds the discrepancy and hands you a queue — and has nothing at all to say about the listing that went down this morning. A person, in-house or outsourced, can do the case work properly but only for as many cases as one person can hold in their head. Dr. Shield is the thing in between that nobody was selling.
| Reimbursement software | A person on it | Dr. Shield | |
|---|---|---|---|
| Scope | Reimbursements only | Whatever is on fire today | All six problem classes |
| Who argues the case | You do | Them, on the ones they reach | Fable 5, on all of them |
| Case follow-up | Not its job | When there is time | Every three days, documented |
| Refused cases | Left in the list | Usually re-appealed | Internal review, one case ID |
| Suppressed listings | Out of scope | Found when sales drop | Watched daily, priced per day |
| Copy that keeps reverting | Out of scope | Re-edited, repeatedly | Diagnosed by contribution source |
| Competitor attacks | Out of scope | Reactive | Timeline built before you need it |
| Reviews | Out of scope | Varies alarmingly | Guideline breaches only, always |
| Reversals | Rarely shown | Rarely deducted | Tracked and netted out |
| Who approves a filing | You, manually | Them, told to you later | Your choice — claims always a human |
| Human expertise | None included | Varies with who you got | Operators from a $500M+ team |
| Software included | It is the software | Billed separately | Full Orbit suite, free |
Price / 14
Start free. Then a number we agree out loud.
- Fable 5 working your case load every day
- All six problem classes, not just reimbursements
- Suppressed and stranded listings watched daily and priced per day
- The three-day call cadence, with every call documented
- One case ID per issue, and internal review instead of re-appeal
- Your choice of approval level — claims always come to a human
- Every case logged with its evidence and its outcome
- Reimbursements shown net of reversals, not gross
- Operators from a $500M+ Amazon team supervising
- The complete Orbit software suite, free
- Leave whenever you want
Questions / 15
The things people actually ask.
Is this just a reimbursement tool with a bigger name?
What happens in the first 30 days?
How much will you recover for me?
Why do you quote net rather than gross?
Why does my listing copy keep reverting?
Can you get a negative review removed?
Do you go after competitors who attack us?
Do you file claims automatically?
What is the "three-day call" actually for?
Is Orbit really included?
Can I run Dr. Shield alongside the others?
Start / 16