A ZSell Brand Report reads a brand's structural position on Amazon from the public record. This standard is the open definition of what that reading covers, how each part is measured, and the lines the report will not cross. It is the contract behind every report — published so a reader need not take our word for any of it.
How to read this standard
Every dimension is stated in two layers, together:
- The assertion — what we claim to check or measure, in plain terms.
- The measurement — the data it is read from, the rule that sets its status, and the limit on what it can show.
They are kept together on purpose. A claim with no measurement beneath it is marketing; a measurement with no claim above it cannot be cited. Binding them means what we publish and what we run cannot drift apart.
Part I — Principles
These govern every dimension.
- We measure and attribute; we do not decide. The report carries no single grade or score. Each dimension is reported as run and attributed, and the weight of any flag is triaged in a risk register, not rolled into a number. The buyer reaches the decision; we deliver the read.
- Reader-neutral. The report certifies the same facts to whoever holds it. The sourced record is identical across the acquirer, diligence-firm and owner editions; only an owner's private reconciliation — their own authorized Amazon figures, which Amazon's rules permit showing to no one else — is ever added.
- Reproducible figures, labelled interpretation. Every figure and status is derived deterministically from licensed public data, so a skeptic can re-derive it. The interpretation that reads those figures — "the read" — is AI-assisted and labelled as such; it reasons over the measured facts, it never produces them. The guarantee is reproducibility of the evidence, not "trust us" — which is what lets a report that changes hands carry the same weight as one you ran yourself.
- Point-in-time and dated. A report is read as of a single date — it carries one report date and states the data window it was read over (a 12-month default for history, rank and buy-box; a 90-day window for availability) in its coverage limits. Individual figures are not separately dated. Structure reproduces run to run; units drift, so a reading is interpreted as of the report's date.
- Categorical, not calibrated. A status is set by the presence or absence of a structural fact, never by a number crossing a threshold, and never by a model grading the brand. A suppressed buy box is a flag because the suppression was observed — not because a figure crossed a line.
- Coverage is stated, not implied. A dimension we cannot read is marked not assessed, never a silent pass. "Clean" is stated affirmatively, per dimension.
- No pay-for-outcome. No one can pay to change a reading, soften a flag, or suppress a report. Price and audience affect what is sold and to whom — never what the report says.
- Anchored where Amazon speaks; defined where it is silent. Each dimension names its Amazon basis where one exists, cited and dated; where Amazon imposes no requirement, the definition is ours, and is marked as such.
- Versioned, not retroactive. The standard carries a version. A material change bumps it and applies to new reports; earlier reports are not re-run or restated — each is a point-in-time artifact stamped with the version in force when it ran. Prior versions stay published and dated, so any report stays interpretable against its own version.
Part II — The anchor
The standard's authority is half borrowed, half built.
- Borrowed — Amazon's own requirements, where they are externally observable: the control-and-identity cluster (Brand Registry enrolment, seller-of-record, buy-box control, brand programs). We certify against Amazon's published bar and cite it.
- Built — the ZSell definitions, where Amazon imposes no requirement on what an acquirer most needs to know (revenue, concentration, tenure, trajectory, competitive position, shock history). Here the standard stands on the measurement discipline in Part I, not a borrowed rule.
- The account wall is a declared ceiling. Amazon's highest-signal requirements — account health, order-defect rate, intellectual-property complaints — live inside the Seller Central account and are invisible from outside it. Before a deal, the standard reads their on-platform shadows and says so; the owner edition, where the owner authorizes account access, is the lane where that half can later become directly measurable.
Part III — Status vocabulary
Each dimension is reported with one of five statuses:
- ✓ checked — run and clean; the structural fact that would flag is absent, stated so.
- ⚑ flag — a structural condition was observed that the reader must weigh.
- ◆ measured — a reading with no pass or fail, such as the revenue bracket or a tenure date.
- ○ note — read only in part, or a standing caveat; never a clean pass.
- ╌ not assessed — not readable on this run; a statement of coverage, not a pass.
Severity is not carried here. It lives in the report's risk register, severity-ordered — so the grading the status vocabulary avoids is not re-introduced through the back door.
Part IV — The dimensions
Each dimension below gives its assertion, its Amazon anchor (or notes that Amazon is silent), and how it is measured — source, status rule and limit.
Identity resolved
- Assertion. The brand is de-anonymised to the selling entity and the catalogue behind it — the report's starting point.
- Amazon anchor. Brand-identity consistency and seller-of-record. Partial.
- Measurement. From the brand resolver and the seller census. Checked when the brand resolves to a selling entity and footprint; flagged when it cannot be resolved on the public channel. Limit: the public channel only.
Control & identity
- Assertion. Who controls the brand's sales — whether a single authorized account holds it, how many accounts carry it, whether the buy box is suppressed, and how contested that box has been. The dimension with the highest kill-rate in practice.
- Amazon anchor. Authorized seller and buy-box control. Full.
- Measurement. From the seller census and buy-box history. Flagged on a categorical fact — more than one non-Amazon account carries the brand, or a buy box Amazon suppressed, or contested listings. Checked when a single account carries it and no box is suppressed. We report the presence of accounts; we never decide which one "is" the brand.
Revenue (measurable channel)
- Assertion. An independent monthly revenue bracket for the brand's channel — a measurement the buyer sets against the broker's number, not a check of the broker's number.
- Amazon anchor. None — Amazon imposes no such requirement. A ZSell definition.
- Measurement. From the subset of listings Amazon publishes a purchase volume for. Reported as a bracket, with no pass or fail. Limit: the measurable subset only — not total revenue.
Concentration
- Assertion. How broadly the brand's revenue and catalogue spread across niches, and whether it rests on a single shelf.
- Amazon anchor. None. A ZSell definition.
- Measurement. From the niche breakdown. Flagged on the categorical fact of a single niche; otherwise measured, with the share each niche carries.
Brand Registry
- Assertion. Whether the brand controls its own detail pages — enrolled in Brand Registry and running a live Brand Store, per marketplace.
- Amazon anchor. Brand Registry enrolment and the trademark behind it (as a proxy), and brand programs. Partial — a Brand Store is registry-gated, and registry requires a trademark, so a live Store is a reliable "enrolled and in control" signal.
- Measurement. From the linked Brand Store on the public channel. Checked when a live Store is found (named, with any partial cross-marketplace enrolment stated); a note when none is found — listing control unverified, never "not enrolled," since a brand can be enrolled without a Store. Limit: legal IP ownership and disputes are post-deal diligence, not a channel reading.
Custody & manner of use
- Assertion. How long the brand has sold on Amazon, from the earliest first-seen date across its listings.
- Amazon anchor. None. A ZSell definition.
- Measurement. From listing first-seen dates. Reported as a measured fact. Limit: ownership-change is not reconstructed — stated, not implied.
Shock-event history
- Assertion. Availability, rank and price shocks over the window — revenue-bearing stock-outs, listings that fell out of ranking, sharp price cuts.
- Amazon anchor. A partial proxy for account-health requirements; otherwise a ZSell definition.
- Measurement. From the top listings' availability, rank and price history. Checked only when stock-outs were read, were clean, and no rank excursion was found; otherwise measured, leading with the revenue-bearing scope — a stock-out on a dead-tail listing is not a channel shock. Limit: a 90-day availability window; the shock type read is named, so a clean line is not read as every shock type being clear.
Competitive position
- Assertion. Where the brand ranks in its category, and which incumbents sit above it.
- Amazon anchor. None. A ZSell definition.
- Measurement. From the category position reading. Checked when rank and the incumbents above were measured; not assessed otherwise.
Growth trajectory
- Assertion. Realized, Amazon-measured demand momentum — units moving over the window, not a projection.
- Amazon anchor. None. A ZSell definition.
- Measurement. From the public demand history. Reported as a measured reading. Limit: realized demand only; variant proliferation is named so it is not read as new products.
Upkeep & review integrity
- Assertion. How the catalogue is kept — fulfilment mix and review depth — and whether review accrual is anomalous. Review authenticity (the content of reviews) is not assessed.
- Amazon anchor. Partial — Amazon prohibits review manipulation; also a proxy for account history.
- Measurement. From the fulfilment mix and review-accrual history. Flagged on a review-accrual spike far above a listing's own pace — a manipulation tell to verify, never proof of fake reviews; a note when accrual tracks pace. Authenticity is not assessed.
Category / platform risk
- Assertion. A standing caveat: Amazon's policy, gating and fee shifts are category-wide, not brand-specific — weigh them against the niches the brand sells in.
- Amazon anchor. Amazon's own evolving policy (category-wide).
- Measurement. Always a note.
Part V — Scope and coverage limits
- Scope: the Amazon public channel and the brand's structural position on it. Not financial diligence, not intellectual-property, legal or supply-chain review, not account-internal performance.
- The account wall: before a deal, account-health requirements are read only as on-platform shadows. The owner edition, where the owner authorizes account access, can later extend the standard across that wall.
- Window: history, rank and buy-box are read over a trailing window — one year by default, or a listing's lifetime if shorter; availability over a 90-day window. The windows are stated once in each report's coverage limits; individual figures are not separately dated.
- Measurable-revenue subset: revenue and share figures cover the volume-badged listings only, never the whole catalogue; niches not read are not assessed, never "absent."
Part VI — Versioning
The standard is versioned. A change to any assertion or measurement rule — adding a dimension, or changing a status rule — bumps the version and applies to new reports from that point.
Versions are not retroactive. An earlier report is never re-run or restated when the standard changes. A report is a point-in-time reading: re-running it would change its results both because the standard may have added or redefined a dimension and because the underlying public data drifts over time — and it would break verification, since a report is confirmed by a content hash over its data, so a re-run is a different report, not an update.
Instead, each report is stamped with the standard version it was run under, as of its date. Prior versions stay published and dated, so any report stays interpretable against its own version; a reader who wants a newer check commissions a fresh report under the current version. Every report issued under this standard carries its standard version, so you always know which definitions it was measured against.