What one account did last time, and how it is likely to do now

The anchor. Each styles[] entry carries the prior book, the prior sell-out and, where the grade is readable, baseline: its state, basis, evidence, grade, pace, expected sell-through, level and rate, a range with the reason it is wide or absent, the curve and its bucket size, the members it voted on and the ones it dropped, the doors and weeks reported, the window and where the window came from, one composed sentence, and by_door. The header carries sellout_reported and sellout_reason. ordered_qty on each style is this season's live booked units for the same account and style, off the same statement as the prior figures, so ordered against predicted is a subtraction over one row rather than a crossing of two reads taken at two instants. A style in scope with no live line carries 0 and not null: the book WAS read, and having taken none of a style is the reading a planner came for. baseline.by_door[] is the account's level delivered to each door the account has open, one row each, carrying door_id, name (the buyer's own key for the same field, character for character, so a face never has the field rename itself under it), door_reference, units, share, rung, rung_label and source. rung is measured (this door sold the styles the grade stood on), category_average (it reported none of them but does sell the category), account_average (neither, so it takes the account's average door) or not_measured (nothing could be sized, so the row carries a null figure and never a zero). The rows sum to level_units by construction, so the account's own figure and the doors under it cannot disagree.

source SAYS WHICH READ SERVED THE ROW, and the two are not the same claim. door_baseline is the engine's own per-door pass: the availability correction taken on that door's own shelf, with the door's evidence beside the figure (sold_last_season, sold_corrected, out_of_stock_days, availability, weeks_reported, styles_reported). account_split is the INTERIM: the account's level divided at read time by each door's measured share of the styles the grade stood on, which cannot correct for availability and never carries category_average. The engine's rows are served wherever it has any; the interim answers only for a graded style the pass has not reached yet, so a face that branches on source can say which it is looking at rather than guess. An account that reports one total has no doors on file, and by_door is then empty rather than one synthetic door holding the whole number. The KEY IS ABSENT, with by_door_reason carrying the sentence, where the split could not be read at all: the grades stand either way, because the split is an addition on top of figures that are already complete. by_door_reason is null where the split was read. basis and evidence are the provenance pair, and a client switching on the grade wants both. basis is the rung the grade was read from, in ladder order: account_predecessor, account_kit_member, account_cohort, account_family, brand_analog, sell_in_only, none. The first rung carrying enough reported weeks at this account votes alone, and the outranked members are dropped and named rather than blended in at a lower weight. evidence is how far to trust it: measured, uncorrected, borrowed, brand, none. It is evidence, not basis, that carries the borrowed against measured label, because one clean predecessor of the account's own doors is a measurement and several blended is a borrowing, and the figure looks identical either way. uncorrected is that same own record read over weeks nobody reported a shelf for, so a slow week may be an empty shelf rather than absent demand. brand_analog with evidence: brand is the analog case named above: nothing at this account, so the brand's own answer is served at brand level and labelled as borrowed from us. grade is winner, sleeper or bleeder, and state is graded, magnitude_only, brand_fallback, sell_in_only or refused. The header also says which morning the grades are from: grades_as_of is the completion stamp of the last grading run that succeeded, engine_state is that engine's state (never_ran, not_measured, running, fresh, stale, partial, failed, not_applicable), next_run_at is when it runs again, and engine carries the same four facts in the shape every engine read on the platform uses. They are stamped on every path, including the one where the grades themselves could not be read, never from the reader's clock, which is what this page used before. grades_state is a different question from engine_state and sits beside it: that one is about the RUN, this one is about the answer in your hands. It is served, not_on_plan or unreadable, and grades_reason carries a sentence only for not_on_plan, the same text a 403 plan_excludes carries. Three responses can hold engine_state: fresh and no grades at all, for three unrelated reasons, and this is the field that separates them. A style with no grade carries baseline: null where the pass simply never assessed it, and a not_assessed block (state, reason_code, sentence) where the last run named why it did not grade this ACCOUNT at all (a read that failed, reading conventions nobody has confirmed). "Assessed by nobody" and "we could not read their record this morning" send a seller to two different places. baseline.by_door[] is the same level, delivered to each of the account's shops: door_id, name, share, units, and rung, the word for the evidence the figure stands on. measured is the door's own history of the styles the grade was measured on, corrected for its own shelf; category_average is a door that reported none of those styles but does sell the category; account_average is a door that reported neither, which where nothing was measured at all is the level divided by the doors on file; not_measured is a door on a style the engine could not size, and it carries no figure. The units sum to level_units exactly, so the door column and the account figure above it cannot disagree. Each row also carries the door's own evidence: sold_last_season as the account reported it, sold_corrected with the shelf accounted for, out_of_stock_days, availability, weeks_reported and styles_reported, because a figure lifted by a third for a shop that was dark for eight weeks has to be able to say so. availability: null means no week at that door carried a shelf reading, so the units are taken at face value rather than corrected. by_door: [] is a real answer and the commonest one on an account that reports a single total: it has no door to break down. The key ABSENT means the split could not be read, the same distinction baseline itself draws one level up. baseline_source on each door row names which rung of the source chain produced its figure: own_sales (our own sales on every channel we sell through: the brand's own forecast, served where this account reported nothing for the style), sell_out (this account's own reported sell-out for a predecessor, kit member, cohort or family member), sell_in (a sell_in_only row, off what we shipped or ordered) or category (a category-level average, the chain's last resort). dtc (our own d2c-typed channels only) is reserved for a figure read off those channels alone; no door row is read that way today. null on a refusal, where no rung answered, or on a tenant this field has not yet reached. Served beside grade_basis.