Every dollar, accounted for.

Last month’s P&L, thirty days of sales, your traffic funnel, your ad efficiency and your Amazon account health all arrive together, from a single call against a summary that was built during the last refresh. Nothing aggregates while you wait, so the page opens at the same speed whether you list four hundred SKUs or forty thousand.

The month-end projection blends three independent methods — a straight run rate, a seven-day trailing pace, and last month’s own shape at the same day. You get the blended number, the low-high spread those three actually span, and a clock that reads Day 6 of 30, so you can act while the month is still running instead of finding out on the last day of it.

Your AHR score sits against its thousand-point baseline, with late shipment, on-time delivery, both order defect rates, invoice defects and policy violations beside it. Every tile carries Amazon’s own target and turns red on Amazon’s own verdict — Sellerbud never invents a pass mark of its own.

Each card names its own period in the header — Last 30 Days, Last Month — and the ones that go deeper carry a circular arrow beside it. Top States is one click from the state analytics screen, Advertising from the ad console, so the morning glance and the actual investigation are never far apart.

No metric ever appears on its own. Every tile shows the focused month next to the two before it, with a twelve-month chart beside it, so a number that is quietly slowing looks different from a number that is holding.

Sales, Advertising and Traffic, eight metrics each, in a fixed order that never shuffles under you. The ratios — CTR, CVR, ROAS, TACOS, Return % — are re-derived from the summed month rather than averaged across days, so one quiet Tuesday can never drag a monthly number sideways.

A grid for scanning trends, one table for reading a whole month across all three sections, another for reading a single metric across every month. The choice rides in the query string, so a particular reading of your data is a link you can hand to your accountant.

Open a finished month and the profit block is waiting: gross revenue, then the six deduction lines — return value, cost of goods, fees, shipping, ad spend and misc expenses — then net profit and net profit percent. Net profit is recomputed from the rows above it rather than read from a stored total, so the column always adds up.

On your India store, seven columns per state — net revenue, net orders, net units, return rate, average selling price, gross profit and gross profit percent — ordered by net revenue descending, and ordered by the aggregation rather than by your browser. Best and worst are decided on the server, which knows that on return rate the low number is the good one. State rows resolve against a seeded reference set, which is why this screen is India-only today.

A green bar for net revenue on the left axis, a red bar for return rate on the right, and the smallest states — the ones that together account for under a tenth of your revenue — folded into a single Others bar at the end. A state that sells beautifully and sends a chunk of it back stops hiding behind its revenue rank.

Pick a state and you get its month-by-month trend across the same metrics. Pick a month inside it and you get that state day by day. The month picker only offers months that state actually has data for, so a drill-in never lands on an empty screen.

Amazon’s ship-to field is free text — short codes, squashed spellings, casing variants. Sellerbud normalises each one through a curated alias map and settles the genuinely ambiguous codes against the order’s PIN code, right down to the Andhra Pradesh and Telangana overlap in the 5xx range. India’s states and union territories are the seeded reference set.

Pick a comparison window and every metric arrives with its prior value and a direction that knows what good means. A falling ACOS reads as a win, rising ad spend does not automatically read as one, and a metric that went from nothing to something says New instead of an infinite percentage.

ASIN, SKU, parent SKU, category and tag — one component and one endpoint serve all five, so switching the grain never changes what a column means. Four of them sit in the breakdown's collation dropdown and tags keep their own page. Sorting runs in the database across the whole catalogue, which is why page one really is the top.

Every row carries the product image, its parent SKU and parent ASIN, the resolved category, and a strip of child thumbnails with a plus-N-more overflow. Return % and TACOS sit right under the product instead of buried in a column, and the metric you sorted by is highlighted in every row.

Cost of goods, Amazon’s fees, ad spend and returned value come off the top line, and what is left is the number that actually matters. The same subtraction runs at every grain — the whole store, a category, a parent SKU, an ASIN, a single SKU — because it is one declared chain of calculations rather than five spreadsheets that disagree.

Every cost you record becomes a new dated row rather than an overwrite, and each order is costed at the price you were paying on its own date. Your very first cost is backdated so historical months fill in, and last quarter keeps last quarter’s cost when you renegotiate today.

Download a template already filled with every SKU, its ASIN, title, selling price and current cost, type into the one blank column and send it back. Only the rows you filled are applied, and the result names exactly what updated, what was skipped, what was invalid, and which cost is more than ten times its selling price and probably a typo.

Gross profit is one of the sortable fields, so ordering the catalogue by it ascending puts the products losing you money on the first screen. SKUs you have not costed yet fall back to a default percentage you control, so no row is ever blank and no product hides in an empty cell.

Settled, Pending Settlement, Not Settled, Overdue and Cancelled sit across the top as counts. Click one and the table filters to it while the rest dim; click again to clear. The statuses are derived rather than guessed — Overdue means sixty days have passed with no settlement row against that order.

Revenue, settlement and payout percentage on every row, with the settlement summed at read time across every line Amazon closed for that order. A refunded order can show a negative payout, and it does — the ledger is not tidied up on the way to your screen.

Open an order and you get the per-SKU settlement matrix: principal, tax, commission, fulfilment fee, closing fee, and the order-level lines merged into their own row. Amazon’s camelCase fee names are title-cased, and duplicate lines for the same fee are summed rather than quietly dropped.

On your India store, every return lands in a worklist built around six counters — four of them clocks that run out (SAFE-T filing, clarification replies, appeal windows and the sixty-day arrival window) and two that keep the score (rupees still at risk, and what you have already recovered). Deadlines are recomputed on every read from the live policy windows, and reconciling a unit snapshots its cost basis on the server. Sellerbud surfaces the claim and the clock; filing it stays manual in Seller Central.

Payment reconciliation, performance, month analytics, state analytics, hourly ad performance and inventory planning — the six you can build today. Each one is a catalogue row rather than a bespoke screen, so they all share the same picker, the same configurator and the same guarantees.

The configurator lists every column the export will contain, counts them, and shows a full sample table underneath, labelled illustrative because it is generated rather than yours. Preview and worker read the same catalogue document, so the columns you looked at are the columns that get written — with one exception: hourly ad performance is answered by Athena and the file keeps Athena's own snake_case headers.

Generation runs in the background and streams straight into the file, so a large export never eats memory and a failed run never leaves half an object behind. Every report ends somewhere you can act on — Done with a download, Empty with no data found, or Failed with a retry — and re-running one copies its parameters into a fresh row instead of overwriting the old file.