Home Docs How It Works How MarginOS turns raw data into your margins (the pipeline)

How MarginOS turns raw data into your margins (the pipeline)

Last updated: August 5, 2026 By Aron Baczoni

Quick Reference

Inputs

Your connected store, plus whichever cost and advertising sources you have connected. Nothing to configure for the pipeline itself.

Outputs

Prepared metrics that every screen reads identically, costs converted to a common basis with unresolvable ones quarantined visibly, and figures that become more accurate as late cost evidence arrives.

Outcomes

Understand why a correction is not instant, why a closed period can become more accurate afterwards, and why two figures that look different are almost always different windows rather than a contradiction.

Your margins are assembled from sources that disagree with each other: a store that knows orders but not costs, a warehouse that knows fulfillment but not customers, ad platforms that each claim the same sale, and invoices that arrive weeks after the orders they describe. This article explains what MarginOS does with all of it, and — more usefully — what it guarantees about the result.

Four guarantees

The mechanics matter less than the promises they exist to keep. Everything below follows from these four.

1. Your source records are never rewritten

Orders, refunds and invoices are kept exactly as they arrived. When something needs correcting — you supply a real product cost, override an assumption, remap a channel — the correction is recorded on top of the original rather than replacing it.

This is why any figure can be explained after the fact. A margin that changed last Tuesday can be traced to what changed and when, because the earlier state was never destroyed. It is also why fixing a cost today does not quietly rewrite your history into a story you cannot audit.

2. Every screen reads the same number

A metric is computed once and every surface reads that same result. The figure in a summary tile, the figure in a detail drawer and the figure behind an alert are not three calculations that happen to agree — they are one calculation displayed three times.

This sounds like an implementation detail and is actually the difference between a tool you trust and one you check. Dashboards that recompute per screen drift, and once a user finds two numbers that disagree, every number is suspect.

3. Costs are standardised before they are compared

Cost data arrives in incompatible shapes — per-unit, per-order, per-shipment, per-invoice, monthly in arrears. Before anything is compared or totalled, it is converted to a common basis so that a fulfillment cost from one provider means the same thing as one from another.

Where a cost cannot be resolved to the product or order it belongs to, it is not silently dropped or averaged across the catalogue. It is held in a visible quarantine with the reason, so your totals stay honest and the unresolved item stays fixable. See why a cost landed in Unmapped Costs.

4. Nothing is invented to fill a gap

Where an input is missing, the result is reported as unknown rather than estimated into existence. This is covered in full in why MarginOS shows N/A instead of a made-up number, and it is the guarantee the other three exist to protect.

What this means day to day

Your numbers are not live to the second, on purpose

Figures are prepared rather than calculated the instant you open a page. That is what makes them consistent and fast, and it means a change you make — a corrected cost, a new mapping — appears once the affected figures are next recalculated rather than immediately.

If you have just fixed something and the screen has not caught up, nothing is broken. The correction is recorded; the display follows.

Late data is normal and is handled

A warehouse invoice for March arriving in April is the ordinary case, not an exception. Costs are attached to the orders they belong to rather than to the date they arrived, which is why a past period’s margin can improve in accuracy after the fact. A number that firms up as evidence arrives is the system working.

Accuracy improves over time, and says so

Early in a period, more of your margin runs on assumptions; as real costs arrive, more of it runs on evidence. MarginOS grades that rather than hiding it — see the Defensible Profit Model for how cost exactness is ranked, and Data Trust for the per-product view.

Common questions

I fixed a cost. Why has the margin not changed yet?

Because figures are prepared rather than recomputed on page load. The correction is stored and will be reflected when the affected metrics are next recalculated. If a figure still looks wrong well after that, check whether the cost was applied to the product you expected.

Why did last month’s profit change after month end?

Almost always late cost evidence — a fulfillment or freight invoice that arrived after the orders it covers. MarginOS attaches costs to the orders they belong to, not to the day the paperwork showed up, so a closed period can become more accurate. The direction is nearly always the same: profit that was optimistic becomes real.

Does correcting a cost rewrite my history?

It updates the figures the cost affects, but it does not destroy what was there before. The original record and the correction both persist, which is what makes a change explainable rather than merely visible.

Two screens show different numbers. Is that possible?

Not for the same metric over the same period and business scope — those are one calculation rendered twice. If two figures disagree, they are almost certainly different metrics or different windows: a 30-day figure against a 90-day one, or a direct-to-consumer figure against one including marketplace and wholesale.

Do I need to do anything to keep this running?

No. Keep your connections healthy and supply cost evidence where MarginOS says it is missing; the rest is automatic.

Related

About the author

Aron Baczoni

Aron Baczoni is the founder of MarginOS and a former decade-long veteran of Google's Ads and Global Business Operations units. His work is focused on bridging the gap between AI's promise and its practical implementation for direct-to-consumer (DTC) brands, helping them build a sustainable competitive advantage through strategic, high-ROI AI solutions.

Read Aron's story

Get Your Profit & Inventory Command Center

See your real margins in real time, not in spreadsheets.

Get your Profit & Inventory Command Center

Or email sales@marginos.com to see your real margins on your own data.