CANOPII

The work

WHAT WE DO

Into the forest

The short version

We run it, and we answer for it.

Canopii is a service firm. Consultants in the field. We hold one definition for every number a business produces, and we stay on the hook when a figure comes out wrong.

Nobody bought a bad system. They bought eight good ones, four years apart, each one right on its own day. The layer that makes them agree does not exist yet. We build it, and you own it.

Where the work actually is

A licence buys the easy 20%.

The easy 20%

A place to put the data.

The hard 80%

  1. Ingesting messy sources

  2. Normalising them

  3. Backfilling history

  4. Automating workflows

  5. Catching a vendor who changed a field without telling anyone

  6. Owning correctness

That’s the part we do.

How it moves

We say what we would do, then keep changing it.

We tell you what we would connect first and what it will not fix. Then we build it, and keep changing it as the numbers come in.

Sometimes the answer is one system fewer. We name the licence to cancel, and do the work that makes cancelling it safe.

YOUR SYSTEMSEACH RIGHT ON ITS OWNORDERSSTOCKSPENDBOOKSSHIPPINGRETURNSEMAILSHEET30 WEBHOOK TOPICSRECONCILED EVERY 2 HOURSONE DEFINITION PER NUMBERWE SAY SOSTILL DARKONE LEDGERFX-NORMALISED · ECB DAILYANSWERSON SCREENTHE 02:00 MAILTYPED READERS, NO SQLACTIONSWITH YOUR SIGN-OFFNOT RUNNING YETONE FIGURE, NOT THREEREAD THEM AS HOURS OLD
  1. 01Your systemsEight sources · each right on its own
  2. 02Connect · backfill · keep current30 webhook topicsReconciled every 2 hours
  3. 03One definition per numberWhere two sources disagree, we say soWhat is still dark stays drawn dark
  4. 04One ledgerFX-normalised on ECB daily rates
  5. 05Answers now · actions not yetAnswers: on screen, in the 02:00 mail, from typed readers with no SQLActions: with your sign-off — written, not running
  6. 06One figure, not threeRead them as hours old

A business does not keep its problems inside one department. We do not either.

What the product shows

Sales, and what they are made of

Every figure is computed once, by one reader, and quoted by the dashboard, the export, the nightly mail and the agent alike. Open sales, discounts or returns to see the constituents.

Total sales · 30 days

vs prior 30d

$1,207,710

62.4k012/715/718/721/724/727/730/72/85/88/810/8

Total sales breakdown

What the traffic does

Where the drop-off actually happens, from the first visit to the paid order. The basket step is drawn faint on purpose — there is no honest source for it, so we say so rather than estimating it.
  1. Sessions184,220
  2. Added to basket—

    No source for this step

  3. Reached checkout11,486
    6.2% of sessions93.8% drop-off
  4. Orders8,341
    72.6% of reached checkout27.4% drop-off

What it cost to get them there

Every channel's spend on one axis, week by week, so a line is expensive only next to the others. Crossings are the point — the week paid social overtook search is a decision, and it is visible without opening a report.
15.2k018/525/51/68/615/622/629/66/713/720/727/73/810/8
Paid searchPaid socialShoppingEmail & lifecycle

Where it ships

Three fulfilment hubs and every lane running out of them, drawn from the shipping address on the order rather than the storefront it was bought through. Hover a lane or a country to see what is moving along it.
Orders shipped by destination country, with per-store fulfillment lanesAmericas — United States: 3,385 orders shippedEurope — Netherlands: 3,793 orders shippedAsia · Hong Kong — Hong Kong: 1,815 orders shipped
8,993 orders in this period
AmericasEuropeAsia · Hong Kong
12,418orders

Whether they come back

Each row is a month's worth of new customers, followed across the months after. The triangle is honest — a cohort acquired last month has one month of history and the grid draws it that way.
CohortSizeM0M1M2M3M4M5M6M7M8M9M10M11
2025-08412100%34%19%24%16%18%11%14%12%9%11%8%
2025-09268100%28%31%17%20%13%15%9%12%10%7%
2025-10455100%41%22%18%23%14%12%16%9%11%
2025-11602100%37%26%21%15%19%13%10%12%
2025-121284100%18%12%9%11%7%8%6%
2026-01344100%44%29%33%21%18%22%
2026-02371100%31%35%19%24%16%
2026-03209100%48%27%31%18%
2026-04356100%26%33%20%
2026-05498100%39%21%
2026-06427100%35%
2026-07314100%

One item, day by day

Units sold against stock on hand, refunds under the axis, and both restocks visible as steps. The view that answers 'are we about to run out' without anyone opening a spreadsheet.
6/126/227/27/127/228/13201,3322 refunded
Units soldStock levelRefunded unitsWeekend columns dimmed

Where we are useful

Your systems disagree.

Two reports, one question, two answers, and a morning lost proving which is right.

Nobody trusts the number.

It exists, it is on a screen, and the meeting still opens by asking where it came from.

A person is the pipeline.

Someone exports it and rebuilds the answer in a spreadsheet every week.

Decisions wait on a spreadsheet.

A price change or a reorder is held until somebody has the time to assemble the evidence.

You would rather keep your tools.

We connect what you already run and take over the tending.

Reporting

The report nobody could run before.

  1. Every platform reports on itself.

    Your storefront cannot see the ad spend that produced the order. The ad platform cannot see what came back three weeks later.

  2. One place, one set of definitions.

    When every source lands in the same layer under the same definitions, a report stops being a project and becomes a question.

  3. The report you actually wanted.

    Margin by channel after returns. The true cost of a discount code. Stock cover against what the forecast says is coming.

We build the ones you ask for, and they are yours.

Next

Tell us what you need.

team@canopii.co