Bring your existing spreadsheet

Import a planning workbook as a dataset, build a model on it, review what came across, and start asking what-ifs.

Most planning already lives in a spreadsheet. This walkthrough brings one in, checks what eode understood of it, and gets you to what-ifs — without rebuilding anything by hand.

A workbook is not a model. It becomes a dataset first, and a dataset can then feed as many models as you like.

Before you start

Any planning workbook works. It does not need reshaping, but two things make the import cleaner:

Sheet names that say what they hold — "Demand", "Capacity", "Costs". They help the mapping.
One scenario per sheet set, if the workbook holds several. eode asks which one is the base.

Step 1 — Import the workbook as a dataset

You can do this from the Datasets rail item, or from inside the wizard while you author the model. From the wizard:

Open New Model

Left rail → New Model → Next → Basic. Name the model and click Next.

Open the datasources row

On Describe your problem, open Upload or select relevant datasources, then click Upload new datasets and attach them to this model — "New .xlsx files, Google Sheets or other datasets. This opens the Import Data window from the Datasets tab; what you import there is attached to this model."

Pick the source and the file

The Import Data window opens on its source catalog. Microsoft Excel is under Popular; so is Google Sheets, which signs you in instead of taking a file. Drop the .xlsx on the drop zone — "Drop a file here, or click to browse".

Choose what to pull

The window lists every sheet it found, all selected. Clear anything you do not want — "Only selected resources are pulled into the model."

Import and attach

Review names the dataset for you if you leave the name empty (the date and time), then Import data → runs the import and Attach to this model → brings it back to the wizard with the dataset attached.

Already imported it? Select data from previously connected datasets lists everything on your Datasets tab; tick it there instead.

The import window is covered in Import data; the full connector catalog and the connect step in Data integration.

Step 2 — Check how your workbook was read

Describe your goals and your business the same way you would without a workbook — what you said in words is what the design is confirmed against. Then Next.

Because a dataset is attached, Clarify opens on the reading of your workbook, and nothing is designed until you agree it. The wait is Reading your data, counting Read {n} of {m} sheets. Then four screens:

Your data, as I read it

WHERE THE NAMES COME FROM shows the sheet and column your products, customers, sites and periods were found in, with a few values from each. If one is wrong, Change it to another sheet and column, or mark it Not in this workbook.

Each sheet has a card saying what RIA took it to be and how sure it is. Press Use it for a sheet the model should read and Set it aside for one it should not — a results tab or a scratch sheet. A sheet marked GUESSING has to be decided before Next moves on.

A report-style sheet, with panels of figures rather than one table, is read out into tidy tables; each says where in the sheet it came from and which cells could not be read.

The columns and the terms

A table per sheet you are using: what each column means, what it names, its unit, and how sure RIA is. A unit or term it could not settle is asked here, once, with the answers to pick from — a money column in dollars or thousands of dollars, say. Pick one, or type another unit.

What is missing

The screen is headed What your data does not say. Each thing the model needs that your sheets do not carry, with four answers: Assume it, with my consent, Supplied: it is in... (name the sheet or column, or say it in a sentence), Scope it out, or Stop here if you would rather come back with the data.

What you want to decide

One sentence saying what the model is for. It starts with what you wrote on Describe, or with a proposal drawn from the reading if you wrote nothing; rewrite it in your words. Then Design on this reading.

The full reference for these screens is in Building models.

If the workbook holds more than one scenario, Clarify also asks "Which scenario is the base?" before it builds. Picking one other than the workbook's own default makes eode set the selector and recalculate the workbook first; the wait says "Recalculating the workbook for" your pick, and takes a little longer than a normal build.

Step 3 — Watch it translate

On a workbook model the build screen says what it is doing: "Translating your workbook into the model's structure and data." It lands on "The structure is built", and Next opens Check the design: every section on one page, each with Confirm or Needs correcting, then Accept the design. On a workbook model, read SCOPE and WHERE YOUR DATA LANDS against your sheets — when a sheet holds more rows than the design names, SCOPE says so. See Building models.

Step 4 — Review what came across

Step 4 of the wizard, Load data, opens on "Review your imported data" for a workbook model. This is the important part, and it is worth reading properly.

The headline compares two things side by side: the plan your workbook recorded, imported exactly as stated and traced to the cell it came from, against what the model chose when it optimized inside the same caps your workbook set. A gap between them is not an import bug — it is what optimization found that your spreadsheet did not. That gap is often the whole reason to do this.

When a figure differs, "Why do these differ?" explains that specific difference, and labels where the explanation came from: Written by RIA, or From known patterns. Work it out again re-runs it.

Below the headline, category panels show every imported field, its value, and the workbook cell it matched, with a traced percentage per category. Fields the importer derived rather than matched are marked "derived — no direct cell match" so you can check them. Mark as reviewed keeps your place as you work down.

Constraints from your workbook lists the caps taken straight from your sheets — demand, capacity, discounts. When one binds, the plan cannot meet everything, and that is your workbook's own intent, not an error.

Step 5 — Accept and promote

Validation findings come with the review; warnings have to be acknowledged before promotion. Accept each category, or Accept all, then Accept Base & promote. The imported data becomes the model's Base and the model is ready.

If something upstream is wrong, do not accept it: send the model back. It returns to its questions with your answers kept, and from the review screen you can Replace workbook before building again.

If the workbook fails validation, the model's own page says so — "Your workbook didn't pass validation. Fix the flagged cells and re-upload — the checks re-run automatically." — with a Re-upload the workbook button. Nothing is promoted until the checks pass.

Step 6 — Ask the questions your spreadsheet couldn't

With Base in place, ask what-ifs in plain English — "What if our biggest supplier raises prices 12%?" Each becomes a scenario with deltas against your imported plan.

When the workbook itself changes, Refresh from workbook on the Scenarios screen re-applies the saved mapping to the updated file. If the new file changed the structure rather than just the numbers, it reopens the reconciliation review instead of applying silently.

Datasets — the dataset library, the import window, and binding a dataset onto a model.
Data integration — the connector catalog, saved connections, and scheduled syncs.
Building models — the five steps in full.
Reading results — how to read what the model chose.