Skip to main content
Use a forecast to decide what to do next. Build an editable Excel workbook with a base case, alternatives, visible assumptions, and checks that help your team compare choices.

Choose the data

Use a project containing the historical data, actuals, targets, or existing plan that should anchor the forecast. For a quick trial, upload a spreadsheet. For a model you intend to refresh, use connected data where possible. If you already have a planning workbook, upload it too. Ask Addison to inspect its formulas and dependencies and preserve the conventions your team relies on. Need to bring data in first? Follow Connect your data.

Ask for a working model

Open the composer, choose the project, and describe the model and its audience:
Build an Excel forecast for next quarter’s gross margin using this project’s data. Show a base case, upside case, and downside case. Make revenue growth, product mix, and markdown rate editable assumptions, and flag any missing inputs.
Addison writes the .xlsx file into the project. Open it from the file card in the response or from the project’s file list.
A spreadsheet open on the Forecast tab, with Line and Measure columns and one column per fiscal month from FY26-M01 to FY26-M12. A header row labels FY26-M01 through FY26-M06 'Actual' and FY26-M07 through FY26-M12 'Forecast', with the forecast columns shaded. Under an Accessories block, rows run from units and orders through markdown_dollars, ticket_demand, discount_dollars, gross_demand, return_dollars_ultimate, return_units_ultimate, shipping_revenue, owned_gross_demand, mp_gross_demand, net_revenue, charged, product_cogs and inventory_writeoff to a bold GM1 line, then through fulfillment_warehousing, outbound_freight, return_processing, payment_processing, reverse_freight and marketplace_commission to a bold 'GM2 (delivered margin)' line. A Bottoms block begins below with the same measures. Sheet tabs along the bottom read Read Me, Assumptions, Overrides, Actuals, Forecast, Membership, Version Compare, Ref Facts, Ref Pnl, Ref Membership, Discover Period, Data Sources and Checks.

A forecast workbook, with actuals and forecast columns side by side

Refine the model

Check that the model uses the right source period, fiscal calendar, currency, and units. Ask Addison to explain the assumptions and which inputs are missing. You can refine these together once you have a model to inspect. For example:
Keep historical actuals separate from planning assumptions, and use formulas so the forecast updates when I change an input.
Before comparing scenarios, make sure you can identify: The workbook should not blur these categories. If it does, select the relevant cells and use Edit with AI to reorganize or label them.

Change one assumption

Change one input that matters (for example, markdown rate or revenue growth) and confirm that the scenario outputs recalculate. Then ask:
Explain which assumptions have the greatest effect on gross margin and where the scenarios begin to change the decision.
The goal is not to generate more scenarios. It is to understand which assumptions carry the decision. Keep historical actuals fixed when changing scenario assumptions. For an update to an existing model, ask which assumptions changed and how much of the change in the forecast each explains.

Verify the workbook

Open the Verification Hub from the shield in the spreadsheet toolbar and run the available workbook checks. Resolve formula errors, inconsistent outputs, or missing documentation before sharing the model.
An Excel workbook 'Olympics Top 5 Leaderboard.xlsx' open in Summation's in-app spreadsheet editor under a 'My Project' breadcrumb. A formatting toolbar (Arial, size 10, bold/italic/underline/strikethrough, alignment) sits above a cell-reference box (B8) and a formula bar reading ='Medals By Country Year'!D2. The 'Leaderboard' sheet, titled 'Top 5 Olympic Nations, Summer Games 2000-2016', lists nations (United States, Russia, Australia, Germany) with Gold/Silver/Bronze/Total, year-over-year change, and athlete-count rows across 1996–2016 columns. The top-right shows a verification shield, an 'Edit with AI' button, and a ⋯ menu; sheet tabs (Read Me, Leaderboard, Medals By Country Year, Data Sources, Checks) run along the bottom.

The workbook editor, with the verification shield in the toolbar

Verification can test the workbook’s construction and its relationship to the sources it used. Forecast results still depend on the assumptions you choose. Keep assumptions explicit and review them with the people accountable for the plan.

Download or share the workbook

Use the ⋯ menu to download the workbook as a standard .xlsx. The file retains its formulas so the model remains editable outside Summation. Open the exported workbook in the spreadsheet application your team uses and change one input to check that the intended outputs recalculate. Review macros, external links, and specialized features separately if your original workbook depends on them. Include the source period, assumption changes, and unresolved questions with the handoff. Need a shared tool instead of a workbook? Summation can also build a purpose-built application, available on Enterprise plans. Contact us to set one up; you cannot start it from the Applications page.

You are done when

You have a model that:
  • connects to the right historical evidence;
  • separates facts, assumptions, calculations, and outputs;
  • compares at least two meaningful scenarios;
  • recalculates when an assumption changes;
  • contains visible checks and a current verification state;
  • can be downloaded as an editable Excel workbook.

Next

Get verified insights

Investigate the historical drivers behind an assumption.

Automate recurring work

Refresh or deliver the model on a schedule after the logic is stable.