Building even a simple financial model in Excel takes real spreadsheet fluency, not just knowledge of the business. You need to separate assumptions from the calculations that depend on them, anchor the right cells with absolute references, and make sure that changing one growth rate doesn’t quietly break every formula downstream of it. For a consultant sketching numbers before a client call, or a small business owner trying to answer “what happens to our margin if costs go up,” that skill gap is usually the real blocker, not the math itself.
An AI virtual assistant built directly into a spreadsheet is designed to close that gap. Instead of building a model cell by cell, you describe what you want in plain language, and the assistant drafts the structure, formulas, and links to a set of assumptions you can change later. Zoom Sheets, one of the four products inside the Zoom AI Productivity Suite, works this way: describe a model in a chat-style prompt, formulas included, and it returns an editable spreadsheet draft for you to review and test.
We asked Zoom Sheets to build a straightforward SaaS revenue forecast: one revenue line growing at a fixed monthly rate, a flat fixed cost, a variable cost tied to that month’s revenue, and a running net income total. It’s a single-line growth model with cost assumptions rather than a full three-statement build or a multi-product forecast — more on exactly what that scope does and doesn’t cover further down.
Ready to try it? Open Zoom Sheets.
Building a 12-Month Revenue Forecast, Step by Step
Here’s what the process looked like, starting from a single written prompt and ending with a formula-based table that still needs review before use.
- Switch the Hub composer to Sheets, then describe the model in plain language. In Zoom Hub, the home for your Zoom Workplace files, select Sheets as the output type and enter a detailed prompt describing the model you want. Once Sheets is selected, the prompt is enough to generate the initial structure and formulas.
We typed a fairly detailed request: a 12-month revenue forecast starting from $50,000 in monthly recurring revenue, 8% month-over-month growth, $15,000 in flat fixed costs, and variable costs equal to 30% of that month’s revenue, “using formulas throughout so the growth rate and cost assumptions can be changed later.”
What came back split the model into two blocks instead of one long table: an “Assumptions” section at the top holding the four inputs, and a 12-row forecast below it with Revenue, Fixed Costs, Variable Costs, Total Costs, Net Income, and Cumulative Net Income columns. Splitting inputs from calculations this way is what makes the “changed later” part of the request possible, since every formula in the forecast rows points back to the Assumptions block instead of repeating raw numbers.
In this walkthrough, spot-checking cells in the forecast rows showed that revenue used =B8*(1+$B$3) to compound the prior month’s revenue by the growth rate, and variable costs used =B9*$B$5 to apply the variable cost rate to that same month’s revenue, with both formulas anchoring the assumption cells in place with dollar signs. Total Costs added Fixed Costs and Variable Costs for the month, Net Income subtracted Total Costs from Revenue, and Cumulative Net Income carried each month’s Net Income forward into a running total. This was a limited check of the displayed example, not a validation of the entire model. Review every generated formula against the assumptions, source data, and expected outputs before using the forecast.
- Ask a follow-up question that requires an actual formula, not a quick answer. We went back into the same sheet and asked for two more things: the implied annualized growth rate from Month 1 revenue to Month 12 revenue, and a cell that identifies the first month the business reaches cumulative profitability, specifying “using a formula rather than you telling me the answer directly.”
The assistant added a new “Key Metrics” block with both, labeling the first one “Implied CAGR.” In this walkthrough, =((B19/B8)^(1/11))-1 used the 11 growth intervals between 12 monthly values and returned 8.00%, matching the example’s monthly growth assumption. Treat that result as a model check to review, not as an independently validated conclusion. A separate annualized calculation would compound the monthly growth factor over 12 months: (1 + 8%)^12 - 1, or approximately 151.82%.
The second formula,
=MATCH(TRUE,G8:G19>0,0), searched the Cumulative Net Income column for the first positive value rather than hard-coding a month number.
- Change an assumption and watch whether the model actually recalculates. To test the “changed later” part of the original request, we edited the Fixed Costs assumption directly, raising it from $15,000 to $45,000 to simulate a heavier cost structure.
Every cell tied to that assumption updated immediately. Net Income turned negative for the first four months, Cumulative Net Income stayed negative through Month 7 (bottoming out around -$2,701.88), then crossed into positive territory in Month 8 at $12,281.97. The Month of Cumulative Profitability cell updated on its own, from 1 to 8, matching that exact crossover point, with nothing re-entered or re-asked for.
The monthly-growth check in the Key Metrics block, on the other hand, stayed at 8.00%, unchanged from before the edit. That makes sense: its formula only references the two revenue cells, so it has no reason to move when a cost assumption changes. The fact that it held steady while everything downstream of Fixed Costs moved is itself a useful signal that the formulas are wired to the right inputs, not just recalculating for the sake of it.
In this walkthrough, the formula bar still displayed =MATCH(TRUE,G8:G19>0,0)after the edit, and the displayed result changed with the revised inputs. That observation is limited to this example; review the formula logic and resulting month whenever assumptions or the forecast range change.
That last step is what actually matters for a financial model, more than the generation step that comes before it. A model that only looks right the moment it’s created isn’t really a model, it’s a snapshot. Changing a single assumption cell and watching a running total and a MATCH-based lookup update together, while a formula that has no business moving stays exactly where it should, is what a financial model is supposed to do when someone asks “what if” during a planning conversation.
Where This Kind of Model Fits (and Where It Doesn’t)
This walkthrough covers one specific, common shape of financial model: one revenue line with a compounding growth assumption, a fixed cost, a variable cost as a percentage of revenue, and a running profitability check. That shape covers a lot of real planning work: a services business projecting monthly revenue against overhead, a small SaaS product estimating monthly profitability, or a consultant putting rough numbers in front of a client before a bigger build-out.
What wasn’t tested here: multi-product revenue lines, tiered pricing, retention- and churn-driven SaaS models, and full three-statement models with tax and depreciation treatment. Those involve materially more structure than the single-line model built in this walkthrough. If your model needs that kind of depth, treat the natural-language, formula-generation approach shown here as a starting point for building it further, not as a claim that it already handles that level of complexity.
There’s one boundary case that shows up directly in the model: if a set of assumptions never produces a month where Cumulative Net Income crosses into positive territory within the 12-month window, the Month of Cumulative Profitability formula returns #N/A instead of a month number. That’s the ordinary behavior of a MATCH formula reporting “no match found” in the range it searched, not a sign that anything broke. It also doubles as a useful signal on its own, since it means the business needs either a longer forecast window or a change to the cost assumptions before it turns profitable.
Getting the Model Out of Zoom Sheets
A generated model only stays useful if it survives contact with the rest of your work. Zoom Sheets supports real-time collaborative editing, version history, and permission controls, so multiple people can review a forecast while controlling who can edit and who can only view it. That matters once a forecast stops being a one-person exercise and starts getting reviewed by a manager or a client.
Sheets also imports and exports Microsoft Excel and CSV files, so a model built this way can move into another workflow. Complex formulas, charts, fonts, or formatting may vary during conversion, so open the exported .xlsx file in Excel and review the calculations and layout before continuing work or sharing it.
The same conversational starting point isn’t limited to a typed description, either. Because Sheets lives inside Zoom Workplace alongside Meetings and chat, Zoom’s broader product direction is for context like growth assumptions or cost numbers raised on a planning call to carry into a sheet like this one, rather than needing to be retyped from memory after the fact.
FAQ
Do I need to know Excel formulas to build a financial model like this?
No. Every formula in this walkthrough, including the compounding revenue growth, the monthly-growth check in the Key Metrics block, and the MATCH-based profitability check, came from describing what was needed in plain language rather than writing the formula syntax directly.
Can I export the model to an actual Excel file?
Yes. Zoom Sheets exports to Excel (.xlsx) and CSV format. Complex formulas, formatting, and other workbook elements may vary during conversion, so open the exported file in Excel and verify the model before relying on it.
Will this work for a more complex model, like a full three-statement model or a churn-driven SaaS forecast?
Not as tested here. This walkthrough covers a single revenue line with growth and cost assumptions, a common but relatively contained kind of financial model. More complex structures, like multi-product revenue, tiered pricing, or retention-driven forecasting, sit outside that scope (see “Where This Kind of Model Fits” above).
Does the AI just fill in numbers, or does it actually write formulas that update?
It can draft formulas. In this walkthrough, changing the Fixed Costs assumption changed the displayed values in dependent cells, including Cumulative Net Income and a MATCH-based lookup for the first profitable month. Review the dependencies and test representative assumption changes before relying on the model.