Financial analysis

An interactive 36-month money model of your project — KPIs, pro-forma statements, sensitivity and adjustable levers, anchored by live web research. · Optional

The financial analysis is the last stop in the Research phase and the deepest. Where the business case asks should we build this?, this asks does the money work? — it turns your idea into a working three-year financial model you can stress-test. It's entirely optional: run it when you want to plan runway, pressure-test pricing or pitch numbers to a co-founder or investor; skip it for internal tools or throwaway prototypes.

Where it lives

It's the second tab on the Research page, next to the Business Case. Open your project, click Research, then the Financial Analysis tab. (The old /financial-analysis link redirects here.) Market validation and the go/no-go verdict stay on the Business Case tab — this tab is purely the money model, derived from the same summary and business case.

Generating the model

1. On the empty state you'll see a readiness meter (scored out of 100) telling you whether your project description has enough for a useful model.
2. Optionally add monetization details in the notes box — pricing you charge or plan to, revenue streams (subscriptions, commissions, one-off sales…), known costs or margins, and current traction (users, orders, MRR). Optional, but the more you give the analyst, the sharper the model.
3. Click Generate Financial Model. Generation runs live web research to anchor its assumptions and typically takes 3–8 minutes — it runs in the background, so you can leave the tab.

⚠️ Watch out: Like the business case, this needs a populated project summary and a description with real substance. If either is missing you'll see a pre-flight message instead of the Generate button.

Reading the dashboard

Once generated, the header shows when it was built, which model ran it and how many web sources it searched. A "weakly grounded" badge appears if fewer than six web searches anchored the numbers — treat those values as assumptions rather than facts. As the header puts it: seeds are researched anchors; every projection is computed locally. That's the key idea — the AI researches the starting values, but the whole 36-month simulation runs in your browser, so it recalculates instantly as you drag things.

The dashboard has a sticky lever panel on the left and the model on the right:

• Levers & scenario presets — grouped sliders for the model's inputs (prices, growth, churn, costs, and so on). Three preset buttons — Base, Optimistic, Pessimistic — snap every lever at once; drag any slider and you're in a custom scenario until you pick a preset again. Your slider positions are saved automatically.
• KPI tiles — headline numbers such as revenue at months 12 and 36, cash at month 36, cash trough, gross margin, breakeven month, runway, 3-year founder income, 3-year ROI, and unit economics (CAC, LTV, LTV:CAC — target ≥ 3×). Tiles turn red when a number goes negative and green when unit economics look healthy.
• Milestones — chips marking when key events (like breakeven) land, dated from the generation month.
• Charts — revenue and cash over time with all three scenarios overlaid, a cost-stack breakdown, and per-stream economics (Recharts).
• What moves the outcome (tornado) — ranks the levers by how much a ±20% move swings month-36 cash, so you can see what actually matters.
• Sensitivity grid — a heat-map of month-36 cash (and breakeven month) as the two most decisive levers move together.
• Pro-forma statements — an annual income statement across the three years and a full month-by-month table.
• Findings & assumptions — a written narrative plus every assumption tagged ANCHORED (backed by a cited web source), USER (from your notes) or ASSUMED (a model default), with links to the sources.

Updating the model

Click Regenerate to open a dialog where you can add or revise monetization notes. The analyst keeps the stream structure and lever keys stable where it can, so your saved slider positions survive and it only re-researches what your notes change. The existing model stays interactive while the new one builds.

⚡ Power-user hints

• Drag, don't regenerate. For "what if we charged more?" questions, just move the price lever — the model recomputes locally and instantly. Regenerate only when the structure changes (a new revenue stream, a different business model).
• Start from the tornado. It tells you which two or three levers actually move the outcome; focus your attention (and your real-world validation) there.
• Trust ANCHORED over ASSUMED. Scan the assumptions list — anchored figures cite a source; assumed ones are defaults you should sanity-check against your own numbers.
• Mind the grounding badge. A "weakly grounded" model is fine for direction but not for a board deck — feed it more monetization detail and regenerate.

↔ The traditional way

Building this by hand means a founder or finance analyst living in a spreadsheet for a day or more: wiring up a 36-month pro-forma, hand-typing scenario tabs, building sensitivity tables with data-tables, and formatting a tornado chart — then re-doing the wiring every time an assumption changes, and hunting down market benchmarks separately. VibeMap researches the anchor values for you, builds the model, scenarios, sensitivity grid and tornado in one pass, and lets you explore trade-offs by dragging a slider instead of rebuilding formulas.

What's next

• Happy the numbers work? Head into the spec: Generating personas.
• When your spec is complete, consolidate everything: Prepare for development.