A seed or Series A round closes on the back of a model the investor can defend internally. The deck wins you the meeting, but the model is what gets pulled apart at IC and either holds or doesn't. We see at least one founder a quarter lose a term sheet because their model didn't survive due diligence. Even though their deck was strong, their traction was real, and the partner liked them.

This guide is the structure we ship in Phase 1 of every Vault Catalyst engagement, paired with what investor-side diligence actually looks at and what the public failure modes are.

1. What investors actually check first

When a partner gets your model, they don't read it linearly. They check three things in roughly this order:

Is it driver-based? Do the numbers come from inputs (channel volume, conversion, AOV), or are they typed in directly? A typed-in revenue number signals an associate built it on a Sunday.
Are the assumptions believable? They will pick 3-4 cells and ask “why this number?”. If the answer is anything except “here's the source / here's the benchmark / here's our actual data,” the model loses credibility.
Does the bear case give 12+ months of runway? The first calculation a partner does is dividing your raise by your bear-case burn. If runway falls below a year in the bear, the round either dies or repricesm.

Everything else , the pretty charts, the multi-tab structure, the headcount plan , only matters once these three pass. Get these three right and the rest is optimization. Get any of these wrong and the rest doesn't save you.

2. The structural failure modes that kill rounds

Industry analyses converge on roughly the same list of model failures that founders make at seed and Series A. Public reporting from CFO Bridge, CLA Connect, MicroVentures, and PrometAI overlap heavily on these [1][2][3][4].

Failure 1: Top-down revenue

A model where revenue is “5% of a $10B TAM” signals zero customer understanding. Investors don't want a market-share assumption; they want a sales-process assumption. Top-down models are dismissed in roughly the time it takes to scroll past them.

The fix: build revenue from the bottom up. Number of accounts × average contract value × win rate, with each input traceable to a benchmark or your own data.

Failure 2: Hockey-stick projections without inflection logic

Every founder model shows a hockey stick somewhere. That's fine. What kills the model is a hockey stick with no mechanical explanation for the inflection point. Why does revenue go from $200K MRR to $1.2M MRR in month 18? “We hit product-market fit” isn't an answer; that's a label, not a mechanism.

PrometAI calls this out specifically: “Among the most common startup financial model mistakes, none is as tempting or as damaging as overestimating revenue potential” [3].

Failure 3: False precision

Models that show revenue to two decimal places three years out are signaling that the founder doesn't understand the difference between calculation and confidence. CLA Connect's seven-mistake framework lists this explicitly [2]: precision implies certainty you don't have, and sophisticated investors discount the entire model when they see it.

The fix: round projections to clean numbers (₹1.4 Cr, not ₹1,42,87,500), and use ranges where ranges are honest.

Failure 4: Cash-flow timing errors

A common pattern: revenue is recognized in the month the contract is signed, but cash actually arrives 30-60-90 days later. Founders confuse these and end up with a model that says they have runway when they don't. Capidel calls cash-flow timing “a common issue” that “threatens survival” rather than just fundraising [5].

The fix: model cash, not just P&L. A separate cash flow tab that shows contract-to-cash lag, GST collected and paid out, and AR/AP timing is often the difference between a model that closes a round and one that doesn't.

Failure 5: Single scenario

A model with only a base case is a model with one belief. Investors want to know what happens if you're wrong. The CLA Connect framework calls this “focusing on point estimates” as one of the seven recurring errors [2].

Three scenarios is the minimum: base, bear, bull. The bear is the most important one because that's the one investors will rebuild in their head before they commit.

Failure 6: Headcount untied to milestones

A model that shows 35 hires in year 2 with no linkage between each hire and a specific milestone is a model that says you'll spend the money the investor is giving you on whatever feels right at the time. Investors want every hire mapped to a function and a milestone.

Failure 7: Disconnected from the deck

The deck says we'll hit ₹2 Cr ARR by month 12. The model shows ₹1.4 Cr by month 12. The pitch says we'll spend most of the round on engineering. The headcount plan shows 60% sales hires.

We see this in roughly 70% of founder models we review. It is the single fastest way to lose investor trust without realizing it. Every number in the deck must be traceable to a cell in the model.

3. The seven-tab structure that holds up

A seed or Series A model needs seven tabs, in this order. Anything more is over-engineered; anything less is under-engineered.

Tab 1: Assumptions

Every input the rest of the model pulls from. CAC, conversion rate, average contract value, churn rate, headcount costs by role, gross margin, working capital cycles. Each input has:

The number (current best estimate).
The source (your own data / sector benchmark / clearly stated belief).
The bear and bull versions for scenario switching.
A comment explaining why this number, in 1 sentence.

The assumptions tab is the tab investors spend the most time on after the summary. Most founders bury it; sophisticated founders lead with it.

Tab 2: Revenue build

Driver-based, bottom-up. For SaaS: new logos × ACV × retention. For marketplace: GMV with take-rate. For D2C: visitors × conversion × AOV × repeat factor. Each line traces back to the assumptions tab.

Tab 3: Cost build

COGS, gross margin, then operating costs broken into headcount, marketing, infrastructure, G&A. Headcount mapped to roles, with the start month and salary band tagged.

Tab 4: Headcount plan

A separate tab that lists every hire by month, role, function, and salary band. Investors look at this tab specifically to see whether your hiring plan is tied to the milestones in your deck. If you're raising to hit ₹X ARR and your sales hires don't accelerate before that target, the model is internally inconsistent.

Tab 5: Scenarios

Base, bear, bull. The base case is your honest expectation. The bear case is “everything I'm worried about happens at once.” The bull case is “the right things break right.”

The scenarios tab should drive the rest of the model via a single dropdown switch. Investors will toggle it themselves and watch the runway change.

Tab 6: Cash flow

P&L is not cash. Cash flow shows actual money in and out, including contract-to-cash lag, GST cycles, AR/AP timing, and capex. The cash tab is what tells you when you actually need to raise next, not the P&L.

Tab 7: Summary / outputs

A single page with the metrics investors care about: ARR or revenue trajectory, growth rate, gross margin, burn rate, runway under each scenario, and the milestones the round buys. This tab should be readable as a screenshot, because that's how it gets shared inside funds.

4. The numbers investors will rebuild in their head

Every partner will redo this math while you're talking. Have your version ready, and make sure it matches.

CAC payback period. CAC ÷ (monthly gross profit per customer). Under 12 months is healthy; under 6 months is strong.
LTV:CAC ratio. 3x is the floor. Under 2x is a yellow flag; under 1.5x is a red flag.
Gross margin. SaaS expects 70%+. Marketplace 20-40%. D2C 50%+. If yours is below sector benchmark, lead with the explanation.
Burn multiple. Net new burn ÷ net new ARR. Under 1.5x is good. Over 2x is concerning.
Months of runway. Cash on hand + this round ÷ monthly burn. Bear-case runway is the number that matters at diligence.

Walk into the partner meeting with these five numbers committed to memory. Hesitation on any of them signals that you don't actually know your business at the level a Series A investor needs you to.

5. The bear case is the most important case

Most founders treat the bear case as a formality. Investors treat it as the primary case. The bear case tells the investor:

Whether you understand your downside risks.
Whether the round actually buys the runway you claim it does.
Whether you have a credible response if your business slows.

A bear case that just turns down the growth rate by 20% is a fake bear case. A real bear case includes:

Slower top-of-funnel. 30-40% drop in lead generation, modeled month over month, not as an annual delta.
Worse conversion. 25-35% drop in your win rate or trial conversion.
Higher CAC. 40-60% increase in CAC, which compresses payback.
Slightly higher costs. 10-15% drift in salaries, AWS, ops.
One key customer churning. If you have customer concentration, model the concentration breaking.

Stack all five and check the runway. If it's under 12 months, the round is under-sized.

6. The cap table that holds up at diligence

A messy cap table can kill a round even with a strong model. Carta has documented this directly: “An inaccurate cap table can lead to costly legal fees and can delay or even kill a funding round” [6].

Common diligence-killing cap table issues we see:

Founder shares not vested. Investors want 4-year vesting with a 1-year cliff on founder equity. If you don't have it, expect to negotiate it in.
Phantom equity to advisors. Verbal promises of “1% to so-and-so” that aren't in the cap table create future disputes that scare investors.
SAFE notes not converted. Outstanding SAFEs from prior rounds need to convert at the priced round; investors will model the dilution carefully.
ESOP pool sized wrong. Most institutional investors expect 8-15% pool. Negotiate where the pool dilutes. Pre-money or post-money. Because the difference can move effective dilution by 100-200 bps.
Unclear cofounder splits. 50-50 splits with no decision-making clarity raise governance concerns. Investors prefer slightly uneven splits with a clear CEO.

Use Carta or Vakilsearch (in India) to maintain the cap table. A model that lines up with a clean cap table closes faster than a model that lines up with a spreadsheet you and your CA built last year.

7. The Indian-specific model considerations

A model built for an Indian seed should additionally have:

GST flows. Output GST on revenue, input GST on costs, working capital impact. This is meaningfully different from a US model.
FEMA-aware structuring. If you have foreign investors or plan to, the model should reflect FDI route assumptions.
Bengaluru / Mumbai cost realism. Engineering rates have moved meaningfully in 2026; old benchmarks underprice your build.
TDS withholdings. Operationally relevant to cash timing; investors notice when these are missing.
ESOP perquisite tax mechanics. The tax treatment of ESOPs in India is complex and a sophisticated investor will check whether you've modeled it correctly.

We covered the broader compliance picture in our India seed playbook; the model implications above are the parts founders most often skip.

8. The 5 questions to ask before sending the model

Before you send the model to any investor, run this checklist:

Does every assumption have a source or stated belief? Open the assumptions tab and confirm.
Does the bear case give 12+ months of runway? Toggle the scenario switch and check.
Does the model match the deck? Pick three numbers from the deck and verify they match cells in the model.
Is the CAC payback period under 12 months in the base? If not, lead with the explanation.
Can you walk through the entire model in 7 minutes? If not, it's either too complicated or you don't know it well enough.

9. Common deal-breakers at diligence

Beyond the model itself, these are the diligence findings that have killed rounds we've been part of:

Unrecorded liabilities. Vendor disputes, undisclosed loans, founder personal advances to the company.
Customer concentration above 30% in a single account without a long-term contract.
Revenue recognition errors. Recognizing setup fees upfront when they should be amortized.
Founder transactions. Founders charging their own consulting fees through related entities. Looks like self-dealing even when it isn't.
Compliance gaps. Missing PF / ESI for employees, GST returns delayed, TDS not deducted.

Goldenegg Check has documented several of these as recurring deal-breakers [7]. None of them are about the model directly, but all of them surface during model review and kill the round.

10. What to do in the next 14 days

Open your current model. Count the cells with hard-coded values that should be formulas. Fix them.
Build the bear case with the 5-stack approach above. Check 12-month runway.
Reconcile every number in your deck with a cell in the model. Fix mismatches.
Run the 5-question checklist before sending the model anywhere.
If your cap table is messy, get it on Carta or Vakilsearch this week. Investors will ask for it before signing.

If the work above is too much to do alongside running the company, this is exactly what we ship in the first month of Phase 1. Book a discovery call if you want to talk through where your model is now and what would need to change.

The financial model that survives investor diligence: a 2026 builder's guide