Why does a hard-coded plug fail in diligence?

A plug is a manually typed number that forces the balance sheet to balance instead of letting it balance by construction. Investors spot it by changing an assumption: in a real integrated model every statement moves; a plugged model breaks or stops tying, revealing that the balance was never real.

Why must SaaS metrics be computed from the same build as the statements?

If metrics live on a separate tab with their own assumptions, they drift from the P&L. An investor recomputes NRR or CAC payback from your revenue build and finds the gap. Deriving every metric from the same ARR bridge and S&M line means there is nothing to reconcile.

What is the most common revenue-build error?

Recognizing a full year of revenue on the ending customer count instead of the average. Customers who join mid-year weren't paying all year, so using the ending count can overstate first-year revenue by 50% or more.

Home · Guides · Model mistakes in diligence

Guide · ~8 min read

Model mistakes that get caught in diligence

A partner doesn't read your model top to bottom. They poke it in five or six specific places, and if it fails any one of them, they quietly downgrade everything else. Here are the recurring errors — the ones that show up again and again — with how each gets caught and the fix.

1. The hard-coded plug that forces the balance

The error: the balance sheet doesn't tie, so somewhere a number gets typed in — "other assets," a mystery equity line — to force assets to equal liabilities plus equity.

How it's caught: an investor changes one assumption (say, churn) and watches. In a truly integrated model, every statement moves and it still ties. A plugged model either breaks or keeps a suspiciously round balancing figure — a dead giveaway that the balance was never real.

Fix: never plug. Build the cash line from the cash-flow statement and retained earnings from accumulated net income, and the balance sheet ties by construction. See how a model ties out.

2. Metrics that contradict the statements

The error: NRR, CAC payback and LTV:CAC are computed on a separate tab with their own, friendlier assumptions — so the metrics tab tells a better story than the P&L.

How it's caught: the investor recomputes NRR from your ARR bridge, or CAC payback from your S&M line, and gets a different answer than your metrics tab shows. Now every number in the model is suspect.

Fix: derive every metric from the same build as the statements. Your NRR should be the expansion and churn lines from the ARR bridge; your CAC should be the S&M line divided by the new logos in the revenue build. Nothing to reconcile because nothing was computed twice.

3. Revenue on ending customers, not average

The error: recognizing a full year of revenue on the year-end customer count — even though customers who signed in month 11 only paid for two months.

How it's caught: the arithmetic doesn't foot. If you started with 14 customers and ended with 55, you did not earn a full year of revenue on 55 customers.

Worked

Correct: average of 34.45 customers × $2,400 × 12 = $992,126. Wrong (ending count): 54.9 × $2,400 × 12 = $1,581,120 — a 59% overstatement of first-year revenue.

Fix: recognize revenue on average customers (or better, monthly cohorts). See the revenue-build guide.

4. CAC payback out of revenue, not gross profit

The error: dividing CAC by monthly ARPA instead of by monthly gross profit, which understates payback and flatters efficiency.

Worked

CAC $11,000. Correct (gross profit $1,776/mo): 6.2 months. Wrong (revenue $2,400/mo): 4.6 months — understated by ~26%. The first "gross or net?" question exposes it.

Fix: pay CAC back out of gross profit — CAC ÷ (ARPA × gross margin). See CAC payback done honestly.

5. A hockey stick with no driver behind it

The error: revenue growth typed straight into a cell — "150% a year" — with nothing underneath it. The chart looks great; the logic is missing.

How it's caught: "Where does the growth come from?" If the answer is "the growth-rate assumption," the model is decoration and the meeting cools.

Fix: make growth an output. Drive it from new-logo adds tied to sales capacity, ARPA, churn and expansion. Then the growth rate is defensible because it's the consequence of assumptions you can point to.

6. LTV:CAC built on an optimistic churn number

The error: a 20×+ LTV:CAC that rests entirely on a churn assumption far below what the company has actually seen.

How it's caught: the investor asks what churn drove the LTV, then asks what your actual churn has been. If the two don't match, the whole ratio evaporates.

Fix: use your observed churn, show it plainly, and don't over-index on the ratio. A defensible 3×–5× beats a fragile 16×.

7. Working capital that doesn't flow to cash

The error: accounts receivable, payables or deferred revenue sit on the balance sheet but their period-over-period changes never appear in the cash-flow statement — so cash is overstated.

How it's caught: the model stops tying, or cash grows faster than net income and financing can explain.

Fix: every working-capital balance needs a matching "change in…" line in the cash-flow statement, driven by the same days-sales/days-payable assumption. Increase in an asset uses cash; increase in a liability is a source of cash.

The checklist a partner runs

Change one assumption — does every statement move and still tie?
Recompute NRR and CAC payback independently — do they match the metrics tab?
Is first-year revenue on average, not ending, customers?
Is CAC paid back out of gross profit?
Where does growth come from — a driver or a typed rate?
What churn drives LTV, and does it match reality?
Does working capital flow through to cash?

The meta-lesson

Every one of these mistakes has the same root: a number computed in one place that doesn't agree with the same number computed another way. An integrated model — where the statements, metrics and valuation all come from one driver-based build — makes these contradictions structurally impossible, which is exactly why investors trust the format.

Start from a model that can't make these mistakes.

ModelKit is a single integrated build: statements, SaaS metrics and valuation method all derive from the same drivers, and 217 automated checks (including the balance-check every year) run on every build. Change an assumption and everything recomputes — and still ties.

Request the kit — $149 Open the worked sample

Educational material only — not investment, legal, financial, accounting, or tax advice, and not a valuation opinion. Figures are illustrative outputs of a deterministic model.

← All guides · Start with: 3-statement model →