The Numbers Your Accountant Uses Are Not Yours

Two reports, two different businesses

Every month I get a set of management accounts from my accountant. A profit and loss statement, a balance sheet, sometimes a cash flow summary. And every day I look at a dashboard that shows MRR, churn, conversion rate, cohort retention. Both claim to describe the same business. They rarely agree, and that used to bother me until I understood why.

They’re not measuring the same thing. Management accounts exist to answer a specific, narrow question: what happened to the money, in a form that satisfies accounting standards, tax authorities, and anyone who might audit the business later. Analytics exist to answer a different question: what is happening to the customers, right now, in enough detail that I can make a decision about product or pricing before the quarter closes. One looks backward with precision. The other looks sideways and forward with less precision but more speed. Neither is wrong. They’re just not interchangeable, and treating them as interchangeable is where solo operators get confused about their own business.

What management accounts are actually built for

A profit and loss statement groups revenue and expenses into categories that make sense for compliance and tax, not for product decisions. Revenue on the P&L is usually recognized on an accrual basis, meaning income gets recorded when it’s earned, not necessarily when the cash lands in the bank. If a customer pays for a year upfront, accrual accounting spreads that revenue across the twelve months it covers, sitting the unearned portion on the balance sheet as deferred revenue. That’s the correct way to represent it for financial reporting. It also means the P&L for the month that customer paid will not show the full payment, and a rolling revenue analytics view that just sums up cash received that day absolutely will.

The balance sheet tells you what the business owns and owes at a single point in time. It doesn’t tell you why a customer churned or which acquisition channel is producing customers who stick around. It was never built to. It’s a snapshot for solvency and ownership questions, useful for a bank, an investor, or your own peace of mind about whether the business could cover its obligations if things went quiet for a while.

None of this is a criticism of accountants. The rules they follow, whatever standard applies in your jurisdiction, exist so that financial statements mean the same thing across different businesses and different years. Consistency is the whole point. But consistency across time and comparability across businesses is a different goal from telling you, this week, whether your funnel is converting worse than last month.

What analytics are actually built for

Churn, MRR, CLV, conversion rate, cohort retention. None of these appear as line items on a P&L or balance sheet, because they aren’t accounting concepts. They’re derived from event-level data: who signed up, who paid, who cancelled, who clicked what, when. MRR (monthly recurring revenue) is a normalized, forward-looking approximation of what you’d collect in a typical month if nothing changed, which is why it needs adjusting for annual plans, discounts, and one-time charges to mean anything. Churn is a rate calculated from customer records over a defined window, and that window is a choice you make, not a rule handed down by a standards board. Cohort retention groups customers by signup month and tracks what fraction are still active N months later, which is a lens built specifically for spotting whether a product or onboarding change helped or hurt.

These numbers are useful precisely because they’re granular and fast. You can look at last week’s signup cohort and see early warning signs long before that cohort’s behavior shows up anywhere on a financial statement. But that speed comes at the cost of the rigor accounting has. There’s no external standard enforcing how you define “active customer” or “churned.” Two SaaS businesses can both report “5% monthly churn” using different definitions and mean different things. That’s not a flaw to fix once and forget. It’s a permanent tradeoff, and it means your own definitions need to stay consistent over time more than they need to match anyone else’s.

Where the two disagree, concretely

Take a subscription business that offers monthly and annual plans. A customer buys the annual plan for a lump sum. On the accountant’s P&L, that revenue is recognized a twelfth at a time over the year. On a naive analytics dashboard that just sums payments received, that customer creates a spike in the month they paid and nothing afterward until renewal. If your analytics tool doesn’t normalize annual payments into a monthly-equivalent figure, your MRR chart will look like it has a heartbeat that has nothing to do with actual business momentum. This is a common, fixable data quality issue, but only if you know to look for it. I’ve had months where MRR looked like it jumped and the honest explanation was three people renewing annually in the same week, not new growth.

Refunds and chargebacks are another point of divergence. A refund issued this month for a sale made three months ago hits the accountant’s books as an adjustment tied to the period that makes sense for tax purposes. On a churn or revenue analytics view, how you attribute that refund, to the month of the sale or the month of the refund, changes your historical churn numbers depending on which convention your tooling uses. I’ve gone back and forth on this and settled on attributing refunds to the month they happen, because that’s when I need to notice the pattern, but I don’t pretend that choice is the only correct one.

Customer Lifetime Value is probably the clearest example of a number that has no accounting equivalent at all. CLV is an estimate, built from assumptions about average revenue per customer and average retention, projected forward. It’s a planning tool. It is not, and cannot be, a verified historical fact the way a bank balance is. Any CLV figure you see, including your own, is only as good as the churn and revenue assumptions feeding it, and it should be treated as a working estimate you revisit, not a fixed truth you build permanent decisions on.

Why this matters for a one-person operation

When you’re running the business alone, it’s tempting to want one number that tells you how things are going. There isn’t one. Ask your accountant for churn and you’re asking them to do work that isn’t part of what management accounts are built to show. Ask your analytics dashboard whether you owe tax on last quarter’s revenue and you’re asking a tool that has no idea what your deductible expenses were. The two systems need to stay separate because they’re doing separate jobs, and neither one should be quietly overriding the other.

What I do check, periodically, is whether the two roughly reconcile over a longer window. Not month to month, where timing differences are expected and normal, but over a quarter or a year, the total revenue in my analytics should land in the same neighborhood as the revenue on the P&L once accrual timing washes out. If they drift far apart over a longer period, that’s usually a sign something in the analytics pipeline is broken, a webhook silently failing, a double-counted event, a metric definition that quietly changed. That’s a data quality check, not a tax question, and it’s one I can run myself without needing to interpret anything an accountant would need to weigh in on.

The two sets of numbers were never meant to be the same report. Knowing which one to reach for, and what each one can’t tell you, is the actual skill. Neither replaces judgment. Both are just inputs to it.

If you want more breakdowns like this on the metrics and methods behind running a business on real data, you can find the rest of the collection here.

Get new guides and videos first — join the Telegram channel.