Realized yield on a business account: why average balance misleads

A quoted yield is an input to a calculation nobody performs. The output you actually want is different: what did this account pay my business, after charges, on the money I really kept in it, over a period that has finished.

That number can be computed exactly, from documents you already have. It requires no rate table and no projection, and it is the only figure that lets you compare one account against another honestly. Nothing here quotes a rate, and no calculator runs on this page. What follows is the arithmetic and, more usefully, where the inputs go wrong.

What to take away

The five inputs, and where each comes from

All of these live in your statements or your account agreement. None come from a marketing page.

Formula one: what the account actually paid

For one completed period:

R = (I - G) / A

That is your realized yield for the period, net of charges. To put periods of different lengths on the same footing:

R annualized = R x (days in a year / P)

This is an approximation, because it does not model compounding inside the period. It is close enough for comparison, and being explicit about the approximation is better than pretending to a precision the inputs do not support.

Run this for the last several completed periods rather than one. A single period can be flattered by a month where every condition happened to be met.

The average balance problem

A is where most calculations quietly fail. There are several ways to compute an average balance and they can produce meaningfully different answers for the same account.

Ask two questions of your institution and write the answers into the account record: which basis is used for interest accrual, and which basis is used for testing any minimum. What has to be disclosed about the method, and about the yield convention it is quoted under, is published as Regulation DD. They are not always the same, and a business that dips on payroll day can satisfy the first while failing the second every single cycle.

While you are asking, find out on what basis a year is counted for accrual, and use the same basis on both sides when you compare accounts. Comparing figures built on different conventions produces a difference that is arithmetic rather than real.

Formula two: tiered balances

A tiered account pays different rates on different bands, and there are two structures. Work out which one you are in before you calculate anything.

In a banded structure, each slice of your balance earns its own rate, so the blended result is the weighted average across bands:

Blended = sum over bands of (balance in band x band rate) / total balance

The advertised number is normally the top band. If most of your money sits below it, the blended figure lands nowhere near the headline, and that gap is the whole reason the calculation is worth doing. For orientation on what national deposit rate data is and how it is compiled, the regulator publishes its national rates material; it will not price your account, and it will remind you that a headline is a different object from a realized yield.

In a threshold structure, crossing a level moves the entire balance to a different rate. Here the useful arithmetic is a break even: how much extra balance you would need to commit to cross the line, and what that money is giving up by sitting here rather than somewhere with better access. Crossing a threshold with money you will need next month is a common and expensive move.

Formula three: the cost of qualifying

If a rate or a waiver requires you to hold a floor, that money is committed capital, not free balance. Put it in the denominator:

Effective yield on committed money = (I - G) / (A + any part of Q you would not otherwise hold here)

This single adjustment reverses rankings regularly. An account paying well on a small slice while requiring a large idle floor can return less on the total commitment than an account with a duller headline and no conditions at all. The related exercise on the operating side, where a floor is held to avoid a monthly charge, is the same idea applied to charges rather than interest, and the charge categories to check are set out in how charges group together.

Comparing two accounts without fooling yourself

Where the arithmetic goes wrong

The short version

Take a finished period, divide net interest by the balance you actually held including anything you were forced to strand, and compare that figure with the same figure for the alternative. The headline rate is a claim about a hypothetical business. The number you compute is a fact about yours, and it is the one worth acting on, as the wider framing on reading a quoted yield sets out.

Common questions

Is there a calculator on this page?

No. The formulas are three lines and every input has to be taken from your own statements. A spreadsheet with one row per period, holding interest credited, charges, average balance and days, gives you a running record and takes a few minutes a month.

What if my statements do not show an average balance?

Ask for it, or build it from daily balances if your export includes them. If neither is available, that itself is worth knowing, because you cannot verify what you were paid without it.

How do I compare this against a product with a fixed term?

Compute the realized yield the same way for the term product, then add the thing the formula cannot show: the cost of not being able to reach the money. If the balance might be needed early, price the penalty and the disruption before comparing anything.

How often should I run it?

Every period at first, then quarterly once the account is behaving predictably. Always run it after a rate change notice, and after any cycle where a qualifying condition failed. Where the balance belongs at all is a question of cash layers rather than yield, and that is covered in matching money to the job it does.