# Build a Budget

Start from last year, enter figures per account, and phase them to match how your business actually behaves.

Phasing is the step that decides whether your monthly variances mean anything.

![The New Budget screen, where the budget is named and given a fiscal year and a period before figures are entered against each account.](https://www.sorviai.com/help/budget/new-budget-form.png)

*The period you choose is the grain every later comparison is made at.*

## Build it

- **Choose the fiscal year and the period**

  The fiscal year sets the span; the period sets the grain. Monthly, Quarterly, Half-Yearly and Yearly are offered, and monthly is what makes a variance report worth reading.
- **Start from last year's actuals**

  Far quicker than a blank sheet, and grounded in something real.
- **Choose which account groups to include**

  Income and expense are the usual pair, and the form can also take assets, liabilities and equity if you budget the balance sheet. Leave those off unless you know you want them.
- **Enter figures per account**

  One figure per account per period.
- **Phase them properly**

  Seasonal businesses ruin their own variance reports by spreading annual figures evenly.
- **Save and compare monthly**

  A budget nobody looks at monthly is decoration.

> **Tip:**
>
> **Phasing matters more than accuracy**
>
> A roughly right annual number, phased to match your actual seasonality, produces far more useful monthly variances than a precise annual number spread evenly.

## Phasing patterns

| Account | Phase by |
| --- | --- |
| Sales | Last year's monthly shape, not one twelfth. |
| Cost of goods sold | The sales phasing, at your expected margin. |
| Rent and subscriptions | Evenly. They genuinely are. |
| Salaries | Evenly, plus the months with a bonus or a pay rise. |
| Marketing | When the campaigns are, not when the money is available. |
| Utilities | Last year's shape. Winter is not one twelfth. |

## Reforecasting

When circumstances change materially, create a second budget rather than editing the first. Comparing against a figure that has been quietly adjusted all year proves nothing, and the original is the only thing that shows what you expected before you knew.