Budget management, planning and analysis

This chapter is not a product guide. It explains budget management itself, at a level that someone who does not use PORTWARE can read and learn from. Chapter 21 shows their equivalent in PORTWARE; what is described here holds in any system.

What a budget is and is not

A budget is not a forecast but an agreement: a commitment on how resources will be used.

• A forecast answers "what will happen?".
• A budget answers "what did we agree to do?".

A good system keeps the two apart. In a company that does not, the budget silently turns into a forecast as the year goes on, and nobody has promised anybody anything.

Types of budget

The cash budget standing apart is no accident: profitability and cash are different things. A profitable company may be unable to pay because its receivables are collected late.

Preparation approaches

In practice most companies work in a mix: management gives a frame, units fill it in, and reconciliation happens below. What matters is not which approach is chosen but that the choice is conscious — "let's add 10% to last year's" is an approach too, it just has no name.

Scenarios and the rolling forecast

Scenarios: keeping base, optimistic and pessimistic plans separately. A single figure hides uncertainty.

Rolling forecast: during the year the plan of closed months is replaced by actuals and a new month is added at the front. So there is always a twelve-month view.

The rule: the annual budget is frozen, the forecast lives. Trying to keep the budget alive deletes the agreement; freezing the forecast is blindness.

Good practice is to change only a few key assumptions in each scenario and to tie each scenario in advance to an action ("if the pessimistic case happens we freeze hiring"). The rolling forecast keeps the base scenario current, the scenarios draw the range around it; the two work together, not instead of each other.

Budgeting with a short history

A business with a few months of data cannot write twelve months as "last year + %". What to do:

1. Take the monthly average per stream from the actual months: product family, service, project.
2. For each stream estimate units and price separately, and the margin separately.
3. Leave an assumption the data cannot support empty; do not write zero. Zero says "it will never happen"; a gap says "we have not decided yet" and keeps visible the question to ask the plan owner.
4. Build an optimistic and pessimistic range around the base plan and compare with the actuals every month. The uncertainty of a short history is shown honestly with a range, not a single figure.

Variance analysis

Variance = actual − plan. The direction of the sign is read by the measure: spending more on an expense and earning less on revenue are both worse than plan.

Materiality. Not every variance is explained. A threshold is set (for example 10% or a certain amount) and only those above it are justified. A process that tries to explain everything turns into a process that explains nothing.

Price / volume split. This is the gift of a driver-based budget:

volume effect = (actual units − plan units) × plan price
price effect  = (actual price − plan price) × actual units

Together they are the whole variance. When several products are examined together a third effect appears: the mix effect, which arises when the split of sales between expensive and cheap products changes. Splitting streams separately leaves the mix inside each stream.

Case

Opening up a variance

Situation

Plan 1,000 units × 50 TL = 50,000 TL. Actual 900 units × 54 TL = 48,600 TL. Total variance −1,400 TL, and since this is a revenue item it is bad. (The figures are an example.)

What was done

The variance was split in two. Volume effect (900−1,000)×50 = −5,000 TL. Price effect (54−50)×900 = +3,600 TL.

Result

The single figure said "we fell behind by 1,400". The split gave two separate messages: to the sales team "why did we sell less", to pricing "the increase worked". The same variance, two different decisions.

A variance with no reason written is a figure; a variance with its reason written is a decision. The real product of the month-end close is not the table but the sentences written in its margin.

Commitment accounting

Spending starts at the order, not at the invoice. A company that follows its budget only by invoices sees the overrun when the invoice arrives — that is, after the fact.

So serious budget tracking keeps three figures together:

remaining = plan − actual − commitment

The control levels come from here too: watch, warn, block. Most companies are advised to start with warn: blocking before the data matures stops real work because of an under-budgeted item and ends trust in the system itself.

The difference between KPI and OKR

• A KPI measures health and is watched constantly.
• An OKR aims at change and is periodic.

In a well-built system an OKR sets a KPI's target; they are not separate worlds. Details in chapter 22.

The FP&A cycle

Financial planning and analysis is not a report but a closing circle:

Strategy → objective (OKR) → budget → control → month-end close →
variance analysis → forecast update → revision → quarterly scoring → Strategy

In most companies the circle fails to close between variance analysis and the forecast update: the variance is found, discussed and not fed back into the plan. The next month the same variance is discovered again.

Eight common mistakes

1. Having no single definition of "the right figure".
2. Writing the budget at the start of the year and never opening it again.
3. Using zero instead of "unknown".
4. Changing an approved figure in place — that is, deleting the agreement.
5. Looking only at invoices and not seeing the commitment.
6. Confusing profit with cash.
7. Confusing progress with confidence.
8. Leaving variances without reasons.

Every item on this list is the reason for a screen design in chapters 20–22: the measure registry closes the first, revisions the fourth, the commitment column the fifth, the cash panel the sixth, OKR's separate confidence field the seventh, the variance explanation the eighth.

Sources and further reading

• Price, volume and mix analysis: Zebra BI — Price Volume Mix Analysis
• Scenario planning: AFP — Scenario Planning, FP&A Trends — Scenario Planning in FP&A
• Rolling forecast: Cube — Rolling Forecast Best Practices
• Commitment control: Oracle PeopleSoft — Commitment Accounting

Summary of this chapter

• A budget is an agreement, a forecast an expectation; the two are kept apart.
• The choice of approach should be conscious; a driver-based plan answers the scenario question.
• Variance is filtered by a materiality threshold, split into price and volume, and its reason is written.
• Spending starts at the order; remaining = plan − actual − commitment.
• FP&A is a circle and most often breaks between variance and forecast.