ComparisonStrategy & Corporate Finance
Zero-based budgeting vs traditional budgeting: choosing a method for each kind of cost
Zero-based budgeting rebuilds spending from a blank page each cycle, while traditional incremental budgeting adjusts last year's figures up or down. Neither suits every cost. ZBB earns its effort on discretionary and indirect spend that nobody has questioned in years; driver-based rolling forecasts suit volume-linked costs; activity-based methods suit shared services. This comparison sets the four approaches side by side and shows how to choose one cost category at a time.
On this page
- Four budgeting approaches compared on effort, data and durability
- Why costs return after a cut, and which methods treat the cause
- Full ZBB, ZBB-lite or neither: matching the method to the spend
- Running a zero-based cycle: packages, decision units and challenge sessions
- What AI-assisted spend classification changes, and what stays a judgment
- A professional services firm applies ZBB to indirect spend only
- Questions and answers
- Sources
Four budgeting approaches compared on effort, data and durability
Read across a row to see what each approach demands and what it returns.
| Dimension | Incremental | Zero-based (ZBB) | Activity-based | Driver-based rolling forecast |
|---|---|---|---|---|
| Starting point | Last year's actuals or budget, plus or minus an adjustment | A blank page: every cost package justified from its purpose | The activities the organization performs and what drives their cost | Operational drivers such as volume, headcount or price, re-forecast each period |
| Effort per cycle | Low; mostly negotiation over the adjustment | High in the first cycle, lower once cost packages exist | High to set up, moderate to maintain | Moderate once drivers are agreed; the model does the arithmetic |
| Data it needs | General ledger history | Spend classified by category, supplier and owner, plus service levels | Activity volumes and how time or resources are consumed | Reliable driver data and tested links between drivers and cost |
| Savings durability | Weak: inefficiencies in the base roll forward | Strong while package owners stay accountable; fades if governance lapses | Good for shared services where internal demand can be managed | Not a savings method; it keeps spend proportionate to activity |
| Cultural impact | Comfortable; rewards spending the full allocation | Demanding; budget holders must defend every line | Exposes cross-subsidies between units, which can be contentious | Moves conversations from annual targets to drivers and triggers |
| Best-fit cost categories | Stable, contracted or regulated costs | Discretionary and indirect spend: marketing, travel, professional fees, software, facilities | Shared services, IT and support functions with chargeback | Volume-linked costs: direct labor, logistics, cloud consumption |
| Typical failure mode | Baseline never questioned, so budgets ratchet upward | Run once as a cut, then abandoned | Model grows too detailed to maintain | Drivers chosen for convenience rather than causality |
The table compares typical characteristics of each method, not measured outcomes. Most organizations end up running a mix.
Why costs return after a cut, and which methods treat the cause
A spending freeze or a headcount reduction lowers costs without changing why they arose. The demand for the work remains, so the cost reappears as contractors, overtime, tools bought on corporate cards or a supplier contract renewed at its old scope. The cost transformation insight on our capability page makes the same argument: lasting results depend on tackling process inefficiency, organizational complexity and technology fragmentation1.
Incremental budgeting cannot fix this because it starts from the inflated base. ZBB attacks the problem directly by forcing a decision on whether each activity should exist and at what service level. Activity-based approaches attack it by making internal demand visible and chargeable, so business units consume less. Driver-based forecasts do not attack it at all; they keep spend proportionate to activity, which is valuable but different.
A practical test is to ask, for each cost category, what would make it grow again next year. If the answer is a decision someone takes, such as hiring another agency, zero-based challenge works. If it is volume, use drivers. If it is internal demand from other units, use activity-based allocation.
Full ZBB, ZBB-lite or neither: matching the method to the spend
- If
Discretionary spend is large, spread across many budget holders and has not been challenged for several cycles.
ThenRun full ZBB on those categories: build cost packages, define service levels and hold challenge sessions.
Fragmented spend hides duplication that only a ground-up review exposes.
- If
You need results within one planning cycle and the finance team is small.
ThenUse ZBB-lite: apply zero-based challenge to the few cost packages with the most spend and the weakest justification, and keep incremental budgeting elsewhere.
Most of the benefit usually sits in a handful of categories.
- If
Costs move mainly with volumes such as orders, shipments, users or compute.
ThenBuild a driver-based rolling forecast and review driver efficiency instead of defending budget lines.
Rebuilding a volume-linked cost from zero each year repeats work a driver model already does.
- If
Shared services are over-consumed because business units never see their cost.
ThenIntroduce activity-based allocation or chargeback before attempting ZBB in those functions.
Cutting supply while demand stays free tends to produce queues and workarounds rather than savings.
- If
Costs are contracted, regulated or committed for several years.
ThenLeave them on incremental budgeting and revisit them at renewal or renegotiation.
There is no decision to take inside the cycle, so challenge sessions would only consume time.
Running a zero-based cycle: packages, decision units and challenge sessions
A zero-based cycle is a planning process with its own calendar, not a spreadsheet exercise. These are the stages for one cycle on the categories you chose above.
Classify the baseline
Map the last full year of spend to a consistent category taxonomy, recording supplier, cost center and accountable owner for each line. Unclassified spend is where duplication hides, so resolve it before anything else.
Define cost packages and their owners
Group spend into cost packages, such as all external marketing agencies or all collaboration software, and give each one an owner who answers for it across the organization, separate from the budget holders who consume it.
Set decision units and service levels
For each package, describe the minimum service level, the current level and one or two enhanced levels, with the cost of each. This turns a budget argument into a choice about what the organization wants to buy.
Hold challenge sessions
A small panel, typically the CFO, the package owner and a business leader, reviews each package against its options and questions the volume, price and policy assumptions behind it. Decisions and their reasons are written down.
Build the decisions into budgets and controls
Load approved package budgets into the planning system and change purchasing policies, approval limits and catalog rules so the decisions hold in everyday buying, not only in the plan.
Track monthly and re-base each year
Report spend by package against the approved level every month, investigate variances with the package owner, and refresh the baseline in the next annual plan rather than repeating the whole analysis.
What AI-assisted spend classification changes, and what stays a judgment
The most expensive part of a first ZBB cycle is classification: matching thousands of ledger lines, purchase orders and card transactions to categories when descriptions are inconsistent and supplier names vary. Machine-learning classifiers and language models can propose a category for each line, normalize supplier names and send uncertain lines to a person for review. The spend taxonomy guide covers how to build and govern the categories themselves.
Two further uses repay the effort. Anomaly detection flags spend that breaks a package's pattern, such as a new supplier in a consolidated category, so owners investigate early. Scenario models price each service-level option and show how it shifts with volume, which shortens challenge sessions.
What stays human is the decision itself: which service level to buy, which activities to stop and how to treat the people affected. A model can show that three teams pay for overlapping research subscriptions; it cannot decide whose needs come first. Our cost transformation offering aims to redesign cost structures rather than simply cut1, and in that framing the AI work supplies evidence, not authority.
A professional services firm applies ZBB to indirect spend only
Questions and answers
How often should a company run zero-based budgeting?
A full ground-up review of a category is usually worth repeating only when something has changed: a new strategy, an acquisition, a shift in working patterns or evidence that spend has drifted back. In between, keep the cost packages and owners, track spend against the approved service levels each month and re-base packages during the annual plan. Repeating the full analysis every year on every category exhausts budget holders.
What is the finance team's role in a ZBB program?
Finance designs the process, owns the baseline and the cost package register, prepares the service-level costings and facilitates the challenge sessions. It should not own the spending decisions. Package owners and business leaders make those choices and remain accountable for them. When finance becomes the only voice arguing for reductions, decisions tend to unravel as soon as the program ends.
Can zero-based budgeting be combined with rolling forecasts?
Yes, and the combination is common. ZBB sets the approved level for discretionary packages at the start of the year, while the rolling forecast re-projects volume-linked costs each period from their drivers. The forecast then compares actual package spend with the approved level and flags drift. The two methods answer different questions: what we should spend on an activity, and what we will spend given current activity.
Is zero-based budgeting suitable for a fast-growing technology company?
Partly. Growth companies rarely benefit from defending every line, because most spend follows hiring and customer growth. But categories such as software tools, non-production cloud environments, contractors and marketing programs grow quickly and often without a clear owner. Applying zero-based challenge to those categories and driver-based forecasting to the rest keeps growth funded without letting overhead outpace the business.