What Six Decades of Zero Based Budgeting Teach a Startup

On 5 November 2015 Kraft Heinz’s chief executive, Bernardo Hees, told analysts that “ZBB is a systematic approach not a one-time event”. His company had promised “$1.5 billion of pre-tax savings by 2017”, to be “sourced from our organization, footprint and zero-based budgeting productivity plans” (Kraft Heinz 10-Q, 2015). It reported about $1,725 million by the end of 2017 (10-K, 2017).

By then, the SEC later found, “the procurement division had largely exhausted its ability to extract synergies from the merger”, yet management “did not adjust the internal targets” (SEC order, 2021). In 2017 and 2018 procurement staff manipulated 54 of about 59 supplier transactions to book savings early; the restatement corrected $208 million, and the company paid a $62 million penalty (SEC, 2021).

Zero based budgeting (ZBB) asks each manager to justify spending from a minimum, in ranked “decision packages”, instead of adjusting last year’s numbers. Trials since 1962, from the US Department of Agriculture to Kraft Heinz, found the ranking useful and repeat savings hard to find. A startup should zero-base once, at a change point, and rank its margin every quarter.

The household method of the same name, “your income minus your expenses equals zero every month” in Ramsey Solutions’ words, is a different tool.

Three ways to build next year’s spending, and what each asks of managers

	Incremental budget	Full zero-base pass (Pyhrr, 1970)	Quarterly ranking of recent and marginal lines (our proposal)
Starts from	Last year’s spending, adjusted	A minimum level per unit, typically 20% to 40% below current spending	The ranking from the last full pass
Each manager submits	Changes to last year’s lines	3 to 10 decision packages per unit, each a level of service with its cost	One ranked list: lines added in the last 12 months, lines below the last cutoff, contracts renewing within 2 quarters, older lines still unproven or running above plan
Leadership decides	Whether the total is acceptable	Where the cutoff falls across all packages	What to stop, shrink or keep on the list
Hours at a 40-person startup	Fewest	About 308 (central case)	About 44 a quarter
What the record shows	The base goes unexamined	Ranking valued; savings rarely measured	Untested as a package; built from GAO’s 1979 advice and Pyhrr’s cutoff line

Source: Pyhrr’s method as summarised by Dean (1978); GAO (1979); hours from the cost model further down (our arithmetic).

Zero based budgeting never started from zero

The working method came out of Texas Instruments. Its staff and research departments, facing a budget decrease in 1970, adopted a procedure that Peter Pyhrr described in the Harvard Business Review that November (Dean, 1978). The name was older than the method. Pyhrr was “using a term formerly used at the U.S. Department of Agriculture”, in Burton Dean’s summary of Pyhrr’s own account.

Managers described every activity in one or more decision packages, then ranked them. The budget became the sum of those funded: “the most important activities (or decision packages), whether they are current or new” (Nevada Legislative Counsel Bureau, 1977, quoting Pyhrr’s 1973 book). The zero was never literal. Each unit’s first package was a minimum level, “usually substantially less than the current level, say 20% to 40% less”, with 3 to 10 packages per unit above it (Dean, 1978). Senior managers could not read every package, so Pyhrr added a cutoff line. In Dean’s worked example, managers “glanced over all the packages, approved the top ones up to a total value of 50% of last year’s expenditures”, and argued only about the rest. The ranked list then served all year as the order in which to cut or add.

Jimmy Carter took the method into government. As governor of Georgia in the early 1970s he brought Pyhrr in to install it (Wikipedia). On 14 February 1977 he ordered every federal agency to build the fiscal 1979 budget with it, citing “the success of the zero-base budget system adopted by the State of Georgia” (Carter, 1977). A year later he asked for more work at the bottom of the range: “I think more emphasis should be placed on the examination of minimum levels” (Carter, 1978). From inventor to president, ZBB was a way to rank the margin.

Every public trial valued the ranking and struggled to count the savings

But the public record is thinner on savings than the method’s reputation. In spring 1962 the Department of Agriculture attempted “a comprehensive and simultaneous evaluation of all departmental programs”. Aaron Wildavsky and Arthur Hammond interviewed the participants and concluded that a zero-base budget “could be described but could not be practiced”, with “few specific changes attributable to this costly method” (Wildavsky & Hammond, 1965). William Carlson, who helped plan the exercise, looked back in The Wall Street Journal in 1976: “It was just too much.” He thought it suited a new Cabinet officer learning a department, while “it’s impossible to do every year”. And “what you discover is that the lowest priority things are there for political reasons” (quoted by the Nevada Legislative Counsel Bureau, 1977).

Carter’s federal version repeated the pattern at scale. The White House’s summary of OMB’s first-year assessment said “No single dollar figure can be pinned down to represent total ZBB economies in the 1979 budget”. It credited the ranking instead: “many large requests never reached OMB, because agencies recognized in ranking priorities that the proposals could not be justified” (White House, 1978). An evaluation of 10 departments and 6 agencies, holding 74% of $560 billion of budget authority and based on their replies to OMB, put “priority ranking of agency programs” first among the benefits. ZBB “did not, in fact, produce significant tangible cost savings in federal agency 1979 budgets” (Sherlekar & Dean, 1980).

The cost was easier to count. The General Accounting Office found an agency that “generated over 90,000 pieces of paper to present 4,478 decision packages for 751 decision units”, against about 22,500 pages before. In one agency the output “was so overwhelming that it was often discarded because they did not have time to review it” (GAO, 1979). Paper was the one line that zero-based budgeting reliably grew.

Companies reported much the same. Of 54 early adopters surveyed by the consultant Paul Stonich, 51% said their operating budget had moved by less than 5%, and 13% reported cuts of more than 10% (Stonich, 1977, as summarised by Dean, 1978); the answers are self-reported. GAO’s remedy is the useful part: each year, agree on a limited number of programmes for full review, and challenge the minimum funding of the rest in detail only if the new request is “much higher than last year’s”. Nothing in this record suggests that repeating the full ritual finds savings the first pass missed.

The corporate revival found one large cut, then ran out of base

Kraft Heinz was the high point of a revival. Rodney Coyte, Martin Messner and Shan Zhou trace it to 3G Capital, which installed ZBB in the companies it bought (Coyte, Messner & Zhou, 2022). In the first quarter of 2016 Kraft Heinz’s US segment’s Adjusted EBITDA rose 32.9% on organic sales up 0.2%, with savings “driven by zero-based budgeting initiatives” (10-Q, 2016). The $1,725 million it reported by the end of 2017 was 15% above the original target (our arithmetic).

The SEC order never mentions zero-based budgeting, and nor do the seven reasons Kraft Heinz gave for the $15.4 billion of impairments it announced in February 2019; its 2018 10-K shows $15.9 billion for the year as a whole. What the order does support is a lesson about targets: a quota set year after year on a base already cut asks people to find money that is no longer there.

In the revival, though, the quota was part of the method. Accounts of the Kraft Heinz merger describe a plan that began with fixed targets for cuts and merger savings, with ZBB as one lever. People who ran or sold such programmes describe category owners who challenge every budget holder and savings locked into budgets, never a minimum level or a cutoff across units. The revival says more about quotas than about ranking.

Other adopters reported results without isolating them. Unilever counted more than €2 billion of 2018 savings from “the supply chain, ZBB and change programmes” together, toward a 20% underlying operating margin in 2020, from 16.4% in 2016 (Unilever 20-F, 2018). It reached 18.5% in 2020, a shortfall it attributed largely to Covid-related costs and product mix (20-F, 2020). Mondelez said it had “embraced and embedded zero-based budgeting practices across the organization” while its adjusted operating margin rose from 12.0% in 2014 to 15.3% in 2016, mainly on “lower overheads from cost reduction programs” (Mondelez 10-K, 2016).

The one large-sample test found little. Coyte and colleagues compared 79 firms that announced ZBB on earnings calls with S&P 500 companies from 2002 to 2018, 7,832 firm-years in all, and found no significant cost reduction among adopters overall. Adopters that mentioned 3G in their filings showed some, at the 10% level. Imitation of peers drove adoption, and the authors suggest the revival “may be more symbolic than substantial”. Annual reports trace the same arc: 78 Form 10-K filings by 35 companies have used the phrase since 2001, 44 of them (56%) between 2015 and 2019, and 9 in 2022 to 2026 (EDGAR full-text search, our count).

Garbage collectors solved the same scheduling problem by sweeping young objects often

Therefore the question for a startup is how often to pay for a full review, and computer science has a worked answer. Garbage collection reclaims memory a program no longer uses (Wikipedia). The simplest collectors stop the program and trace everything, and while they run “the program can perform no useful work”. Generational collectors cut that cost with one regularity: “Generational garbage collection schemes are based on the empirical observation that most objects die young.” On most cycles they test only recent objects, and only “occasionally a mark-and-sweep is performed” across everything (Wikipedia). The price is stated plainly: “some garbage is not detected as such for longer than normal.”

In a budget, the young generation is every line added in the last 12 months plus discretionary programmes. The old generation is established payroll, core infrastructure and signed multi-year contracts, and the full mark-and-sweep is a zero-base pass. Age only stands in for an unknown return, though: one fast-growing B2B software company stopped re-arguing sales-development hires and paid ads once more spend brought no more opportunities, and kept re-budgeting every line of unknown yield, however old.

Three differences limit the mapping. No study measures the infant mortality of budget lines, so a company has to test it on its own records. Reachability in memory is a fact, while a line’s value is a judgement made partly by its owner, which was Carlson’s point about political survivors. And an object keeps its allocated size, while an established line drifts up as prices rise, usage per head grows and policy compliance slips.

Savings creep back once monitoring stops. One large adopter reassigned the staff who governed its law firms after rating that spend among its best managed; fee bills grew inaccurate until the contracts were re-sourced. In Accenture’s 2016 survey of 682 senior executives, self-reported and covering all cost cuts, only 36% strongly agreed that their business sustains the benefits. The time spent watching a line therefore belongs in its minimum level. A line running well above what the last full pass approved should rejoin the quarterly list, which turns GAO’s exception into a trigger.

What one full ZBB cycle costs a hypothetical 40-person startup

	Low	Central	High
Decision units (one per function)	8	8	8
Packages per unit	3	6	10
Decision packages	24	48	80
Hours to write and cost a package	2	4	8
Finance hours to check a package	0.5	1	2
Writing and checking, hours	60	240	800
Unit rankings (unit head and finance lead), hours	16	32	48
Leadership ranking session (6 people), hours	24	36	72
Total hours	100	308	920
Person-weeks at 40 hours	2.5	7.7	23.0
Cost at $100 an hour	$10.0k	$30.8k	$92.0k
Share of an $8.0M annual budget	0.12%	0.39%	1.15%

Source: hypothetical company. Packages per unit: Dean (1978), 3 to 10; the central 6 matches the GAO agency’s 4,478 packages across 751 units, about 6.0 a unit. Hours per package, the $100 hourly cost (a $200k loaded salary over 2,000 hours) and the $8.0M budget are our assumptions. A quarterly ranking of recent and marginal lines (8 ranked lists at 3 hours, 1 finance hour each, a 2-hour session for 6 leaders) takes 44 hours, or $4.4k. CX Cash calculated the totals.

At startup scale the full pass costs less than its reputation: about 308 hours, or 0.39% of the budget, on central assumptions, plus setup that shrinks more slowly than headcount, since a year of spending must first be restated into clean categories. The reason to keep the pass rare is yield. The record suggests a first pass finds most of the slack, and Kraft Heinz shows what follows when a target assumes the next pass will find as much.

A six-step cycle for a 40-person company

1. Zero-base once, at a change point: a round that changes the plan, a cut, a pivot, a new chief executive or finance lead. Daniel Ogden’s 1978 rule fits: when demands shift, “a good manager will call for a zero based budget to re-establish a defensible base” (Ogden, 1978).
2. Write packages from a minimum. Each function sets its minimum level 20% to 40% below current spending, then prices each increment above it and what it buys. Price growth increments only as far ahead as their yield is known, often a quarter or two: yields learned on revenue already won tend to break as revenue doubles.
3. Draw the cutoff before the meeting. Approve the packages above it on a read-through and spend the meeting below it.
4. Keep the ranked list. It becomes the order of cuts if the next round slips.
5. Each quarter, rank the young lines: one ranked list per function of every line added in the last 12 months, every line below the last cutoff and every contract renewing within 2 quarters. A new department head’s first-quarter additions go first.
6. Set no year-over-year savings quota. Follow the results in the monthly budget variance review.

Who questions the base if nobody starts from zero?

The best case for the full method comes from its inventor and its most aggressive users. Pyhrr argued that, properly run, “this kind of budgeting need not add heavily to the burdens of budgetmaking” (Nevada Legislative Counsel Bureau, 1977). Carlson’s remark supports the other half of the case: the lowest-priority items sit in the base, and only a review of the base finds them. Kraft Heinz did beat its original savings target, Mondelez’s margin rose 3.3 points in two years, and Coyte’s 3G-aware adopters showed some cost reduction.

Pyhrr and Carlson are right that the base hides spending, and at 40 people Pyhrr is right about the burden too: about 308 hours is a modest price at a change point. That is the case for the one full pass, and a weak case for repetition. Each success above was a first pass on a base nobody had questioned, usually under a new owner. At Kraft Heinz the quota that outlived the synergies ended in a restatement.

One full pass per change point, and a ranking of young lines every quarter

The record points to a schedule more than a method. Every pass in it starts from a minimum level, since a package without one is only a request.

When	What to do	What the record shows
At a change point	Run one full zero-base pass, budgeting about 300 hours at 40 people, with every function starting 20% to 40% below current spending	Each documented success was a first pass on a base nobody had questioned
Every quarter	Rank the young and marginal lines instead of running a second full pass	GAO’s 1979 advice: full review for a few programmes, the minimum of the rest left alone
At the board review	Ask for the ranked list and where the cutoff fell, before any percentage saved, and for what was added in the last 12 months and what was stopped	OMB credited the ranking and could pin no dollar figure on the savings
When a company says “we zero-based the budget”	Treat it as a claim to check, starting with whether anything was ranked	Unilever reported its savings bundled with other programmes, and the large-sample study found none on average
When savings targets come up	Resist year-over-year targets on a base already cut	The Kraft Heinz quota outlived its merger synergies and ended in a $208 million restatement

ZBB’s value has always been the ranking. The savings it can show came from first passes on an unexamined base; where it became a government-wide ritual or a year-over-year quota, the paper and the pressure grew instead. The quarterly ranking has no trial of its own, but a company can test it on its own records by counting how many of last year’s new lines it has already stopped.

References

• Accenture (2016). 82 Percent of Executives Aim to Cut Costs to Fuel Growth, But Majority Struggle to Sustain the Cost Reduction Benefits, press release, 12 January (online survey of 682 senior executives in 13 industries).
• Carter, J. (1977). Memorandum for the Heads of Executive Departments and Agencies on Zero-Base Budgeting for the Fiscal Year 1979 Budget, 14 February. The American Presidency Project.
• Carter, J. (1978). Memorandum From the President on Zero-Base Budgeting, 14 April. The American Presidency Project.
• Coyte, R., Messner, M. and Zhou, S. (2022). The revival of zero-based budgeting: drivers and consequences of firm-level adoptions. Accounting & Finance 62(3), 3147–3188.
• Dean, B. V. (1978). Methods in Zero-Base Budgeting. Research report, Department of Operations, Case Western Reserve University.
• Kraft Heinz Company. Form 10-Q for the quarter ended 27 September 2015; Form 10-Q for the quarter ended 3 April 2016; Form 10-K for 2017; Form 10-K for 2018.
• Mondelez International (2017). Form 10-K for 2016.
• Nevada Legislative Counsel Bureau, Office of Research (1977). Zero-Base Budgeting, Background Paper 1977 No. 2 (quoting Pyhrr, 1973, and The Wall Street Journal, 9 July 1976).
• Ogden, D. M. (1978). Beyond zero based budgeting. Public Administration Review 38(6).
• Pyhrr, P. A. (1970). Zero-base budgeting. Harvard Business Review, November–December, 111–121 (cited through Dean, 1978). Pyhrr, P. A. (1973). Zero-Base Budgeting: A Practical Management Tool for Evaluating Expenses. New York: Wiley (quoted through the Nevada paper).
• Ramsey Solutions. How to Make a Zero-Based Budget.
• Securities and Exchange Commission (2021). In the Matter of The Kraft Heinz Company and Eduardo Pelleissone, Release No. 33-10977; press release 2021-174.
• Securities and Exchange Commission. EDGAR, Form 10-K filings containing “zero-based budgeting”, as of 3 October 2026.
• Sherlekar, V. S. and Dean, B. V. (1980). An evaluation of the initial year of zero-base budgeting in the federal government. Management Science 26(8), 750–772.
• Stonich, P. J. (1977). Zero-Base Planning and Budgeting. Dow Jones-Irwin (survey results as summarised by Dean, 1978).
• Unilever PLC. Form 20-F for 2018; Form 20-F for 2020.
• US General Accounting Office (1979). Streamlining Zero-Base Budgeting Will Benefit Decisionmaking (PAD-79-45).
• White House (1978). Zero-Base Budgeting: Announcement of a Report Concerning First-Year Results in the Federal Government, 2 May. The American Presidency Project.
• Wildavsky, A. and Hammond, A. (1965). Comprehensive versus incremental budgeting in the Department of Agriculture. Administrative Science Quarterly 10(3), 321–346.
• Wikipedia. Garbage collection (computer science); Tracing garbage collection; Peter Pyhrr.

The figures come from the filings, government reports and papers above; the findings of Sherlekar and Dean, Wildavsky and Hammond, and Ogden are taken from their published abstracts. Hours, page counts and percentage changes printed without a source are our own sums on those records, and so is the EDGAR count. The 40-person startup is hypothetical, with its assumptions stated under the table.