Zero-Based Budgeting in the Age of AI: A Practical Playbook

Table of Contents

1. Why ZBB and Why Now
2. Where ZBB Fits Between Top-Down and Bottom-Up
3. Two Practical Ways to Run ZBB
4. The 7-Phase ZBB Roadmap
5. Start Small: Where the Quick Wins Hide
6. Overcoming Stakeholder Resistance (and Making It Fun)
7. How AI Cuts ZBB Cycle Time
8. Metrics That Matter

Key Takeaways

• AI makes zero-based budgeting faster and more practical by automating spend classification and data preparation.
• ZBB is a mindset shift that challenges teams to ask “Do we still need this?” instead of “What’s new?”
• Starting small with a focused pilot helps uncover quick wins and builds confidence before scaling companywide.
• Reframing cuts as reinvestment opportunities encourages engagement and rewards high-ROI ideas.
• Finance leads the process, but collaboration across departments ensures realistic and balanced plans.
• AI accelerates analysis, while human judgment guides strategic decisions and trade-offs.
• Success should be measured not only by savings but also by re-justified spend, efficiency gains, and faster decision cycles.

Traditional incremental budgeting focuses on what changes next year. Zero-based budgeting (ZBB) also asks a tougher question: what should stop?  This playbook shows how to roll out ZBB in seven phases, where to find quick wins, and how AI lightens the heaviest lifts (cleaning data, classifying spend, and creating a defendable baseline). You’ll leave with a practical roadmap, a stakeholder strategy, and a prompt framework you can use today.

Why ZBB and why now

Most finance teams still rely on incremental or hybrid approaches. That’s fine for stability, but it bakes in old choices and “SALLY” (same as last year) line items. ZBB flips the mindset: every dollar must be re-justified against outcomes.

Benefits:

• Reveal savings quickly (SaaS, travel, overlapping vendors).
• Combat status-quo bias (old contracts, predecessor decisions).
• Reinvest smartly using the Pareto lens (the 20% of spend driving 80% of outcomes).

Watchouts:

• It’s time-intensive in the first cycle.
• It can over-reward short-term ROI unless you intentionally protect long-horizon items (brand, retention, sustainability).

Where ZBB fits between top-down and bottom-up

• Top-down sets ambition aligned to strategy and external factors (market, tariffs, competitors).
• Bottom-up shows how to execute at the project/transaction level.
• ZBB is a bottom-up exercise — but the best teams link ZBB to a top-down view to calibrate realism vs ambition.

Two practical ways to run ZBB

1. Blank slate: build from zero only the line items that clear your effectiveness/ROI bar. Powerful, but heavy.
2. “Start with last year, justify every line”: the enterprise-friendly path. Inspect last year’s transactions at a granular level, group by outputs (e.g., one TV campaign spans multiple lines), then keep, renegotiate, or cut.

Tip: “Outputs first” avoids death-by-GL-code. You’re evaluating decisions, not just vendors.

The 7-phase ZBB roadmap

Phase 1 — Goals & Governance

• Define savings targets or reallocation goals.
• Decide enterprise vs pilot (start small!).
• Nominate owners (CFO/VP Ops as steering; FP&A as PMO).
• Clarify “what zero means” for you and pick your approach (blank-slate vs justify-every-line).

Phase 2 — Taxonomy & Drivers

• Create a simple taxonomy (Category → Subcategory → Driver).
• Typical drivers: FTEs, square footage, licenses, tickets, shipments, ad impressions, etc.

Phase 3 — Baseline & Commitments

• Pull 12–24 months of transactions; cleanse just enough to be decision-grade.
• Separate commitments (locked contracts) vs discretionary. Flag one-time vs recurring.

Phase 4 — Cost-Center Packages

• Send each owner a “package”: historic spend, output groups, contract notes, renewal dates, driver context.
• Ask owners to propose keeping, leaning, renegotiating, or cutting, along with a brief rationale.

Phase 5 — FP&A Challenge & Dependencies

• Facilitate review sessions; pressure-test assumptions; surface dependencies (kill X → Y breaks).
• Protect long-horizon investments with explicit criteria (e.g., retention, brand, talent).

Phase 6 — Consolidate & Decide

• Build scenarios (Base / Lean / Reinvest).
• Trade-offs are the point: publish a concise “Service Catalog” (decisions, policies, and PO limits by category).

Phase 7 — Operate & Learn

• Translate decisions into budget controls (approval flows, limits, renewal rules).
• Run mini-ZBBs quarterly for one category at a time (SaaS, Travel, Events).
• Post-mortem: planned vs realized savings; update benchmarks.

Start small: where the quick wins hide

• SaaS: right-size seats, kill duplicative tools, renegotiate renewals with usage data in hand.
• Travel & Events: pre-approve by purpose; cap per-diems; consolidate vendors.
• Marketing outputs: evaluate campaigns as bundled outputs, not as fragmented costs.

Overcoming stakeholder resistance (and making it fun)

Cut talk stalls momentum. Try ring-fencing: savings you identify go into a central pot that teams can pitch to win back for high-ROI ideas. It turns ZBB into a growth game.

Other ways to help stakeholders say “yes”:

• Promise more flexibility later (better budgets → faster approvals for in-year opportunities).
• Give ready-to-use packages (clean history, driver context, renewal calendar).
• Time-box requests (e.g., “Please classify just your top 30 vendors this week.”).

How AI cuts ZBB cycle time

The heaviest lift in ZBB is classifying messy spend and creating a defendable baseline. AI helps you:

1. Classify vendors to taxonomy from descriptions, invoices, and memos.
2. Group transactions into outputs (e.g., all lines that support “Q3 Product Launch”).
3. Confidence-score classifications and flag out-of-policy or one-time items.
4. Draft owner packages with summaries, renewals, and suggested options (keep/lean/renegotiate/cut).

Prompt framework (use this skeleton):

• Goal: “Classify spend for a ZBB project.”
• Returns: “For each transaction: Category, Subcategory, Driver, Confidence (0–100%), One-time? Y/N, Notes.”
• Format: “Return CSV with headers: id, category, subcategory, driver, confidence,one_time, notes.”
• Warnings: “Use only my taxonomy; if <80% confident, set driver to ‘Unknown’ and add a research note.”
• Context: “We’re a B2B SaaS co. Drivers include FTEs, licenses, impressions, square footage, and tickets.”

Where humans still decide:

• Long-term ROI (brand, retention, sustainability), sensitive vendor relationships, and cross-functional trade-offs.

Metrics that matter

• % of spend re-justified (target >80% in pilot scope).
• Savings realized vs target (and % reinvested).
• Cycle time (kickoff → decision).
• AI coverage (% transactions with >80% confidence).
• Policy adherence (% spend pre-approved vs after-the-fact).

See how Paystand maps and monitors spend in real time with our full stack of offerings for your finances.