An annual marketing budget is built months before the year it covers, using assumptions about growth, channel costs, and market conditions that will inevitably be partly wrong. Finance still treats it as a commitment because the rest of the business plans headcount, cash flow, and investor guidance around it. The planning skill isn't predicting the future perfectly — it's building a budget with enough structure and flexibility that it can absorb being wrong in a few assumptions without needing to be rebuilt from scratch.

Two starting philosophies: zero-based vs. incremental

Incremental budgeting starts from last year's actual spend and adjusts it up or down by a percentage ("marketing gets a 10% increase to support the revenue growth target"). It's fast, low-friction, and preserves institutional continuity — but it also quietly preserves last year's mistakes, since underperforming programs that were funded last year tend to get funded again by default.

Zero-based budgeting (ZBB) starts every line item at zero and requires it to be justified from scratch against this year's goals, regardless of what was spent last year. It forces a harder conversation about what's actually working, but it's slower, more politically charged, and if done every single year without exception, can create planning fatigue and under-invest in things that need 2-3 years to compound (brand building, SEO, content).

Zero-based vs. incremental budgeting

Dimension	Zero-based	Incremental
Starting point	$0 for every line, rebuilt from justification	Last year's actual spend, adjusted by a %
Speed to build	Slow — every line needs a fresh business case	Fast — mostly arithmetic on last year's numbers
Risk of perpetuating waste	Low — underperformers must re-earn funding	High — stale or underperforming programs often get re-funded by default
Risk to long-horizon bets	Higher — slow-compounding programs (SEO, brand) can lose out to channels with faster, easier-to-prove ROI	Lower — continuity protects multi-year investments from being cut on a single bad quarter
Political cost	High — every team re-litigates its budget annually	Low — mostly maintains the status quo
Common real-world use	Applied selectively to a subset of lines each cycle	Default method, often ZBB'd every 2-3 years as a reset

Most mature organizations blend the two

A common pattern: apply zero-based scrutiny to the newest or most volatile 20-30% of the budget (test channels, new programs) every cycle, while using incremental adjustment for the stable core (brand, retention, core paid channels) — then do a full zero-based reset every two to three years to catch anything that's quietly calcified.

Allocating budget across channels

Once a total number exists, allocation across channels should be driven by three inputs working together, not any one alone: historical ROI/incrementality data (what has actually worked, adjusted for the lift corrections covered in the effectiveness lesson), marginal return curves (most channels have diminishing returns — the 10th dollar in a channel rarely returns as much as the 1st), and strategic goals that aren't captured by short-term ROI (brand investment, new-market entry, retention programs that pay back over a longer horizon than a quarter).

A common allocation framework is 70/20/10: roughly 70% of budget to proven, reliably-performing channels; 20% to channels with a track record but room to scale; 10% to experimental or emerging channels with uncertain but potentially high upside. The exact split should move with the business's risk tolerance and growth stage.

channel_marginal_return.txt

text

Paid search marginal ROI by spend tier (illustrative)

Spend tier         Incremental revenue   Marginal ROI
---------------------------------------------------------
$0 - $50K              $220,000             340%
$50K - $100K            $150,000             200%
$100K - $150K            $90,000             80%
$150K - $200K            $55,000             10%   <- near saturation

Lesson: the AVERAGE ROI across all $200K looks healthy,
but the MARGINAL ROI of the last $50K is barely break-even.
Budget decisions should be made on the marginal curve,
not the blended average.

Scenario planning: budgeting for more than one future

Rather than building a single-point budget, mature planning builds at least three scenarios — typically a base case, an upside case (growth or market conditions beat plan), and a downside case (a pullback or macro shock) — each with a pre-agreed set of budget levers: what gets cut first in a downside, what gets accelerated first in an upside. Doing this work in the planning cycle, rather than improvising it mid-year under pressure, is what lets a marketing org respond to a revenue miss in days instead of weeks.

Pre-agree the downside cuts before you need them

If the first time anyone discusses which programs get cut in a downturn is during the downturn itself, the decision gets made under pressure, politically, and usually protects whoever is loudest in the room rather than whoever is actually driving the least incremental return. Agreeing the cut order in advance, based on the marginal-return data above, removes that dynamic.

What's next

Planning sets the budget; predictive modeling is what makes that budget work harder by targeting the people most likely to respond in the first place.

Next: Improving Response Through Modeling & Analytics →

I build these systems professionally.

Whether it's a RAG pipeline, analytics migration, or AI workflow — let's talk.

Work With Me →

Need production AI or MarTech expertise? I build custom LLM workflows, server-side GTM tagging, and analytics infrastructures professionally.

Need custom AI or MarTech setup? Let's build together.

Let's talk →