- Budget is an output, not an input. Work backwards from a new-customer revenue target through allowable cost per acquisition.
- Every channel has a floor. Below roughly 30 to 50 conversions a month, automated bidding never stabilises and you are paying for noise.
- Fund proven, promising and experimental separately, at roughly 70/20/10, and actually spend the ten.
- Demand capture has a hard ceiling. Once impression share plateaus, extra budget buys worse traffic, not more customers.
- Most accounts do not need a bigger budget. They need the same budget in different places.
Budget is an output, not an input
The usual way a paid media budget gets set is that someone looks at last year, adds a percentage, and divides it across channels in roughly the same proportions. This is not planning. It is inheritance, and it carries forward every allocation mistake made by whoever set the original split, usually for reasons nobody in the room can now remember.
A defensible budget runs in the opposite direction. Start with what the business needs paid media to produce, convert that into customers, multiply by what you can afford to pay for one, and the budget falls out of the arithmetic. If the resulting number is larger than you can fund, that is useful information: it means the target is not currently reachable through paid media at your margins, and the conversation that follows should be about price, product or channel rather than about trying harder.
# Illustrative only. All amounts AUD, per month.
1. THE TARGET
New-customer revenue needed $400,000
Average order value $420
New customers required 952
2. WHAT YOU CAN PAY FOR ONE
Allowable CAC (see the strategy guide) $131
3. THE BUDGET
952 × $131 $124,712
4. THE REALITY CHECK
Current blended CAC $164
Budget required at TODAY's efficiency $156,128
Gap to close through efficiency: $31,416/mo
# The gap is the plan. Either find $31k of efficiency,
# fund the higher number, or move the target.
# Pretending the first number is achievable on day one
# is how budgets get blown in month two.
That last block is the part most budget exercises skip. There is almost always a gap between the budget the target implies at your desired efficiency and the budget it implies at your current efficiency. Naming the gap turns it into a work plan. Ignoring it turns it into a surprise in the second month.
Use new-customer revenue, not total revenue
If the target includes revenue from people who would have bought anyway, the budget derived from it will be too large and the reported efficiency too flattering. Repeat purchase and retention are real and valuable, and they are not what the acquisition budget is buying.
Every channel has a floor
Modern bidding is a statistical process. It needs a certain volume of conversion events inside a certain window to model anything useful, and below that volume it is guessing with confidence. This produces the single most wasteful pattern in paid media: a channel funded at a third of what it needs, running permanently in a learning state, generating data too noisy to act on and results too poor to justify the budget it does have.
A workable floor for each channel is the spend required to produce roughly 30 to 50 conversions in 30 days at your expected cost per acquisition. The arithmetic is deliberately simple:
Channel floor = expected CAC × 40 conversions
# Worked, using the $131 allowable CAC above.
# Demand-creation channels usually run 1.3 to 2x the
# capture CAC before they are optimised, so budget for that.
Non-brand search $131 × 40 = $5,240
Paid social $131 × 1.6 × 40 = $8,384
Video $131 × 2.0 × 40 = $10,480
# Against a $30k monthly budget, that is two channels
# funded properly, or three funded badly. Pick two.
These are rules of thumb rather than laws, and they move with conversion volume and sales-cycle length. A business with a 500-dollar cost per acquisition and long consideration cycles will not hit 40 conversions a month on a modest budget at all, and needs to optimise toward a higher-volume upstream event with a proven relationship to closed revenue rather than to the sale itself.
The most expensive budgeting decision
Splitting a limited budget across too many channels so that nothing is starved of attention politically, while everything is starved of data mathematically. It feels balanced in a spreadsheet and produces a portfolio of channels that all look mediocre for reasons nobody can diagnose. Concentration is not a risk here; it is the correct move.
Proven, promising, experimental
Once the total is set and the floors are known, the split has three jobs to do at once: defend the revenue you already have, grow what is nearly working, and find the next thing before the current thing stops working. Those three jobs compete, which is why splitting by channel name rather than by role produces such muddled allocations.
| Bucket | Share | What goes here | Judged on |
|---|---|---|---|
| Proven | ~70% | Channels and campaign types consistently at or inside allowable CAC, with enough volume to trust | Cost per acquisition against the ceiling, and impression share |
| Promising | ~20% | Working but not yet at scale, or at scale but not yet efficient. Needs time and volume, not a verdict | Trend across 60 to 90 days, not this month's number |
| Experimental | ~10% | New channels, new formats, new offers. Most will fail, which is the point | Did it produce a clear answer, either way |
Two rules make the shape hold in practice, and both get broken constantly.
The ten percent gets spent. Experimental budget that is quietly reabsorbed into the proven channel whenever the month looks tight is not a budget, it is a comfort blanket. Over a year that habit costs you every channel you might have found. Ring-fence it, name who is responsible for spending it, and review whether it was spent, not just how it performed.
Promising means a fixed term, not a permanent state. Give it a defined window, usually one or two quarters, and a stated condition for promotion or removal. Channels drift in the promising bucket for years because nobody ever set a date by which they had to prove something.
How much should demand capture get?
As much as it can absorb, and not a dollar more. Unlike almost everything else in paid media, demand capture has a genuine hard ceiling: there are only so many people searching for what you sell this month, and once you are showing for most of those searches, additional budget buys progressively looser matches at progressively worse economics.
The diagnostic is impression share rather than return on ad spend, which is the part accounts get wrong. Return can look strong right up to the ceiling, because the account keeps buying the profitable clicks and declining the marginal ones. What tells you the truth:
- Search lost impression share (budget) above roughly 10 percent on your core non-brand campaigns means there is captured demand you are simply not funding. This is the cheapest growth available and should be closed before any new channel is considered.
- Search lost impression share (rank) is a different problem with a different fix. It is telling you about bid, quality or relevance, and throwing budget at it changes nothing.
- Impression share above 80 to 85 percent with efficiency holding means the channel is close to full. Extra money here produces incremental volume at rapidly declining quality.
When capture is full and you still need growth, the answer is not more capture budget. It is demand creation, which is slower, less efficient per click, and the only thing that increases the size of the pool capture is drawing from. Accounts that refuse to make that move tend to plateau with excellent reported metrics and flat revenue, which is a comfortable place to be right up until it is not.
The account audit guide covers how to read these signals quickly, and the negative keywords guide deals with the other half of the problem: capture budget going to searches that were never going to convert.
Pacing, seasonality and the annual budget problem
An annual budget divided by twelve is wrong in almost every business, and wrong in a specific direction: it under-funds the periods when demand is highest and cheapest to convert, and over-funds the flat months where the same money buys considerably less.
Three adjustments worth making before the year starts:
- Index the months. Pull two or three years of your own conversion volume by month, index each against the average, and shape the budget to that curve rather than to the calendar. Your seasonality, not the category's.
- Fund the run-up, not just the peak. Demand creation needs to be in market before the peak arrives. Spending it during the peak means competing at the most expensive moment for attention you could have bought cheaply a month earlier.
- Hold a genuine contingency. Five percent of the annual figure, unallocated, for the competitor who exits, the product that unexpectedly works, or the month where cost per click moves against you. Budgets without slack get raided from the experimental bucket instead.
Daily budgets are not daily
Google and Meta both spend against a monthly average rather than a strict daily cap, and can overspend a given day substantially while staying within the period total. Pacing panic on a Tuesday is usually a misreading of that mechanic. Judge pacing on a rolling seven-day basis and on the month to date, never on a single day.
When to move money and when to add it
The default response to disappointing results is to ask for more budget. It is occasionally the right answer and usually the expensive one, because a budget increase applied to a misallocation just buys more of the same mistake.
| What you are seeing | What it usually means | Move or add |
|---|---|---|
| Lost impression share to budget above 10% on profitable non-brand | Funded demand is going unserved | Add, to that campaign specifically |
| One channel well inside allowable CAC, another well outside | The split is stale | Move, in increments of 10 to 20% |
| Everything sitting near the ceiling, volume flat | Capture is saturated | Move to demand creation |
| Blended CAC rising while every platform reports steady | Double counting, or declining incrementality | Neither — measure first |
| A channel in learning for months | It is below its floor | Move the budget out, or commit to the floor |
| Strong results in a short test window | Possibly real, possibly noise | Add slowly — 20% a fortnight |
On the last row: large sudden increases reset the bidding model's assumptions and frequently produce a fortnight of poor performance that gets misread as the increase failing. Twenty percent every two weeks lets the account absorb the change. It feels slow and it is considerably faster than increasing by eighty percent, panicking, and reversing it.
Reallocation deserves a fixed monthly slot on the calendar. Left to judgement, it happens when something goes wrong, which means it is always a reaction rather than a decision.
Budgeting mistakes worth avoiding
| The mistake | Why it persists | What it costs |
|---|---|---|
| Setting budget as a fixed percentage of revenue | It is simple and it is what the board expects | Spend shrinks exactly when demand softens and you most need it |
| Paying an agency a percentage of ad spend | It aligns with how agencies are structured | Your adviser earns more when you spend more, regardless of whether you should |
| Equal splits across channels | It looks fair and avoids an argument | Everything sits below its floor and nothing produces clean data |
| Judging branded search inside the blended number | The blended figure looks excellent | Non-brand underperformance stays invisible for years |
| Cutting demand creation in a soft quarter | It is the line with the worst visible efficiency | Capture volumes fall two quarters later, cause unidentified |
| Annual budget divided by twelve | Nobody owns the seasonality work | Under-funded peaks and over-funded troughs, every year |
Frequently asked questions
How much should a small business spend on paid media per month?+
There is no useful universal figure, because the answer is set by your cost per acquisition rather than your size. The practical minimum is whatever produces roughly 30 to 50 conversions a month in a single channel: at a 50-dollar cost per acquisition that is around 2,000 dollars, and at 400 dollars it is closer to 16,000. If your budget cannot reach the floor for even one channel, paid media is not yet the right growth lever, and that is a legitimate finding rather than a failure.
What percentage of revenue should go to paid media?+
Benchmarks in the range of 5 to 15 percent get quoted often and are close to meaningless across different margin structures. A business at 70 percent gross margin can sustain far more than one at 25 percent, and a fixed percentage also has the perverse effect of cutting spend precisely when revenue dips. Derive the budget from allowable cost per acquisition and use the resulting percentage as a sanity check afterwards, not as the input.
Should I increase budget if return on ad spend is strong?+
Only where there is unserved demand to buy. Strong return with impression share already above 80 percent means the channel is close to full, and extra budget will buy looser matches at worse economics. Strong return alongside lost impression share due to budget is a different situation entirely, and is usually the best-value increase available to you.
How quickly can I scale a paid media budget?+
Roughly 20 percent every fortnight is a reasonable pace for an established campaign with stable conversion volume. Larger jumps disturb the bidding model's assumptions and often produce a temporary dip that gets misdiagnosed as the increase failing. Brand-new campaigns need a different approach: start at or above the floor, because starting below it and scaling up wastes the entire learning period.
Should paid media budget include agency or consultant fees?+
Track them separately but include them in cost per acquisition. Media spend alone understates your true acquisition cost, sometimes materially at lower spend levels where a fixed fee is a large proportion of the total. The number the business should be managing to is fully loaded: media, management, creative production and tooling.
What should I do if the budget the target implies is unaffordable?+
Say so early and in writing. That gap is genuine information: at current efficiency and current margin, the target is not reachable through paid media alone. The productive responses are to improve conversion rate or average order value, revisit pricing, extend the payback horizon if cash allows, or lower the target. Quietly under-funding the plan and hoping is the one response that reliably fails.
Where this fits
Run the numbers
Both halves of this are built as tools. The Google Ads budget calculator works out how much a target needs, or what a fixed budget will return, and the ad budget calculator suggests how to split a monthly figure across Google, Meta, retargeting and local.
Most accounts do not have a budget problem. They have an allocation problem wearing a budget problem's clothes: money spread too thin across too many channels, capture under-funded while a saturated channel absorbs increases, and an experimental bucket that exists only on the plan.
The budget sits downstream of the strategy and upstream of the measurement framework that tells you whether the allocation was right. If you would like a second opinion on where your money is currently going, the free audit covers exactly that, and paid media consulting engagements open with it.