How should I plan my marketing budget for next year?

Start from last year's actual spend, not the approved budget, and re-cut it into five lines: paid media, people, martech licences, usage-based AI and data, and agencies and services. Index the total to 100, tag each line by purpose (acquisition, retention, brand, enablement), then move each line within a range rather than to a fixed target. Attach a proof condition to every increase, such as an incrementality test with a date. With budgets flat at 7.8% of revenue in Gartner's 2026 survey, every increase has to be funded by a cut elsewhere.

What percentage of revenue should go to marketing in 2026?

Gartner's 2026 CMO Spend Survey, published on 11 May 2026, put the average at 7.8% of company revenue, up from 7.7% in 2025. More AI-ready organisations in the same survey reported 8.9%. Treat these as reference points, not targets. The sample was 401 marketing leaders in North America, the UK and Europe, mostly at companies with more than $1 billion in revenue. A mid-market business in Singapore, Malaysia or Australia with a different growth stage, margin or sales cycle can be right well above or below that figure.

How much of the marketing budget should go to AI?

Gartner reported an average of 15.3% of marketing budgets allocated to AI initiatives in 2026, and 21.3% at more AI-ready organisations. The more useful question is which line that money sits in. AI spend lands in several places: licences, metered usage charges, training and new roles, and agency fees. Plan it where the cost actually lands, give metered usage its own line, and tie every AI use case to a unit of value such as cost per approved asset or cost per qualified conversation.

Why are marketing people costs rising if AI saves time?

Labour's share of the budget is rising, which is not the same as headcount. Gartner's June 2026 release reported labour rising from 21.9% to 24.5% of marketing budgets in one year, and linked it to AI value depending on the people who execute it. A share can also rise because other lines shrank while FTEs stayed flat, so check your own labour spend. Generating drafts and variants gets cheaper. Deciding what to ship does not, and neither does checking claims or keeping data clean. Teams need fewer people producing first drafts and more editors and analysts. Budget for training before adding headcount.

How do I budget for usage-based AI tools in marketing?

Give metered AI its own budget line instead of hiding it inside martech or a channel. Forecast in units first (generations, conversations, records processed per month, per market, per use case) and multiply by unit cost last. Set a soft alert at around 70% of the monthly allocation and a hard cap that a named owner must lift. Reforecast quarterly, because a successful rollout can multiply usage within a single quarter. Chief Marketer, reporting Gartner's survey, found 41% of CMOs had built or were building real-time controls.

Should I cut martech to fund paid media?

Often, partly. Gartner's 2026 data shows paid media rising to 31.4% of marketing budgets while martech fell to a five-year low of 19.4%, as reported. That shift makes sense when you are paying for unused seats or overlapping tools. It goes wrong when the cut reaches the data layer that measurement and AI workloads depend on. Audit seats and consolidate overlapping platforms first, keep the data foundation intact, and require any extra paid media to come with an incrementality read.

Is it a mistake to cut retention spend to fund acquisition?

Frequently, yes. Gartner's 8 June 2026 release found loyalty and retention had fallen 29% since 2024 to less than 15% of media spend, while the most AI-mature organisations allocated a larger share to retention. In our view, retained customers also produce the reviews and referrals assistants can draw on, though Gartner did not study that link. Check your retention share before moving anything else, and make any cut to it pass the same proof test as an increase elsewhere.

How does AI search affect marketing budget planning?

When buyers ask ChatGPT, Gemini, Copilot, Perplexity or Google's AI Mode first, the assistant frames the category and names a shortlist before your search ads appear. In our view, search ads then work more on confirming a choice than introducing one. In our view that raises the value of brand and retention spend. Ad placements inside assistants vary by assistant and market, so check each before budgeting. Treat answer-engine visibility as a purpose tag across the five lines, test it against a fixed set of buyer questions quarterly, and keep it only if your presence in those answers moves.

How do I justify a marketing budget reallocation to the CFO?

With 56% of CMOs in Gartner's 2026 survey saying budget cannot fund their strategy, extra money is unlikely; trust is what buys room. Answer the four questions most CFOs ask. What happens if we do not make this move: show a holdout or geo test, or the date one will report. Are the platform numbers real: show incremental return next to platform-reported return and explain the gap. What is the downside: name a stop condition with a date, plus the line the money returns to. Why is headcount rising: list the decisions and quality checks people now own. Ranges with proof conditions earn more trust than point targets.

Can small and mid-sized businesses use the same budget model?

Yes, because the model tracks cost behaviour rather than company size. A mid-market team still has paid media, people, licences, metered AI charges and outside services, even if some lines are small. Indexing the total to 100 makes the shape comparable whatever the absolute spend, and across currencies and fiscal years. What does not transfer is the benchmark. Large-company averages from Gartner describe the direction budgets are moving; your own actual spend should set the starting shape and the size of each shift.

Marketing budget planning when AI answers first

What does marketing budget planning look like when the total stays flat?

Marketing budget planning for 2027 is a reallocation job, and this guide is for the marketing lead who has to defend one to a CFO. After reading it you will be able to split spend into five lines by how each cost behaves, then decide which lines move and what proof each increase needs.

Gartner's CMO Spend Survey, published on 11 May 2026, puts the average 2026 marketing budget at 7.8% of company revenue, up from 7.7% a year earlier. Gartner's 2025 headline said budgets had flatlined. A tenth of a point does not change that.

So next year's argument is about what moves inside the envelope. Ewan McIntyre, VP Analyst and Chief of Research in the Gartner Marketing Practice, framed it in the 11 May release: "CMOs are being asked to deliver growth, efficiency and transformation without meaningful budget expansion. Those who succeed will make deliberate, data-driven trade-offs and treat AI as a force multiplier."

Read every Gartner figure here with its sample in mind. Gartner surveyed 401 CMOs and marketing leaders in North America, the UK and Europe from January to March 2026, most at companies with more than $1 billion in revenue. For a mid-market team in Kuala Lumpur or Brisbane, the numbers show direction, not a target.

Where is marketing money moving in 2026?

Money is moving into paid media and people, and out of agency fees and martech licences. Paid media takes 31.4% of the average budget and people 24.5%, per Gartner's 2026 CMO Spend Survey. Proof for that paid media line is what AI performance marketing work is built around.

Source: Gartner 2026 CMO Spend Survey (sample above); martech low as reported by Chief Marketer. "Everything else" is our arithmetic. AI overlaps the other lines, so it is drawn apart.

Gartner released the headline findings on 11 May 2026 and a media and labour follow-up on 8 June 2026, then summarised both in its CMO spend article. The martech and consumption-pricing figures come from the same survey as reported by Chief Marketer on 24 June 2026.

Where the 2026 marketing budget is moving, and what each figure means for your plan (Gartner 2026 CMO Spend Survey)
Line	2026 figure	What it means for your plan
Total budget	7.8% of revenue (7.7% in 2025); 56% of CMOs say it cannot fund their strategy	Every increase is funded by a cut somewhere else.
Paid media	31.4% of budget, funded by cuts to agencies	More money in platforms that grade their own homework, so measurement must grow too.
People	24.5% share, up from 21.9% in 2025	A share, not a headcount. Check your own labour spend and FTEs.
Martech	19.4%, a five-year low (26.6% in 2021), as reported	A lower share does not prove seat cuts. Audit seats before cutting.
AI initiatives	15.3% on average; 21.3% at more AI-ready firms; only 30% report mature AI readiness	This money sits inside the other lines. Find out where.
Awareness and conversion media	62.6% of media, up more than 10% since 2024	Spend is going where measurement is easiest.
Loyalty and retention media	Under 15% of media, down 29% since 2024	Check your own retention share before moving anything.

Labour's share rose 2.6 points in a year, against the common expectation that AI would shrink people costs. It is a share, so it can rise while headcount stays flat. Generating 300 ad variants is cheap. Choosing the dozen to ship takes people.

Retention runs the other way. Gartner's 8 June release found the most AI-mature marketing teams allocate a larger share to loyalty and retention, and suggests less mature teams over-index on what is easiest to measure and automate.

Why do usage-based AI bills break the annual marketing budget?

Metered AI bills grow with adoption, not with a contract term, so an annual marketing budget set in January misses them. Chief Marketer, reporting the Gartner survey on 24 June 2026, found 56% of CMOs had moved more martech budget onto consumption-based pricing and 41% had real-time cost controls in place or in progress.

Usage-based (metered) pricing means you pay per generation, token or conversation rather than per seat. AI sold as a per-seat add-on belongs in the licence line. A licence costs the same whether the team uses it once or ten thousand times. A metered feature costs more every time someone finds a new use for it, so the team that adopts fastest overspends first.

Illustrative example (made-up round numbers)

A team budgets a metered chat agent at 0.5 points of a 100-point budget, based on a one-market pilot of 20,000 conversations a month. In Q2 it goes live in four markets. Volume reaches 100,000 conversations a month, five times the pilot, at the same unit cost. The line now runs at about 2.5 points a year, and nobody approved the extra 2.

Illustrative round numbers, as in the example above.

Warning

Metered AI is the one line that overspends without anyone deciding to. Give it its own budget line before the year starts. Otherwise it surfaces months later, buried in a channel or agency invoice.

Set these up before the year starts:

Forecast in units, then price. Estimate generations, conversations or records per month for each market and use case. Multiply by unit cost last.
Set two thresholds. A soft alert at 70% of the monthly allocation and a hard cap that only a named owner can lift. Where a vendor contract supports no caps, alert on your own billing export.
Tie each use case to a unit of value, such as cost per approved asset or cost per qualified conversation. No unit means it is still a pilot.
Reforecast quarterly. Annual is too slow for a line that can grow fivefold in a quarter.

To see which tools are seat-priced and which are metered, our AI marketing stack audit sorts the layers.

How do you build a marketing budget reallocation model?

A marketing budget reallocation model re-cuts spend into five lines by cost behaviour, indexes the total to 100, and limits how far each line moves in one cycle. Indexing to 100 means stating each line as points of the total, so plans in different currencies compare on shape.

The five-line model: what goes in each line and how its cost behaves
Line	What goes in it	Cost behaviour
1. Paid media	Search, social, programmatic, retail media, AI-assistant ads where sold, test cells	Elastic; can change daily
2. People	Salaries, contractors, training	Sticky; slow and costly to rebuild
3. Martech licences	Seats for CRM, automation, analytics, CDP, content tools	Fixed per contract term
4. Usage-based AI and data	Credits, tokens, per-conversation charges, metered data feeds	Grows with adoption
5. Agencies and services	Retainers, project fees, production, specialists	Semi-fixed; renegotiated yearly

Brand and retention are purpose tags here, not lines, because both draw on all five. Tracking them as tags is what stops retention being quietly defunded while every line looks healthy.

Do this first

Re-cut last year's actual spend into the five lines, not the approved budget. Our rule of thumb: any line whose actuals landed more than 10% away from its approved figure gets explained before it gets a new number. That gap is usually where metered charges have been hiding.

Then, in order:

Index the total to 100.
Tag spend by purpose: acquisition, retention, brand and answer-engine visibility, enablement. If retention sits below about one sixth of the media line, flag it before moving anything else. One sixth is our heuristic, set near Gartner's under-15% retention share, not a Gartner benchmark.
Give each line a range, not a point target. Ranges invite evidence. Point targets invite haggling.
Attach a proof condition to every increase: a named test that confirms or reverses the move by a set date. More media needs an incrementality read, meaning a holdout or geo test that measures the sales that would not have happened without the spend.

Step 4 is where plans go soft. The test mechanics are in our incrementality testing guide, and our marketing ROI calculators show how to check platform-reported returns before they drive a move.

Marketing budget split calculator: is any line out of range?

Enter your total and the share for each of the five lines. The calculator returns the amount per line and the gap against Gartner's 2026 average where Gartner reports one, and it flags any line outside our heuristic range.

Illustrative calculator: a planning aid, not a benchmark. Not legal or investment advice.

Total budget (any unit) Paid media % People %

Martech licences % Usage-based AI and data % Agencies and services %

1 of 5 lines flagged

Showing the default split below.

Formula: amount = total × share ÷ 100. Gap = your share minus Gartner's 2026 average. Shares that do not add to 100 are rescaled to 100 first.

Calculator result (illustrative; defaults shown, total of 1,000 budget units)
Line	Share	Amount	Gap vs Gartner large-firm average (context, not a target)	Flag
Paid media	32%	320	+0.6	Inside the heuristic range
People	24%	240	-0.5	Inside the heuristic range
Martech licences	20%	200	+0.6	Inside the heuristic range
Usage-based AI and data	1%	10	not reported	Below 2: hold a reserve, the bill arrives anyway
Agencies and services	23%	230	not reported	Inside the heuristic range

With the defaults, only the metered line gets flagged. One point is too thin a reserve for a cost that grows with every new use case.

The flag thresholds, and how far each flagged line should move

Flag rules and suggested shift bands (Leapbuzz heuristic built from the direction of the Gartner data; it does not know your margins)
Line	Flag when share is	Suggested move this cycle
Paid media	Above 38	-3 to 0, into measurement and retention
People	Below 18	+2 to +4, editors and analysts rather than producers
Martech licences	Above 22	-5 to -2, through a seat audit and consolidation
Usage-based AI and data	Below 2	+1 to +3, held as a capped reserve
Agencies and services	Above 25	-6 to -3; keep specialists, reprice execution

How does AI search change what the marketing budget buys?

Some buyers now start category research in ChatGPT, Gemini, Copilot, Perplexity or Google's AI Mode, and the answer names brands before any search ad is seen. Our hypothesis, not a Gartner finding: search advertising then confirms that choice more often than it introduces one.

Ad placements differ by assistant and by market (see ChatGPT Ads by market), so do not budget "ads in AI assistants" as one line. The brand side is in our brand vs performance marketing guide.

Our decision rule for answer-engine visibility spend (ours, not Gartner's): before approving it, run 20 to 30 category questions your buyers actually ask in two or three assistants, quarterly, and record who gets named. If the spend has not moved your presence in those answers within two quarters, it goes back into the pool. Fund it as a purpose tag drawn from existing lines, with any AI-assistant ad placements run as a test slice of paid media. It never gets a new envelope.

How do you defend a marketing budget plan to a CFO?

Defend it with evidence that could have gone the other way. A flat budget will not grow because the slides got better, but a CFO who trusts your method will allow more movement inside it. Expect four questions.

The four CFO questions about a budget reallocation, and answers that hold up
CFO question	Weak answer	Answer that holds
What happens if we do not make this move?	"We fall behind on AI."	A holdout or geo test result, or the date one will report.
Are the platform numbers real?	"ROAS is up."	Incremental return from a controlled test next to platform-reported return, with the gap explained.
What is the downside if it fails?	"We will optimise."	A named stop condition with a date, plus the line the money returns to.
Why is labour's share up if AI saves time?	"We need more capacity."	Your own labour spend and FTEs versus last year, and the work people now own.

The fourth question is new. Gartner's labour figure (21.9% to 24.5% of budget in a year) is a share, not proof anyone hired. Bring your own labour spend and FTE counts, then show which roles changed: fewer people producing first drafts, more editing and testing.

Our AI ROI board proof guide sets the evidence standard. The CMO AI roadmap covers sequencing, and the board marketing scorecard shows which metrics survive that room.

Calendars and currencies differ across Singapore, Malaysia, Australia, the US and Canada, the markets where Leapbuzz builds AI strategy and budget models. Indexing to 100 absorbs most of that. A single benchmark does not travel: a Singapore insurer and a US software company can both be right with very different splits.

Your next step for marketing budget planning this cycle: pull last year's actuals, re-cut them into the five lines, and list every line that landed more than 10% off its approved figure. That list is the agenda for your first budget meeting.