Where should the extra marketing budget go in 2026?

Why does the “extra” budget matter more than the rest?

Most of a marketing budget is committed before the year begins — salaries, the channels that already work, the renewals. The part that is genuinely free to choose is small, and it carries most of the strategic weight.

What remains is the surplus — the test line, the “let’s try something” money. It is a minority of the budget, and the only part free to chase whatever the next advantage turns out to be.

That makes the surplus a question on its own, separate from the tier-by-tier allocation most planning decks start with. The point here is narrow: when there is a little extra, where does it tend to earn the most right now? Over the last eighteen months, the most defensible answer appears to have shifted.

None of this argues against ads, social or events; starving a working channel shows quickly, and the core spread is usually close to right. The argument is only about the marginal line — and resisting the reflex to send it wherever last year’s surplus went.

Where is the field actually putting the marginal budget in 2026?

Mostly into paid media — which, on current evidence, may be the harder place to win the marginal line, not the easier one.

Gartner’s 2026 CMO Spend Survey is fairly blunt about the year. Budgets are flat, holding at roughly 7.7% of company revenue, so every additional unit of spend is contested. Within that flat envelope, paid media has risen to 31.4% of the budget — funded, Gartner notes, largely by cutting agencies — while around 15.3% is earmarked for AI, though only about 30% of CMOs feel ready to scale it.

So the dominant move is to push the marginal money toward rented reach — a reasonable instinct in a world where attention is bought. It reads differently where a growing share of discovery now resolves inside a generated answer the buyer never clicks through: paid reach buys the visit, but the decision is increasingly made inside the answer, before any visit happens.

There is a quieter consequence. When most of the field crowds into the same paid auctions, the ground everyone is stepping away from gets cheaper and emptier — and for a surplus free to go anywhere, that contrarian ground is often where the marginal unit stretches furthest.

Why does the owned foundation tend to win the marginal unit now?

Because discovery is collapsing into a single answer, and the engine assembling that answer reaches first for sources it can quote and check.

The click is thinning out. A 2025 Pew study found people click a result about 8% of the time when an AI summary is shown, against 15% when it is not; by early 2026, SparkToro’s clickstream analysis put zero-click Google searches at roughly 68%. So the prize is shifting from the visit to the mention — being named in the answer itself.

What a model reaches for is instructive. It tends to start with a source it can read plainly and quote, then look for corroboration. A 2025 Semrush study spanning hundreds of thousands of prompts found engines leaning heavily on a handful of third-party sources — community sites, editorial pages, reference works such as Reddit, LinkedIn and Wikipedia — alongside whatever a brand states clearly on its own pages.

That pairing is why the foundation tends to earn the marginal unit better than another ad flight. Three plain mechanics sit underneath it.

• You own it. Rented channels stop the instant the invoice does; a clear, citable page keeps working long after.
• It compounds. Each answer-first page adds to a body of material an engine can associate with you, where ad impressions evaporate as they are served.
• The cost per citation falls over time. One well-built page can surface across many answers, on several engines, for years, at little marginal cost.

None of this makes the foundation a guaranteed return — citation behaviour is volatile month to month — only the more durable home for money that has to keep working after it is spent.

Does any of this change for the Indonesian market?

The local market is moving hard toward measurable, mobile-first spending — which makes owning the answer one of the more measurable outcomes available.

Indonesia’s digital-ad market sits at about US$3.41 billion in 2026 and is still growing, with advertisers leaning toward performance and measurable results. The buyers are already on the other side of the screen: PwC’s 2025 workforce survey found a high rate of AI use among Indonesian workers, widely reported at well over two-thirds.

The implication is mundane and local. A prospective customer is already asking an engine about your category — in Indonesian, on a phone — and the answer either names you or names a competitor. When everyone is buying measurable reach, the cheapest measurable outcome may simply be to own the answer that appears the moment the question is asked.

So where exactly should the extra go?

It helps to treat the surplus as a ladder rather than a menu. Monitoring is the constant; each rung adds the next move, and buying the top before the bottom is solid tends to waste the money.

Where the surplus tends to earn most — add the next rung, don’t skip ahead

The common error is funding amplification before the foundation exists — sending the surplus into ads that point at a brand with no canonical source behind it, driving people into a conversation where someone else is the recommended name. Build the owned answer first; let paid make it travel faster afterwards.

Read in that order — owned, then earned, then amplification — the rungs are less a checklist than a single coordinated motion, each step making the next worth more. That is roughly what a deliberate approach to the AI answer is built to produce. For the tier-by-tier numbers behind each rung, our H2-2026 budget guide picks up the operational detail.

What do people most often get wrong with the surplus?

Most of the wasted spend traces back to an older reflex about what marketing money is for.

The frequent mistakes rhyme: pouring the extra into paid reach while invisible in the answer; chasing rankings rather than citations, when most searches no longer produce a click; buying volume on the assumption that output equals visibility, when engines appear to reward retrievable, corroborated material over raw quantity; and skipping corroboration, the unglamorous earned-media work that is much of what makes a brand quotable. None was reckless in the click-and-rank era. The shift is only that the marginal unit now tends to pay back one rung lower than instinct suggests — in the owned source and its corroboration — before it is handed to amplification.

Frequently asked questions