Performance marketing and SEO in 2026: building demand you own
Performance marketing has stopped being an arbitrage game. With auction costs rising and a growing share of search resolved inside AI answers, the durable advantages are first-party data, creative quality, and content structured to be cited rather than merely ranked. Measure incrementality, not attribution, and treat organic visibility as an asset with a balance-sheet logic.
What is answer engine optimisation (AEO)?
Answer engine optimisation is the practice of structuring content so it can be retrieved, understood and cited by AI answer systems and search assistants — through clear question-led headings, self-contained factual answers, defined entities, structured data, and demonstrable expertise. It complements traditional SEO rather than replacing it.
What changed, and what did not
The mechanics of paid media have been progressively automated. Manual audience construction, bid management and placement selection — the craft that produced advantage a decade ago — are now largely platform-side decisions. What remains in the advertiser's control is the offer, the creative, the data you feed the system, and the measurement you use to judge it.
In parallel, a growing share of informational search never reaches a website. Answers are synthesised in the interface. For businesses whose funnel began with a search click, this is a structural change: visibility now means being the source that gets cited, not only the link that gets ranked.
What has not changed is the underlying economics. Demand you have to buy every quarter costs more each quarter. Demand created by brand strength and owned visibility keeps compounding. That is the whole argument for treating this as an asset-building exercise rather than a bidding one.
Building for search and for answer engines
The two overlap more than the discourse suggests, but the emphasis differs. Traditional SEO rewards topical authority, link equity, technical health and intent-matched pages. Answer systems reward retrievable, self-contained, verifiable statements attached to a recognisable entity.
In practice that means writing pages that can be quoted in a sentence.
Measure incrementality, not credit
Attribution models allocate credit among touchpoints that were already going to be there. They are useful for optimisation inside a channel and misleading for budget decisions between channels — branded search in particular absorbs credit for demand created elsewhere.
The corrective is experimentation. Geographic holdouts, audience holdouts, on-off tests, and where volume allows, incrementality studies at the platform level. The discipline is to accept a noisier number that is directionally true over a precise number that is systematically wrong.
Alongside that, marketing mix modelling becomes worth the effort once spend is meaningful across three or more channels. It will not resolve tactical questions, but it corrects the structural bias of last-touch thinking.
Creative as the performance lever
When targeting is automated, variance in outcomes concentrates in the asset. That reframes creative production as a performance function: enough concepts to test genuinely different propositions, disciplined variant structure so results are interpretable, and a hypothesis behind each test rather than volume for its own sake.
The most common waste we see is a hundred variants of the same idea. Ten genuinely distinct propositions, each tested cleanly, will teach more in a month than a thousand cosmetic permutations.
Own the audience you can
Every business should be building a first-party asset: a consented email or messaging list, an account base with behavioural history, a community, a product with logged-in usage. These are the only audiences that do not get repriced when a platform changes policy.
The corollary is that lead capture and content should be designed to build that asset, not just to convert this quarter. Gated content that adds nothing, or lists collected without permission or purpose, are liabilities rather than assets — expensive to maintain and legally exposed.
Rented demand vs. owned demand
Frequently asked questions
Is SEO still worth investing in if AI answers reduce clicks?
Yes, with a shift in emphasis. Being the cited source in an AI answer requires the same foundations — authority, clarity, structured content, verifiable expertise — plus content written to be quoted. Commercial-intent queries still produce clicks, and brand-shaping visibility now happens inside answers whether you get the click or not.
How do we measure marketing that does not produce a click?
Through incrementality testing and modelled contribution rather than click attribution: geo holdouts, share of search movement, direct and branded traffic trends, and lift studies. The absence of a click is not the absence of an effect.
How much should we spend on performance marketing?
Spend to the point where incremental contribution stops covering the cost of capital, not to a fixed percentage of revenue. That threshold can only be found by testing, which is why measurement design precedes budget expansion.
What is the difference between SEO, AEO and GEO?
SEO optimises for ranking in traditional search results. AEO optimises for being used and cited by answer engines. GEO — generative engine optimisation — is used for the same idea with emphasis on visibility inside generative responses. All three depend on clear structure, authority and verifiable claims.
A short diagnostic conversation is usually enough to tell you whether there is a real opportunity here — and what it would take.