The Efficiency–Effectiveness Gap: Harvesting Till There Is Nothing More To Replant

Inflation is supposed to be the moment when every pound counts. And it is. But there is a version of that logic that is quietly making things worse — one where data abundance has made precision feel like prudence, and caution is mistaken for discipline.

We are optimising media with increasing precision, and steadily eroding the conditions that create future demand.

Capital is flowing to the most optimisable environments. Not the most effective ones. That gap has rarely been more costly.

According to the latest survey from Advertiser Perceptions / Premion, marketers overestimate the impact of tight targeting by a factor of two. Les Binet has already been precise on the mechanism: tight targeting is efficient but rarely effective. You reach buyers who were already close. You do not grow the pool. The assumption is that precision improves performance. In reality, it often improves short-term efficiency while limiting long-term growth.

This is the core of the efficiency–effectiveness gap: the growing divergence between what is easy to optimise and what actually creates demand.

Tight targeting is efficient because it reaches people already close to buying. Branding operates differently. It builds future demand by shaping memory, preference, and price sensitivity over time. Ipsos data shows that the share of Americans willing to pay a premium for brands that reflect their identity has risen 15 percentage points over the past decade. In an inflationary environment — where the ask of the consumer is already higher — brand distinctiveness is not a creative luxury. It is the condition under which pricing power survives.

Less waste is not the same as more growth. Confusing the two leads to systematic underinvestment in the assets that sustain it.

This tension is visible in how different media environments are being optimised.

The Brand Reset of dentsu reports long-term sales uplift after a single exposure at 4.5% for linear TV, 3.2% for CTV, and 2% for digital video. The differences are not marginal — they reflect how environment shapes effectiveness at a fundamental level.

Yet CTV is increasingly planned and optimised using digital performance logic: completion rates, short-term attribution, efficiency-led buying. It has inherited digital's KPIs, digital's creatives, and digital's success metrics. The lean-back conditions that make television effective at building memory and desire are being deprioritised in favour of measurable efficiency.

CTV does not simply sit between TV and digital. It sits between two fundamentally different economic models: one built on demand creation, the other on demand capture.

That distinction becomes even more important as AI-driven environments scale.

At first glance, AI-driven environments appear to be the purest expression of demand capture.

The early data is striking. Criteo's data across 500 US retailers shows ChatGPT referrals converting at 1.5x the rate of other channels. Walmart's AI assistant users carry basket sizes 35% higher than average. These are not small numbers.

But at the surface, this optimisation is built on the most recent signals — queries, prompts, observed intent. What it obscures is where those signals come from.

63% of LLM visibility is driven by brand equity according to WARC’s Guide, Brand Building in the Age of Gen AI. In other words, the outputs of AI systems are not neutral reflections of intent. They are shaped by the accumulated effects of long-term brand investment.

AI does not create demand in the moment. It compresses and expresses it.

That creates a structural asymmetry. Brands focused on performance marketing appear highly efficient within AI environments because they are optimising against existing demand. But brands that have invested in memory, distinctiveness, and salience are more likely to be surfaced, recommended, and chosen within those same systems.

If budgets continue shifting toward environments that capture demand while underinvesting in those that build it, the signal pool does not stay stable. It thins — not just in volume, but in quality. Optimisation becomes more precise on a shrinking base of people who were already going to buy.

The most expensive mistake in inflationary markets is not inefficiency. It is over-optimising the visible system — and quietly defunding the environments that create future demand.

And in AI-driven markets, that imbalance compounds: brand is increasingly not just a layer above performance, but the condition for being retrieved at all.

Sources: Advertiser Perceptions / Premion, dentsu The Brand Reset (Kantar, Lumen Research, Les Binet, Dan White), Ipsos, Criteo, WARC