Has Advertising Become Less Effective — and Where Are Brands Really Competing Now?

Main conclusion

After reviewing academic research, field experiments, the IPA effectiveness database, Kantar, Ehrenberg-Bass Institute and global digital-advertising data, the conclusion is that the original hypothesis describes a real symptom but not the full cause. There is no comparable global time series proving that advertising ROI fell two-, three- or five-fold in the past 10–15 years because social media appeared. Channels, attribution methods, consumer behaviour, market structure and the definition of ROI all changed at the same time.

There is, however, strong evidence of another process. The long-term effectiveness of many campaigns weakened from roughly the 2010s alongside the growth of short-term performance practice, media fragmentation and optimisation around easily measured indicators. IPA documented a fall in campaigns producing very large business effects and connected it with reduced brand building and short-termism. This can create a company with an excellent dashboard but a weakening future demand pool.

People have not simply begun to hate all advertising. Kantar Media Reactions 2025 reported that 57% of respondents felt positive about advertising overall. People have gained more power to filter, skip, block and recognise intrusive communication. Competition has therefore not moved from advertising to packaging, social or the website. It has spread across the whole decision chain. Advertising is not dead; what has died is the assumption that the fact of exposure is enough.

What actually happened to advertising ROI

The first problem is that marketers often compare different kinds of ROI. Sales attributed by a platform are not the same as incremental sales that would not have happened without advertising. A large eBay field experiment showed that observational attribution substantially overstated paid-search impact because many people who clicked a branded ad already intended to buy. Another large experiment for Edmunds.com found a different result: turning off paid search lost more than half of paid traffic. A channel has no single true ROI; effect depends on brand, category, awareness and whether the communication creates demand or intercepts existing demand.

Randomised Facebook experiments also found that observational models often failed to reproduce causal lift. The paradox of the digital era is that marketing became much more measurable without necessarily becoming better measured. Algorithms optimise the function they receive: clicks find click-prone people and purchase optimisation finds purchase-prone people. But the person most likely to buy after seeing an ad is not necessarily the person whom the ad caused to buy.

A 2025 Marketing Science study covering more than 70,000 Facebook and Instagram advertisers estimated a median cost per incremental customer of $38.16 under purchase optimisation and $49.93 in a modelled loss of offsite purchase data—about 31% higher. Variation was enormous: the 10th and 90th percentiles were roughly $5.86 and $241.95. That makes claims such as ‘Facebook ROI equals X’ almost meaningless without context and causal design.

Privacy changes did weaken parts of the performance machine. Research on Apple App Tracking Transparency estimated about a 37% fall in CTR for conversion-optimised Meta ads after the regime change, with relatively worse revenue effects for ecommerce businesses more dependent on Meta. Attention also fragmented across feeds, creators, streaming, search, gaming, ecommerce and subscriptions. The harder task is often to be noticed, correctly identified, remembered and recalled at the moment of choice.

Short-termism and the demand pool

The most serious evidence of declining effectiveness concerns short-termism. Performance marketing repeatedly finds people already close to purchase, reports a strong ROAS and encourages the company to reduce broad-reach brand investment. Two or three years later the pool of people who already want the brand has shrunk, and performance becomes more expensive. IPA describes this as a shift from effectiveness to efficiency: optimising the cost of an action instead of scaling total profit.

The more accurate reformulation is therefore not that advertising technology became radically weaker. It became much easier to build a marketing system that optimises short-term attribution brilliantly while creating too little future demand. Brand building and activation perform different jobs and need different time horizons. The familiar 60:40 rule should not be applied mechanically, but the principle of funding both remains important.

Do people avoid advertising more aggressively?

People now control their attention through scroll, skip, mute, unsubscribe, block, ad-free subscriptions, tracking controls and algorithmic feedback. Academic reviews show that ad avoidance is driven by perceived intrusiveness, irritation, relevance, privacy and format—not by a universal rejection of brands. The barrier for bad advertising is lower because people no longer have to tolerate it.

The most toxic experience is a sense of invasion. Transparency research shows that personalisation itself does not automatically damage effectiveness, but a ‘Why am I seeing this ad?’ explanation can reduce response when it reveals a data flow the person considers inappropriate. The reaction is less ‘I hate brands’ and more ‘Why are you following me?’, ‘Why are you interrupting this?’ or ‘How do you know that about me?’

This is consistent with Kantar’s global receptivity data. Consumers do not avoid advertising as a phenomenon; they avoid a poor exchange for their attention. Useful point-of-sale media, relevant sponsorship, entertaining cinema or carefully integrated creator content can be welcomed. The future is not an advertising-free consumer but a consumer with an increasingly effective spam filter.

Where brand competition moved

• Packaging became more strategically important as a one-second recognition and conversion interface, but cannot replace reach or memory building. It works best when brand cue, reason to choose and proof are immediately legible.
• Social media became a cultural, discovery and recommendation environment rather than merely an advertising channel. It gives a brand a human voice, creator interpretation and social proof, but cannot universally replace paid reach.
• Search and retail media moved competition closer to the transaction. Retail-platform advertising roughly doubled its share of digital spend between 2019 and 2024. Brands increasingly win or lose through ranking, availability, thumbnail, rating, delivery, product page and last-metre clarity.
• The website became an evidence and conversion layer rather than the main mass medium. It explains, substantiates and converts interest that often began elsewhere.
• Traditional media did not disappear. Even when digital took almost three quarters of global advertising investment, television, out-of-home, radio, print, sponsorship and physical presence continued to contribute reach, fame and memory.
• The product and company behaviour became the proof layer. No amount of content can sustainably compensate for a promise that is contradicted by the actual product, service or business model.

How brands should communicate values

A value should first exist in the behaviour of the business. Product and packaging then turn it into physical proof. Social should document and culturally translate that proof rather than repeat a declaration. Paid media can make the proof famous. At the moment of purchase the value must be compressed to one-second clarity: brand cue + reason to choose + evidence.

Social and political values require particular caution. Research on brand activism shows that perceived authenticity and brand–cause fit shape positive response, while a gap between statement and corporate reality creates scepticism and woke-washing perceptions. A useful test is: if the brand stopped talking about this value tomorrow, would customers still be able to see it in products and actions? If not, communication is ahead of reality.

The five-layer model: Proof → Encode → Distribute → Convert → Measure

Proof means making the promise real in product, service, business model and behaviour. This matters even more as synthetic content weakens trust; in Media Reactions 2025, 57% of consumers expressed concern that generative AI could create fake ads. Verifiability can become a competitive advantage.

Encode means building a small number of consistent distinctive assets: shape, colour system, device, sound, character, typography, pack architecture, motion style or verbal construction. Ehrenberg-Bass makes the criterion clear: an asset must not merely be attractive but uniquely and widely associated with the brand. Constant redesign in a cluttered environment can destroy accumulated memory.

Distribute combines emotionally strong broad-reach media with platform-native social, creators, PR, events, out-of-home and physical experience. Reach is bought, but attention is earned. Convert means winning the final metres through search, marketplace presence, pack, rating, availability, delivery and a clear reason to choose—without moving all investment to channels that harvest existing demand.

Measure means replacing exclusive reliance on ROAS, CPA, last click and platform attribution with holdouts, geo tests, incrementality experiments, econometrics and Marketing Mix Modelling. Brand building needs measures such as mental availability, unaided awareness, consideration, distinctive-asset recognition, category entry points, pricing power, penetration and market share. Final business measures remain incremental revenue, contribution margin, profit, customer-base growth and long-term enterprise value.

The next frontier: algorithms of choice

Kantar Marketing Trends 2026 forecasts a shift from competition for human attention alone to competition for agentic intention as AI assistants increasingly help people find, compare and potentially buy products. In Kantar’s 2026 work, 24% of AI users already reported using AI shopping assistants, and 41% of AI shoppers used them to compare retailer prices. This is an emerging frontier rather than proof that AI has replaced search or marketplaces.

The strategic implication is that a future brand must be understood by two kinds of intermediary. Human memory needs emotion, distinction and familiarity. Algorithmic systems need structured product data, availability, reviews, trustworthy information, consistent naming and verifiable claims. Algorithmic availability may become an additional layer beside mental and physical availability.

Final answer

There is no universal proof that advertising ROI collapsed several-fold. There is serious evidence that long-term effectiveness weakened where marketing shifted from brand building to short-term activation, while digital attribution often overstated causal ROI and privacy, fragmentation and attention scarcity made part of the performance system harder.

Competition did not relocate to packaging, social content or websites; it expanded. Packaging is a recognition and conversion interface, social a cultural and recommendation layer, search and retail high-intent battlefields, the website an evidence layer, the product the proof layer, and paid media remains critical for fame, reach and memory.

The strongest architecture is value → business behaviour and product → tangible proof → distinctive brand encoding → paid reach plus creator and social culture → search, retail, point of sale and package → experience → reviews, word of mouth and community → memory → next purchase. The centre of power has moved from message control to system consistency.

Sources

• Marketing Science and Management Science field research on Meta offsite data, Apple ATT, paid search and causal advertising measurement.
• IPA effectiveness publications, including channel mix and Go Big or Go Home; Kantar Media Reactions 2025 and Marketing Trends 2026.
• Ehrenberg-Bass Institute work on distinctive assets and packaging; academic reviews of digital ad avoidance, transparency and authentic brand purpose.
• DataReportal global advertising trends and supporting research on brand building, channel roles, retail media and incrementality.