Personalized advertising at scale without losing brand control

Personalized advertising adapts what an ad shows and says to the person or situation it reaches, using signals such as a customer segment, stated preferences, past behavior or the context of the moment. It usually works. A 2025 meta-analysis of 53 experiments in the Journal of Advertising Research found personalized ads generally more effective than generic ones, and found that perceived relevance, not intrusiveness, explained the difference.

People are also uneasy about it. In a 2023 Pew Research Center survey of more than 5,000 US adults, 73% said they have little or no control over what companies do with their data. This guide explains what ad personalization means, what research says about when it works and when it turns creepy, which rules apply in 2026, and how brands can produce personalized content at scale without losing control of the brand.

What is personalized advertising?

Personalized advertising changes the content of an ad based on who sees it or where they see it. It is easiest to understand as three different kinds of personalization:

• Segment personalization adapts ads for groups, such as new versus returning customers, regions or life stages.
• Individual personalization adapts ads for one person based on data about them, such as products they viewed or bought.
• Context personalization adapts ads to the moment rather than the person: the page or video around the ad, the weather, the time of day, the language or the market.

Personalization is different from targeting. Targeting decides who sees an ad. Ad personalization decides what that ad contains. Most campaigns use both, and they rely on different data.

Does ad personalization work?

The 2025 meta-analysis by Yeo, Chu and Li is the broadest summary so far, and its answer is a qualified yes: personalized ads perform better on average, through relevance. Individual studies show where the qualifications lie.

In a field experiment with an online travel company, Anja Lambrecht and Catherine Tucker found that ads showing the exact product a person had viewed were, on average, less effective than generic ads, except when the person had already narrowed down what they wanted. More specific is not automatically more relevant. The guide to dynamic creative optimization covers what this means for ad-serving systems.

Control matters too. In a randomized field experiment on a social network, Tucker found that after users were given more control over their privacy settings, they were nearly twice as likely to click on personalized ads. Ads that were targeted without personalized text showed no such change. People respond better to personalization when they feel the data behind it is theirs to manage.

Where personalized ads turn creepy

In 2004 the privacy scholar Helen Nissenbaum proposed a way of thinking about privacy that explains much of the unease. Under her idea of contextual integrity, privacy is respected when information flows in line with the norms of the context in which it was shared. Telling a pharmacist about a medication is fine. The same information appearing in an ad on a social network is not.

Research on ads supports this. Tami Kim, Kate Barasz and Leslie John found in 2019 that being open about why someone sees an ad backfires when it reveals information flows that break those norms, namely when the data was collected outside the site where the ad appears or was inferred rather than stated by the person. When the flows were acceptable, revealing them on a trusted platform increased ad effectiveness. Elizabeth Aguirre and colleagues found a similar pattern in 2015. When data was collected openly, more personalization raised people’s intention to click. When it was collected covertly, more personalization lowered it, because it made people feel vulnerable.

The business stakes are real. In a Gartner survey of more than 2,500 customers, published in March 2019, 38% said they would stop doing business with a company if they found its personalization “creepy.”

Four questions before you use a signal

1. Was it stated or inferred? Preferences people tell you are safer ground than conclusions drawn about them.
2. Was it collected in this context? Data from your own site or app, used in your own channels, fits expectations better than data gathered elsewhere.
3. Is it sensitive, or could the person be a minor? Health, religion, sexuality, politics and children’s data are restricted in law and in platform policies.
4. Could you explain it? If you would be uncomfortable showing the person “why am I seeing this ad”, the signal is probably the wrong one.

The 2026 rules for personalized ads

The rules around personalized advertising have changed considerably since 2022. The summary below is general information, not legal advice.

• Consent in the EU. In its 2023 ruling in Meta v Bundeskartellamt, the Court of Justice of the EU found that personalized advertising does not appear necessary to perform the contract with a social network’s users, and on the facts of that case it also rejected legitimate interest as a legal basis.
• Sensitive data and minors. Under the Digital Services Act, online platforms may not show ads based on profiling that uses special categories of personal data, and may not show profiling-based ads to users they know with reasonable certainty to be minors.
• A less personalized option. Under the Digital Markets Act, the European Commission fined Meta €200 million in April 2025 for failing to offer users a service that uses less of their personal data. Meta committed to give EU users a choice between fully and more limited personalized advertising from January 2026.
• Cookies. Google decided in 2025 not to introduce a new standalone prompt for third-party cookies in Chrome, and in October 2025 announced it would retire several Privacy Sandbox technologies, including Topics and Protected Audience. Chrome keeps its current approach of offering users a choice about third-party cookies.
• Apple’s tracking prompt. App Tracking Transparency remains in place, though competition authorities in France and Italy fined Apple in 2025 over how it was implemented.
• Platform policies. Meta removed detailed targeting options based on sensitive topics such as health, sexual orientation, religion and politics in 2022, and Google restricts personalized ads in sensitive interest categories.

The direction is consistent across these changes. Personalization based on detailed personal profiles is becoming harder and riskier, while personalization based on context and on data people share knowingly remains open.

Contextual personalization: the privacy-safe route

Context has always been part of advertising, and research suggests it holds up well when personal data is restricted. Avi Goldfarb and Catherine Tucker studied 3.3 million survey responses across 9,596 campaigns and found that after the EU’s earlier privacy rules came into force, banner ads lost 65% of their effect on stated purchase intent. Ads on specialized sites, such as baby or travel sites, “experienced far less of a decline.” A later study by Veronica Marotta, Vibhanshu Abhishek and Alessandro Acquisti of one large US media company found that ads served with a tracking cookie raised publisher revenue by only about 4%.

Contextual personalization adapts the ad to the situation instead of the person: a different scene for a rainy morning, a different product for a cooking site, the right language and claims for each market. It needs no profile of the individual. It does need many more creative variants, because every context the brand wants to respond to requires an ad that fits it.

AI for personalized advertising

AI does two different jobs in personalized advertising. On the platform side, Meta, Google and others use AI to predict who is likely to respond, to set bids and to decide which ad variant each person sees. On the brand side, AI helps produce the variants: adapting images, copy and formats for each segment, context and market.

The first job is largely automated by the platforms. The second decides what they have to work with. In NCSolutions’ analysis of sales-effect studies for consumer packaged goods brands, creative accounted for 49% of the incremental sales from advertising and targeting for 11%. As targeting data becomes scarcer, the variants a brand supplies carry more of the weight.

How to produce personalized content at scale without losing brand control

Personalized content at scale multiplies the number of assets a brand needs: segments times contexts times formats times markets. Three practices keep that manageable.

Build the variant matrix from signals you can use

Start from the signals that pass the four questions above and list the variants each one needs. Keep the matrix to what your campaigns can actually use and test. As the DCO guide explains, traffic limits how many variants an engine can learn from.

Apply brand and legal rules when the content is made

Hundreds of variants cannot each be reviewed with the same care as one hero ad. The rules have to be applied when the variants are produced: logo placement, typography, product depiction, approved claims and legal lines per market. Rules applied at production make every variant approvable. Rules checked afterwards create a review queue that grows with every new signal. Tools for personalizing visual content at scale differ most on this point.

Pre-test and close the loop

Attention prediction before launch and performance data after it both help decide which variants to keep. The goal is a loop in which results inform the next set of variants, while the brand rules stay fixed.

What it looks like in practice

Grip and Realeyes announced a partnership in 2024 to bring personalized, attention-optimized ad content to CPG brands, combining Grip’s rule-based content generation with Realeyes’ attention predictions. In 2026 Grip and Claritas announced a closed-loop system that connects content production with audience data and performance insights. Beiersdorf uses digital twins of its portfolio to create hyper-localized content for NIVEA and Eucerin across formats and channels.

Grip is the enterprise automation platform for publishable marketing content at scale, built by INDG. For personalized advertising, it produces the variants while the brand keeps control of them. Art direction, brand and compliance rules are encoded in its Brand Center and applied whenever content is generated. Adaptations produces compliant variations of a master asset, from Originals, a new campaign or an existing shoot, for each segment, context, market and format. Originals creates new on-brand content from a brief without a physical shoot, for contexts that need a different scene. Digital Shelf generates product imagery from digital twins of the actual product, so the product stays accurate in every version.

Where to start

List the signals your campaigns use today to personalize ads, and run each one through the four questions. Drop the ones that fail, and for the rest, count how many variants they require across formats and markets. That count shows how much of your personalization depends on production rather than data. For the wider view, read about AI in advertising, or talk to the Grip team.

Frequently asked questions

What is the difference between personalized and non-personalized ads?

Personalized ads use data about the person, such as their interests or browsing history, to choose or adapt the ad. Non-personalized ads rely on context instead, such as the content of the page, the general location, the time or the language, without building a profile of the individual.

Do personalized ads work better than generic ads?

On average, yes. A 2025 meta-analysis of 53 experiments found personalized ads generally more effective than generic ones, mainly because people find them more relevant. But personalization can backfire when the data behind it feels intrusive or was collected covertly.

Why do personalized ads feel creepy?

Research suggests ads feel creepy when the information behind them flows in ways people don’t expect, for example when data was collected on another site, inferred rather than stated, or gathered covertly. Ads based on information people knowingly shared in the same context are generally accepted.

Are personalized ads legal under GDPR and the DSA?

Personalized ads are legal in the EU when there is a valid legal basis, usually consent, for the personal data used. The Digital Services Act adds that online platforms may not show ads based on profiling with special categories of personal data, or profiling-based ads to users they know are minors. This is general information, not legal advice.

How is AI used for personalized advertising?

Ad platforms use AI to predict which people are likely to respond, to set bids and to choose which ad variant to show. Brands use AI to produce the variants, adapting images, copy and formats for segments, contexts and markets. The second part is where brand control is won or lost.

‍

References

• Yeo, Chu and Li, “How Persuasive Is Personalized Advertising? A Meta-Analytic Review,” Journal of Advertising Research 65 (4), 2025.
• Pew Research Center, “How Americans View Data Privacy,” 18 October 2023.
• Anja Lambrecht and Catherine Tucker, “When Does Retargeting Work? Information Specificity in Online Advertising,” Journal of Marketing Research 50 (5), 2013.
• Catherine E. Tucker, “Social Networks, Personalized Advertising, and Privacy Controls,” Journal of Marketing Research 51 (5), 2014.
• Helen Nissenbaum, “Privacy as Contextual Integrity,” Washington Law Review 79, 2004.
• Tami Kim, Kate Barasz and Leslie K. John, “Why Am I Seeing This Ad? The Effect of Ad Transparency on Ad Effectiveness,” Journal of Consumer Research 45 (5), 2019.
• Elizabeth Aguirre, Dominik Mahr, Dhruv Grewal, Ko de Ruyter and Martin Wetzels, “Unraveling the Personalization Paradox,” Journal of Retailing 91 (1), 2015.
• Gartner, “Gartner Survey Shows Brands Risk Losing 38 Percent of Customers Because of Poor Marketing Personalization Efforts,” press release, 11 March 2019.
• Fieldfisher, analysis of CJEU C-252/21 Meta Platforms v Bundeskartellamt, 2023.
• Digital Services Act, Article 26 (advertising on online platforms).
• Digital Services Act, Article 28 (online protection of minors).
• European Commission, “Commission finds Apple and Meta in breach of the Digital Markets Act,” 23 April 2025.
• European Commission, “Meta commits to give EU users choice on personalised ads under the Digital Markets Act,” 8 December 2025.
• Google, “Next steps for Privacy Sandbox and tracking protections in Chrome,” 22 April 2025.
• Google, “Update on plans for Privacy Sandbox technologies,” 17 October 2025.
• Autorité de la concurrence, decision on Apple’s App Tracking Transparency, 31 March 2025.
• Italian Competition Authority (AGCM), press release on Apple, 22 December 2025.
• Meta for Developers, “Updating Meta’s detailed targeting options,” 8 December 2021.
• Google Ads policy, “Personalized advertising.”
• Avi Goldfarb and Catherine E. Tucker, “Privacy Regulation and Online Advertising,” Management Science 57 (1), 2011.
• Veronica Marotta, Vibhanshu Abhishek and Alessandro Acquisti, “Online Tracking and Publishers’ Revenues: An Empirical Analysis,” WEIS 2019.
• NCSolutions, Five Keys to Advertising Effectiveness, August 2023.
• Westwood One, on NCSolutions’ analysis of creative’s sales effect, 22 April 2024.