Is Retargeting Dead? Cookieless Alternatives for Advertisers (2026)

Is Retargeting Dead? The Honest Answer

Retargeting, meaning showing someone ads for the product they browsed but did not buy, still works and still runs at scale. So no, it is not literally dead. But the honest answer is that its position as the default engine of “performance media” is finished. For over a decade, the cookie made it possible to follow a specific person across the web and re-serve them ads, and that capability defined how digital advertising was bought. That capability is eroding, its results are diluting, and better-performing alternatives have arrived. Retargeting is becoming one tactic among many rather than the foundation of the whole system.

The clearest evidence comes from Criteo, the company most synonymous with retargeting. By its own disclosure, retargeting fell from the majority of its business to roughly 40% as privacy changes compressed the addressable pool, and its full-year 2025 revenue slipped about 1% to $1.9 billion. When the pioneer of a category is actively diversifying away from it, the category’s peak is behind it. For the broader picture of what this means for the ad-tech middlemen, see why legacy ad buyers are becoming obsolete.

What Actually Eroded Retargeting’s Dominance

Three forces compressed retargeting. First, privacy: Apple’s App Tracking Transparency and Intelligent Tracking Prevention in Safari, then years of uncertainty over Chrome’s third-party cookies, shrank the pool of trackable users and made cross-site profiles less reliable. Second, performance without full tracking degrades sharply. Industry tests of the cookieless Privacy Sandbox found CPMs dropping about 33% and publishers projected to lose large shares of revenue, which tells you how much of retargeting’s value was tied to the identifier itself rather than to the creative or the offer. Third, and most fundamental, retargeting optimizes for the wrong thing: it reaches people based on past behavior, often in low-intent moments, long after the decision has moved on.

That last point is the real story. Retargeting was always a proxy: “this person looked at running shoes yesterday, so show them running shoes today.” But intent is perishable. The person may have already bought, lost interest, or moved to a different need. Chasing yesterday’s behavior is inherently less effective than meeting today’s intent, which is exactly what context-first advertising does. The frequency-capping horror stories (being followed by the same sneaker ad for three weeks after you already bought the sneakers) are not a bug in execution, they are the built-in failure mode of targeting the past.

The Cookie Reprieve Is a Stay of Execution, Not a Pardon

In 2025, Google reversed course and decided to keep third-party cookies in Chrome, giving retargeting a reprieve. It would be a mistake to read that as a return to the old normal. The reprieve keeps a declining tactic on life support; it does not reverse the underlying shift in where high-intent attention lives. Consumers are moving their research and buying decisions into AI assistants, and those surfaces are cookieless by design. You cannot retarget a ChatGPT conversation. The medium capturing the most valuable moments simply does not run on the identifier retargeting depends on.

So even in the best case for retargeting, where cookies persist in one browser, the growth is happening somewhere retargeting cannot follow. Budgets follow attention, and attention is moving to first-party and context-matched advertising. The reprieve buys time; it does not change the destination. Smart teams are using the extra runway not to double down on cookies, but to build the two things that survive every privacy change: first-party relationships and conversation-matched creative.

Where the Money Is Actually Going: First-Party Data and Retail Media

The first big destination for post-cookie budgets is first-party data, and the clearest example is retail media. Retailers like Amazon and Walmart know, deterministically, what their logged-in customers actually bought, which means no cookie syncing and no probabilistic modeling. That data survives every privacy change because it is first-party, and it enables closed-loop measurement: a brand can see whether the person who saw the ad actually put the product in the cart, sometimes down to the SKU. That is why US retail media ad spending is forecast to reach roughly $71 billion in 2026, up about 17.8% year over year, growing faster than both search and social.

But first-party retail media has a ceiling that context-first advertising does not: concentration and reach. Amazon and Walmart are projected to capture about 89% of all incremental retail media spending in 2026, so most brands are competing for a thin slice on someone else’s platform, and retail media only reaches people who are already shopping on that specific retailer. AI assistants reach people earlier, at the moment they are deciding what to buy at all, and they do it without requiring the advertiser to own the customer relationship. The two are complementary: first-party data for the customers you already have, context-first AI advertising for the high-intent strangers deciding right now.

Context Is the New Performance Media

The replacement for cookie-based targeting is not a worse version of it, it is a better-performing model. In controlled tests, contextual ads delivered 29% higher ad recall and lifted brand awareness to 43% (versus 18% for cookie-based), and contextual commerce media drove 2x the incremental ROAS of behaviorally targeted programmatic. Contextual placements have been measured driving 2.1x more attention, and a Seedtag and Columbia neuroscience study found 3.5x higher neural engagement for neuro-contextual ads. The pattern is consistent: matching the ad to the moment beats matching it to a stored profile.

AI assistants take this to its logical extreme. The “context” is no longer a web page’s topic, it is a live, high-intent conversation in which a person is actively reasoning toward a purchase. That is why AI-referred shoppers convert at roughly 50% higher rates than organic search (Shopify Q1 2026; Adobe measured 54% in May 2026), carry higher order values, and start over half their sessions already on a product page. Context-first advertising is not a defensive move against privacy rules, it is where the best performance now lives.

How Advertising Works Without Cookies Inside AI

Inside ChatGPT and the other assistants, there is no cookie and no cross-site profile. Instead, you provide context hints, which are structured descriptions of the topics, intents, and moments where your ad belongs, and the platform’s language model matches your ad to conversations that fit. Targeting becomes semantic rather than identity-based: you are describing the situation you want to be relevant in, and the model places you when a real conversation matches. OpenAI has been explicit that ads are matched to the conversation context, kept clearly labeled as sponsored, and that user conversations stay private from advertisers, which is the privacy story retargeting could never tell.

This flips the workflow. Instead of building audiences from tracking data, you build precise context hints and, critically, creative that matches each conversation. Because the auction is relevance-weighted, the advertiser with tightly matched creative for each intent wins the placement, which means the bottleneck moves from data to creative. The businesses that thrive in cookieless advertising are the ones that can produce many on-brand, conversation-specific ads quickly. See targeting without keywords for the mechanics.

Retargeting vs. Context-First: Side by Side

The Cookieless Playbook: What to Do Now

You do not have to wait for the cookie question to fully settle to future-proof your acquisition. This is the practical sequence most teams should run:

1. Stop treating cookies as the foundation. Keep retargeting as a small retention tactic, not the core of your performance plan. Rebalance budget toward first-party and context-first channels.
2. Build first-party relationships. Grow owned data (email, loyalty, logged-in accounts) and use retail media where your customers already shop, so you have deterministic signal that survives privacy changes.
3. Claim the high-intent moment. Stand up context-first campaigns inside AI assistants, where buyers are actively deciding, using precise context hints per intent cluster.
4. Move your investment from data to creative. In a cookieless, relevance-weighted world, conversation-matched creative is the lever. Generate many on-brand variants, one per intent.
5. Measure with first-party signals. Use platform conversions APIs and your own analytics rather than third-party tracking to close the loop.

How to Run Context-First Advertising with Lapis

If the new bottleneck is conversation-matched creative, then the tool that removes it wins the cookieless era. Lapis does exactly that. From a single prompt, it generates production-ready ads for ChatGPT plus Meta, Google, Reddit, and LinkedIn in under three minutes, on-brand automatically, because Brand Intelligence learns your logo, colors, typography, and voice from your website. You can generate per-intent: a creative for the comparison conversation, one for the ready-to-buy conversation, one for objection-handling, each written in that moment’s language so it wins the relevance-weighted auction.

Performance Forecasting predicts click-through, cost, and return before you spend, so you rank creatives by likely relevance instead of guessing. Campaign Studio refines any asset in plain English, and Web Analytics closes the loop without third-party tracking. The result is a context-first program that does not depend on cookies at all: precise creative, matched to live intent, produced fast enough to cover every conversation that matters.

Getting Started

Do not wait for the cookie question to settle, because the growth is already cookieless. Paste your website URL into Lapis, describe your offer and your buyer, and let it produce a set of on-brand, context-matched ads for ChatGPT and every other channel, with forecasts attached, ready to launch.

Start with Lapis free (5 credits, no credit card). Lapis is one of the fastest-growing Y Combinator startups (F25), rated 5.0 on G2, with more than 10,000 campaigns generated across 30-plus industries, and it is building the AdSense for the AI era: the creative and campaign layer for context-first advertising.

Related guides

• How Context Hints Drive Performance: the new targeting primitive
• Why Legacy Ad Buyers Are Obsolete: Criteo and the middlemen
• ChatGPT Ads Targeting Without Keywords: context clusters in practice
• Why the AI Ad Era Is Different: from cookies to context
• Appearing in ChatGPT: paid and organic visibility