Marketers Were Blind to AI in 2025. In 2026, It’s Back to Basics

Marketing operates a predictable loop: measure performance, optimize based on data, repeat. But in 2025, we’ve watched the fastest-growing technology adoption in human history unfold while our ability to measure it lagged years behind.

The pace caught marketers flat-footed. 2026 is when they catch up.

2025 Reflections

Artificial intelligence (AI) growth is unprecedented. Mary Meeker, the referee on industry growth benchmarks for the last decade, published an authoritative 340-page report on AI trends that makes all previous digital adoption curves look positively sluggish. Case in point: PCs took 20 years to reach 50% adoption in the US; AI is estimated to take three. OpenAI now predicts their models will run autonomous companies and simulate human minds within a decade.

We lost control of the funnel. “LLMs are not search engines, they’re answer engines”— a quip I first heard this year— perfectly captures an emerging problem for marketers. By design, large language models (LLMs) provide a contained experience that decreases the need to click out to other sources. As LLM usage grows, we’re hemorrhaging organic search and sitewide traffic because discovery is happening entirely within the chatbot experience.

A SEMrush study found that an LLM referral is worth 4.4x more than someone coming from organic search. This makes sense; traffic that clicks out of an LLM is much higher intent. The same study estimates LLM traffic will overtake organic search by 2028, which means this visibility problem is only going to get worse. The reality is the top level of your funnel is happening inside ChatGPT, Claude, and Gemini. You just can’t see it yet.

LLM measurement is in the dark ages. We have almost no ability to measure the impact of LLMs. Right now, we’re stuck literally asking the LLM itself where we rank. But getting back a simple list of where you rank against competitors feels a lot like the early days of search: interesting, but not very actionable. The SEO adage from the early 2000s, “I know where I rank. I just don’t know what to do about it,” is back in a big way.

Just like the days before Google Analytics, the most advanced marketers have resorted to combing through server logs to find the LLM search-agent traffic, which still doesn’t show up reliably in most analytics platforms.

2026 Predictions

OpenAI will start advertising — and create a land-grab moment. Analysts are already lambasting OpenAI for taking too long to implement an ad model. In 2026, OpenAI will finally move into advertising. Sam Altman expects this to be some type of affiliate fee, but what matters to marketers isn’t the mechanics; it’s what it represents: the birth of an entirely new paid media channel.

All models will explore some type of monetized traffic or commerce model. Remember when Facebook’s e-commerce advertising spawned an entire DTC category and kicked off a performance marketing boom? LLM-sponsored placements will create the same land-grab opportunity for brands willing to experiment early. And critically, once LLMs start monetizing traffic, they’ll be forced to create methodologies to help brands measure the media efficacy.

The brands testing placement strategies, bidding approaches, and learning how to measure conversions will have years of advantage over those who wait.

All major analytics platforms will start measuring AI traffic. Yes, platforms technically track LLM click-out traffic today using generic UTM codes like “chat.openai.com,” but most referrals still come through AI web crawlers — and because those are classified as bots, analytics platforms miss the full picture.

In 2026, expect LLM crawler identification, AI agent referral classification, visibility into event logging, and the emergence of new attribution models (conversation to conversion) — provided LLMs emit the necessary metadata. This will accelerate as they test and scale monetization and advertising.

Measurement is step one. What to do about it comes next.

Marketers will start to optimize for LLM traffic. Here’s the good news: Emerging academic research shows us what influences LLM recommendations, and it’s surprisingly familiar territory. LLMs have human-like biases that respond to classic optimization techniques:

• Words carry weight. LLM selection can be influenced by an average of 5% based on a single word in the description. In certain categories, this jumps up to 25%.
• Position matters. ChatGPT is 3x more likely to choose a product in the top row.
• Social proof wins. LLMs show a positive bias towards strong reviews and social validation, and negative bias towards higher prices and sponsored posts.

These are familiar tactics, but the playbook isn’t written yet, and the experiments need to start now. Marketers who treat LLM optimization like they treated early SEO— test relentlessly, measure obsessively, iterate quickly— will establish dominance before this becomes table stakes for the industry.

The Opportunity Ahead

The bad news: The world has changed faster than our measurement systems. We spent 2025 unable to track one of the most valuable emerging traffic sources. The good news: That’s already shifting. The infrastructure is coming, and the measurement tools will catch up.

We’ve been here before: new channels, broken attribution, unclear ROI. Every time, the brands that moved early and stayed disciplined won. AI traffic is no different. The opportunity is massive. The window to establish an advantage is now.

Adam Landis was the founder and CEO of AdLibertas, a mobile app data platform acquired by Branch in 2022. At Branch, he serves as Head of Strategic Growth, where his deep experience in mobile advertising and data helps foster innovation in products that increase the ability to measure marketing performance in an increasingly difficult ecosystem.