Data: AI is chewing away at DTC search performance

AI Overview, ChatGPT, and Claude are here. Their daily active users are skyrocketing. What does that mean for us, as marketers?

Chatbots like Claude and ChatGPT are most likely to cannibalize traditional search tools like Google. A user has a question, and AI Overview is a superior solution to answer it.

Stuff like “agentic commerce” (probably not gonna happen) and “instant checkout” (weird and counterintuitive) are still pure speculation. For now, anyway.

We dug into our data on AI and came to this conclusion: If you are reading this, you must not ignore AI’s impact on your business. AI is chewing away at search, specifically.

Much has been said already about how Google’s AI Overviews are cratering search traffic to publisher and brand sites. Do you know how AI is affecting the awareness, consideration, and conversion behaviors of your customers?

Adobe claims some pretty insane numbers: AI traffic to retail sites up 393% year over year, with 85% of AI users claiming it “improved their shopping experience.”

So what does the Northbeam data say?

If our customers are shifting their behaviors and paths to purchase because of AI, we need to understand how. As growth marketers we must maintain high-quality, high-converting funnels.

What we can track is the amount of revenue attributed to AI touchpoints. The below chart shows percent of total organic visits across all Northbeam customers (summarized) that came from AI, and the percentage of organic search revenue that came from searches within an AI tool. 

This is aggregate data across all Northbeam accounts, not an average per account. 

While the chart may look dramatic, perceive the timeframe. It has taken nearly two years for the AI percentage of revenue to rise from zero to just over 7%. 

Given the excitement over AI, one would think the shift would be more dramatic. Still, this is a marked trend upwards. We have no precedent for this technology, and no way of predicting if this trend stays linear or goes exponential in the coming months. 

Somewhat hypocritically to this chart, public opinion about AI continues to waver: more than 88% of Americans would consider themselves “concerned” about AI in some way. Unlike the advent of early digital advertising, Americans are far more literate and opinionated about the rise of AI than the rise of Facebook, for example. 

While there’s a lot of questions about the future of this channel, we can see that it’s influencing search behavior. Generative Engine Optimization (GEO) and Agentic Engine Optimization (AEO) are important now, not later. Your product needs to be visible to AI.

I hypothesize that planting good AEO/GEO seeds now will create strong trees of organic growth later. It’s similar to SEO, in that aspect.

We’ll continue to publish aggregate AI reporting in this newsletter as we find new trends.

The early whisperings I’m hearing about ChatGPT ad performance are not positive. This is not a surprise. When new ad placements appear, they are often good at capturing attention (the first ever TV commercials, for example) but worse at driving sales. It takes time for users to trust ads enough to buy from them.

Nobody has good aggregate data on ChatGPT ad performance yet, so I can’t say if ChatGPT ads are “good” or not. In God we trust, all others must bring data.

What we do know: OpenAI needs to make $2.5 billion this year on ads, and scale that to $100 billion by 2030 for their business to be solvent. To compare: eMarketer expects Meta to bring in $250 billion in digital ad revenue this year. TikTok might bring in $30-$33 billion.

OpenAI’s plan is to build the best-performing ad channel in human history. No pressure?

They need to do this in less than four years and win 20% to 40% of your marketing budget, in an environment where many Americans dislike advertising, distrust AI (in theory), and oppose data centers. Not to mention consumer confidence is hitting record lows, and inflation is gutting the average American’s buying power.

Currently OpenAI is on track to miss it’s own targets on ad revenue by 90%.

This matters a lot for the stock market, but it doesn’t matter for marketers, at least not right now.

Let’s focus on the current reality of AI’s impact on marketing, rather than some speculative future.

It’s important to not conflate investment speculation with logical marketing strategy. The bold predictions made by financiers about AI shouldn’t necessarily influence our decision making. We are marketers, not hedge fund investors.

For example: just because OpenAI predicts that ChatGPT Ads will make $100 billion doesn’t necessarily mean they’re mandatory placements for you to try. Gesticulation about OpenAI’s revenue doesn’t matter if the placements can’t convert profitably for you. Focus on the results, not the headlines.

AppLovin is often on the receiving end of this conflation, unfairly. Marketers often read headlines aimed at retail investors who own AppLovin stock and mistake these headlines as some type of judgement on the performance of AppLovin as an ad platform. Wall Street and marketers have two separate perspectives with two separate goals. Don’t get them mixed up.

I am long on AppLovin as an advertising platform. I have been for more than a year. If you are considering trying out AppLovin, I implore you to read their case studies, not content aimed at retail investors.

The same goes for AI. As users continue to grow and headlines abound, stay focused on results. Questions to ask yourself:

⏰ Google pulled ad group targets into the upcoming changes to bidding. Complex stuff here, but interesting read on how you’ll be affected.

🎙️ Vibe and Audien are crushing CTV. A thorough case study on how you can actually get started with CTV.

🎯 Early research on ChatGPT placements suggests they’re firing all over the place. But do we really have a baseline for this performance?

🏍️ Northbeam Apex is blowing up, just in time for BFCM. 10% more revenue on Meta with the flip of a switch? Sign me up.

Love this newsletter? Please forward it to your friends, coworkers, or Julie Masino. Subscribe here to get this data in your inbox every week.