Adam's Take
OpenAI shares *some* data on brand mentions
Tracking brand influence within an LLM is tough. ChatGPT appends a simple UTM code to referral traffic, but what happens inside the LLM’s results has been a complete black box for brands. In a survey earlier this year, Branch found that over 80% of enterprise marketers face fundamental challenges tracking AI’s influence.
So there was reason for hope last week when OpenAI introduced “Businesses” in ChatGPT, which promises insights into "discovery topics,” or “common themes across conversations where [a brand’s] website is cited.” The most exciting part, though, was the promise of brand “mentions” as a metric. That could finally give us a quantitative view of how brands are referenced with the LLM.
Unfortunately, a deeper dive shows it’s not as complete as we’d hoped. It seems a mention only counts when the business URL is included in an answer. And only half of conversations include a link, often a single URL. You’ve likely seen a ChatGPT answer that will only include a single link from a list of recommendations; simply put, sharing the occasional link inclusion doesn’t capture the full picture.
That’s not to say brands won’t adopt it. The ChatGPT roadmap increasingly makes it clear that brands that don’t lean in will be left out of results. Since this release gives businesses their interface for connecting with ChatGPT, it will likely be the first stop for any brand that even remotely cares about its presence there.
Marketers knowingly lose over a quarter of their budget to fraud
A quarter. Let that sink in. Over 25% of marketing dollars are stolen, and marketers know it. And AI is going to make it a whole worse. Over 80% of marketers are already seeing it happen, and 98% are changing the way they buy as a result.
A recent Branch survey explores how senior marketers are fighting ad fraud and preparing for the onslaught of AI-powered attacks.
The strategies to combat fraud basically boil down to better measurement and a flight to trusted platforms. Understanding the true ROI of ad spend fundamentally beats any method of fraud, and consolidating spend helps with better measurement. It’s a disarmingly simple concept, but that doesn’t make it any easier to accomplish.
Amazon impression counts get honest
Did you know that Amazon currently counts an impression when an ad is served, not when a shopper actually sees it? Now guess how many served impressions are never seen. 63%.
That’s how many fewer ad impressions Amazon predicts advertisers will record when it switches to measuring impressions as seen rather than served on Jan. 1.
That may sting for those who’ve been running CPM campaigns; for those running CPC campaigns, it’s a fairly moot point. But it’s another reason to keep moving downstream from proxy metrics to measuring true ROAS. Amazon impressions are cheaper because most of them are never seen. When you judge advertising by proxy metrics, you get what you pay for.
OpenAI ad tech moves fast with tCPA buys
OpenAI made CPC bidding generally available in May, less than half a year later, it launched targeted cost-per-action (tCPA) buying. Facebook took five and a half years to make the same progression.
This means advertisers can move the system’s algorithm from optimizing for a proxy metric like ad clicks to optimizing for a revenue-generating user action much further downstream, like a purchase.
This allows advertisers to do far less guessing about an ad’s ultimate success. Sure, there may still be shared attribution with other ad sources, but dividing up credit for a purchase is far easier than discerning the influence of incoming traffic.
Podcast
How I Grew This: Strategy, Taste, and the Hybrid Future: How AI and Community Are Rewiring Growth with Lomit Patel
In the latest episode of How I Grew This, Amanda and Adam welcome back growth veteran Lomit Patel, author of Lean AI and growth executive across Roku, IMVU, Tynker, and TYB. Lomit breaks down how AI is reshaping marketing team structures, why human judgment and taste are more critical than ever, and how digital-first Gen Z brands are leveraging offline retail ecosystems and gamified communities to build long-term retention.