Beyond B2B Marketing

The episode contains several genuinely non-obvious insights - the 1,400-prompt average needed for consistent AI results, SparkToro's doubling of AI visibility with zero measurable sales lift, and Google's AI Mode 0.13% click-through rate being obscured by a 'billion users' framing. However, roughly a third of the runtime is origin stories, video game studio anecdotes, and affirmatory back-and-forth that dilutes the density significantly.

“you would have to ask uh these AI tools the same question 1400 times on average, 1400 times to get the same list of brands in the same order. So ranking ranking is just pointless.”

“YouTube, by the way, hugely undervalued for B2B. So is Reddit. Like when you look at the SparkToro data for any given B2B audience, and then you look at the number of people investing in those places, it's just embarrassing.”

The most original contribution is Fishkin's own first-party data showing SparkToro went from 23% to 61 - 63% AI visibility with no detectable business impact - a genuinely contrarian data point against the prevailing AI-visibility hype. Other takes (brand > demand, don't be everywhere, AI slop can't rank) are increasingly common in B2B marketing circles.

“Has it done anything for our business? Have we gotten any more leads? Have more people come and signed up? No. As best we can tell, it has had no impact whatsoever. We have more than doubled our AI visibility with no impact.”

“I don't believe that is because Lee Google has, you know, these incredible algorithms that can spot AI. That's not what I think is happening. What I think is happening is Google has very good algorithms for identifying whether human beings pay attention.”

Fishkin is a genuine practitioner - he founded and scaled Moz, built SparkToro on top of proprietary clickstream panel data, and draws on live operational experience rather than punditry; his AI visibility statistics come from his own product's tracking. He is well into 'career podcast guest' territory which tempers the score, but the substance he brings is real and differentiated.

“we published a a study on SparToro, and we don't compete with this company. I don't do anything with AI tracking, but I just found it to be such BS.”

“SparkToro was visible in audience research, you know, prompts across Chat GPT and Claude and Google AI mode, uh like 20% of the time, 23% of the time. Now we're in the 60s, 61, 62, 63%.”

The episode is notably rich in specific numbers and named entities: the 1,400-prompt consistency threshold, 0.13% AI Mode click rate, SparkToro's 23%→63% visibility arc, Profound.com named and its Andreessen Horowitz backing called out, Datos as the clickstream provider, the $2.15M Snack Bar raise, and the DOJ trial engineer testimony on click/bounce signals as top ranking factor. Minor deductions for some statistics (e.g. the 83%/17x influencer figures) being cited without published methodology visible in the transcript.

“when Google says AI mode reached a million users in a month... That number is super, super tiny. Now, granted, super tiny means like 0.13%”

“we raised two, two point one five million dollars”

The host opens with excessive flattery and several questions are broad or leading rather than probing; there is no meaningful pushback or productive disagreement across 70 minutes. Lee partially compensates by contributing his own research data (always-on influencer program stats, agency inquiry close-rate data), which elevates the exchange above a pure PR softball session, but the craft ceiling is low.

“He's not only the best answer for all things search and audience intelligence, but also as an entrepreneur and founder of multiple companies in the Martech space”

“83% our research found of uh B2B companies that are implementing influencer marketing in an always-on fashion are rate their programs as being extremely successful”