Deconstructing Data

A handful of genuinely useful ideas surface - using aggregate client data across thousands of accounts to benchmark a single client's slowdown, the MCP trick to get keyword-level ad performance that Google's own UI hides, and Facebook banning MCP users to protect ad revenue - but they are surrounded by long tangents, filler agreement, and obvious observations (digital is measurable, content matters, small businesses should focus on their core job).

“the PPC analyst who's managing that dentist could literally go into Claude and say my client is having a slow month. Is it slow for everyone? Is it geographically slow for those that's not slow for what are they doing differently?”

“you can't know which ad does the best for a specific keyword... With the Claude mcp, you could literally ask it that”

There are a few genuinely fresh angles - the observation that AI search is skewing upper-funnel while transactional intent still flows to Google, and the Facebook ad-spend incentive argument for banning MCP - but the episode leans heavily on circulating takes: wild-west AI analogy, 'content is king,' and GEO/AEO-is-just-SEO, which Google itself has already published.

“AI search continues to be more upper funnel... ultimately when people kind of make a decision... they wind up actually on traditional search finding those businesses”

“Facebook, if all of a sudden everyone is not spending money on certain audiences, they can't monetize them. So that's why they were banning people”

Alex Mellon is a genuine long-tenure operator - co-founder of a 400-employee agency, former digital lead for Walmart and Samsung at Publicis, and in the space since 1997 - giving him real practitioner credibility; however, the conversation rarely extracts the depth that caliber should yield, and several claims (e.g. 'most cited brand in AI') go unsourced.

“we are the largest private digital marketing company in the U.S. um, a couple of our competitors took on equity or sold... we have over 400 employees, um, work with thousands of clients”

“I was in charge of digital for Walmart for a little bit Samsung before that”

The episode contains some good concrete anchors - Forbes's 67 apple pie recipes, Reddit stock up 30%, negotiating a car sale up $15,000 overnight via Claude, Facebook spend accounts at $1M/month getting banned - but there are no campaign performance metrics, conversion rate lifts, or before/after data from actual client work, which is the kind of evidence that would truly validate the agency's approach.

“They actually have uh, 67 apple pie recipes recipes on Forbes”

“It bid people up by like $15,000. Is that crazy?”

The hosts ask broad scene-setting questions ('Walk us through this topic, Alex') and consistently validate rather than probe; there is no pushback on unchecked claims like SmartSites being the most-cited agency in AI search, no follow-up on methodology, and Jesse cuts topics short due to time rather than pursuing depth, resulting in a largely promotional chat.

“That's a great point. And that's a really interesting data set that aggregated data across an industry is super valuable.”

“Yeah, I mean I remember that for years everybody was sending emails out saying we'll rank you number one”