Confessions Of An Agency Owner

The first half of the episode is almost entirely personal narrative - accidental agency founding, sabbatical plans, beard growth, Disney trips, book titling mishaps - with minimal operational takeaways. The second half on intent scoring contains genuine substance (direction vs. strength of intent, critique of conversion rate as retrospective and binary, the 120 micro-behaviour model), but the insight-per-minute rate is dragged down by sustained football banter and lifestyle anecdotes.

“There's two forms, there's direction of intent. So I'm in the car, I'm in the industry to buy a Maserati or a Ford. And then there's the strength of intent, which is I'm in the how much do I want to buy the Maserati or the Ford?”

“The two biggest metrics, the two biggest attributes that we look at are time and specificity.”

The XG/expected-goals analogy mapped onto e-commerce conversion metrics is a genuinely fresh framing, and the direction-vs-strength-of-intent distinction adds a useful conceptual layer. However, the Moneyball reference is overused, and the broader 'be more human / personalization matters' thesis is well-worn territory in CRO and martech circles.

“instead of expected goals, it's expected conversion. Instead of expected assist, it's expected uh, add to bag. Instead of expected possession, it's expected engagement”

“we focus so much on this aggregated figure of conversion rate that is binary. It's rare in the similar way that football did in the 2000s”

David Mannheim is a legitimate practitioner: he founded and scaled a CRO agency to acquisition, now builds a data product he has already validated with paying clients and published a book on the domain. He is speaking from direct operational experience, not as a career thought-leader. His profile is real but not widely recognised beyond a niche CRO/e-commerce audience.

“After scaling and selling his agency User Conversion to Brain Labs”

“I stayed on for a little bit, yeah. Two days a week for 12 months just to oversee the transition”

There are concrete data points - Disney's 40% per-day cost increase, the 120 micro-behaviours in the model, Google's RankBrain in 2015, the Land Rover car-configurator example with 14-year-old boys skewing GA demographics - but actual client conversion-lift figures are conspicuously absent ('really good results' is as specific as it gets), and many product claims remain at the level of architecture description without outcome evidence.

“on average you were paying 40% more Ah, per day at ah, a park”

“we take about 120 or so micro behaviors”

The host occasionally steers toward useful clarification ('So how does it work then?', 'Is this a piece of software that sits on the site?') but does not challenge any substantive claims - there is no pushback on whether the XG analogy actually holds, no request for hard client metrics, and no probing of competitive differentiation. The conversation frequently drifts into Disney fandom, football scores, and personal life, which the host does little to redirect.

“So how does it work then?”

“But we're not getting manipulated by the platforms there. So Google Ads, you know, we go and retargeting. I'm not sure maybe we can, but can we decipher that they've visited X, Y and Z pages”