What this covers

This is a conversation in which Jonathan, who runs the agency Thrive, explains how he approaches performance marketing, with Matt and Dom asking questions. He starts from his own origin story: he realized every Google search keyword was really an audience, and that much of the inventory went unbid. From there he describes a method that works backward from customer LTV to the clicks and audiences that produce good customers. By that logic, a twenty-dollar click is not irrational if it brings in high-quality buyers.

The middle of the discussion is about what can and cannot be known. Jonathan says privacy changes, including GDPR and iOS 14.5, broke the link between ad platforms and iPhone data, so one dollar in and one dollar out can no longer be traced. He walks through the substitutes: media mix modeling with third-party tools like Recast, geo holdout tests he admits are never clean experiments, and user surveys that suffer from self-selection. He also covers Attribution Models as subjective crediting, lead-generation latency of one to 24 months, and why a return that works at small spend won't hold at scale (see Diminishing Returns at Scale).

The practical advice is on how campaigns fail: creative treated as an afterthought, unrealistic timelines, and weak measurement. Jonathan also argues that creative should match the Marketing Funnel stage. For new channels he proposes an experiment inside a small 80-20 Budget Allocation, with a stated thesis and results measured three ways. Podcast advertising is the contested case: attribution is murky, and Dom relays a claim from a veteran podcaster that paid growth in the medium rarely works.

Sharpest takeaway

Becker argues paid growth works like value investing: undervalued ad inventory is bought against lifetime-value economics, success requires patience and the right metrics, and because platform measurement has degraded, advertisers need independent third-party validation.

  • Ad inventory that others overlook is mispriced, so it can be bought cheaply relative to the lifetime value it yields
  • Measurement has gotten worse due to privacy changes, so platform-reported results need independent validation through holdout tests, media mix modeling, and surveys
  • Campaigns need time and correct success metrics, and creative is a core driver rather than an afterthought

The argument · threads3 threads · 74 claims
0.85

Podcasting's ad underperformance stems from weak measurement linking exposure to outcomes and data

6 pointscentrality 5/5
  • Matt argues that programmatic advertising produced much more data than traditional magazine and television advertising, which lack effectiveness data, and that the shift toward data-driven advertising was inevitable, though the industry still leaves much to be desired.

    “you had so much more information, and I just thought to myself, this is a $700 billion industry”

  • Walled-garden platforms such as Overcast tell advertisers what their dollars returned in listeners but provide no way to track what happens to those listeners afterward, making comparative data across platforms very hard to produce.

    “it's that concept of walled gardens, particularly make it very hard to make any type of comparative data”

  • Amazon's advertising business generates about $40 billion while podcasting, after 20 years, is worth roughly $2 billion to $3 billion, which Matt attributes largely to the difference in measurement.

    “Amazon started an advertising business two days ago, and they're generating $40 billion. Podcasting has been around for 20 years now, and it's $2 billion to $3 billion.”

  • Programmatic advertising, when it emerged about 15 years ago, offered far more data than magazine and television advertising, and the ad market now runs at roughly $700 billion to nearly $1 trillion, though measurement quality remains a weak point.

    “this is a $700 billion industry, probably close to $1 trillion industry right now.”

  • Podcasting is stuck between brand and performance advertising models, and the friction of connecting ad exposure to outcomes is the central problem that a solution would unlock.

    “podcasting is stuck in this, in between of, should this be brand advertising, should it be performance-based.”

  • Given the loss of data after iOS changes and privacy shifts, brands will likely converge on blended measurement approaches and build their own measurement tools.

    “I think a lot is actually changing with the lack of data that is now accessible post iOS changes and all the privacy changes.”

0.47

Large platforms' ad spend may reflect scale and self-fulfilling beliefs rather than superior returns

4 pointscentrality 3/5
  • Where money is spent is not necessarily where the highest return is, and Google and Meta spend could reflect a self-fulfilling prophecy rather than superior returns.

    “It's spent there, but that doesn't necessarily mean that it's the most valuable.”

  • Smaller channels have less inventory than Meta and Google, so advertisers can hit a ceiling on them and see diminishing returns, while the largest platforms can absorb large budgets.

    “Whereas some of the smaller channels might have less inventory, you can definitely cap out”

  • Matt argues that awareness from large organic reach can drive later conversions as listeners who saw the content lag in visiting or buying, so the impact may show up in numbers over time.

    “maybe there's a lagging impact from those 900,000 people making their way from the Twitter.”

  • Matt describes the mobile gaming industry as a closed ecosystem in which mobile games are advertised on top of mobile games, built on the back of Instagram and Google spending, which he considers a separate conversation.

    “that's a weird sick ecosystem, though, where it's very much a closed ecosystem where it's mobile games on top of mobile games”

0.24

The podcast's audience is intelligent, senior, and high in buying power

2 pointscentrality 2/5
  • Matt believes the Making Media audience is among the most intelligent and buying-power-rich audiences, with many senior professionals, though he acknowledges this is his own bias.

    “I think it's the most intelligent and community or audience with the most buying power, honestly”

  • Matt argues the podcast's audience is highly intelligent, senior in their careers, and has substantial buying power, which makes it attractive to advertisers but also raises the measurement challenge.

    “it is, I think, the most intelligent and community or audience with the most buying power, honestly”