This article is based a Q&A with Breaker Founder Ben Billups and Greg Kihlström for The Agile Brand Guide’s Expert Mode series.
For most of 2026, the email industry has been selling marketing leaders a story about trust. Validity’s Q1 research found that 40% of consumers would trust a retailer’s emails less if they knew AI had written them. That number has been in a lot of decks since. The fix it implies is some combination of disclosure, human review gates, and writing less with the machine.
Ben Billups, founder of Breaker, a B2B newsletter platform in Austin, thinks the diagnosis is wrong. Asked point-blank whether trust or relevance is the binding constraint in B2B email — and which one he’d defend if forced to choose — he didn’t reach for the more marketable word.
Trust Is What Relevance Earns
The trust framing is appealing partly because it’s easy to act on. A brand can publish an AI disclosure policy in an afternoon. Billups’s argument is that doing so addresses a symptom of something structural, and that in B2B the reader isn’t really evaluating sincerity at all.
“Relevance. Trust is a function of relevance (and reliability) over time. A B2B reader is evaluating whether you understand their problem well enough to be worth a meeting. The first touchpoint isn’t really about trust (yet), but demonstrating relevance. Demonstrate relevance 15 or 20 times in a row, and you have trust.”
The distinction has budget consequences. If the problem is trust, the response is governance, and governance is cheap. If the problem is relevance, the response is audience definition and targeting, which costs real money and takes longer to show results. Billups adds a line that should make a few CMOs wince: what AI changed, he says, is that generic content is now free, so “irrelevance is cheaper to produce at scale.” The marketers who get hurt are the ones who read that as an opportunity for volume rather than a targeting problem. Worth asking where your own AI investment actually landed this year. If it went into production speed and not into knowing who’s on the list, you bought the cheaper half.
The Signals That Look Predictive and Aren’t
Billups is unusually blunt about the reporting most teams stare at every Monday. Open rates get inflated by automated opens. Security scanners click links to check for malware, which shows up as engagement that never happened — Breaker filters those out of its own reporting. And a growing subscriber count feels like progress whether or not anyone on the list could ever buy.
“There are helpful leading indicators (new subscribers in your ICP, verified clicks, etc.) and then there are definitive lagging indicators (calls on the calendar and contracts signed)… Someone who has opened 5 newsletters is far less predictive of purchase than someone who has clicked your demo link once.”
That’s a weighting argument, not a scoring one, and it cuts against how most lead scoring models are built. But he attaches a caveat that matters more: don’t discount the weak signals too early, because they compound. Someone still opening their thirtieth newsletter is a different person from someone who opened five and stopped, even though both look like “opens” in a dashboard. He puts the typical B2B path at 12 to 21 interactions before a prospect takes any action, which is his figure rather than a published benchmark — treat the shape of it as the point, not the precision. Failure detection, by contrast, is crude and cheap. Unsubscribes climbing, or engagement flatlining, and you’ve aimed at the wrong people.
The Awkward Part: Buying the Audience
Breaker doesn’t only send newsletters. Its own marketing describes growing a customer’s list with verified B2B prospects for as little as $1.50 each. So an argument that brands should build owned audiences and stop broadcasting at people who don’t care is coming from a company that sells subscribers who never went looking for that newsletter. Asked to argue against his own product, Billups went after the premise of the question instead.
“I’d push back on ‘interested.’ Nobody knows they’re interested in your newsletter before they’ve read it. Engagement with the content is what demonstrates interest.”
It’s a self-serving answer and a reasonable one at the same time, which is what makes it useful. He does give ground in two places. Paid growth fails, he says, when the leads are the wrong people — or when they’re the right people and nobody sent them anything worth reading. And there’s such a thing as growing too fast: before a company has product, offer, and audience lined up, buying speed just funds a wrong direction more efficiently. His counter-argument on organic growth is the sharpest thing in the answer. Organic list growth, he points out, is usually a residual of money already spent on events and social, “just without the attribution.” That’s true. It’s also exactly what a vendor selling attributable list growth would say. A marketing leader can hold both of those at once.
Ten Months of Sending, Then Nine Deals
The patience problem is where newsletter programs usually die. Billups says teams commonly quit around month three. Asked for a real case rather than a principle, he gave one — with a detail he clearly didn’t expect.
“A great example is an agency we worked with that currently has about ~80,000 subscribers (ecommerce niche). They initially invested in growth, then quit growth but kept sending their newsletter, and later closed 9 six-figure deals from subscribers that joined 10 months prior. I asked the owner if they had changed their offer or their content, and to my surprise, the answer was ‘no.'”
Nothing on the marketing side changed. What changed was on the buyer’s side — a development budget got approved, or someone new took a seat. That’s an uncomfortable story for anyone defending a line item every quarter, because it means the mechanism is mostly waiting. Billups frames the economics as a total addressable market problem: if your TAM is 50,000 people and you get to speak directly to 25,000 of them every week, “every founder in their right mind would take that deal.” The more useful part of his answer is what to watch in months four and five, before any of that pays. Two things only. Are the new subscribers actually in the ideal customer profile, and are they engaging at or above benchmark with unsubscribes below it. Both are answerable by week six, and neither one needs attribution to work.
Which points at the two numbers a marketing leader can calculate this week without buying anything. Billups divides cost per lead by ICP match rate to get a cost per qualified lead, and divides it by open rate to get a cost per engaged lead. Most B2B teams he talks to land somewhere between $10 and $50 on both — again, his estimate from customer conversations rather than survey data. The formulas don’t require any new tooling. They require knowing what share of your list could plausibly become a customer, which is the number most teams have never actually computed. From there it’s the familiar chain down to customer acquisition cost, plus one question worth asking your demand gen lead directly: when a subscriber books a demo or replies asking for information, does sales ever hear about it?
The trust conversation asks whether a reader can tell a machine wrote the email. That’s the easier question, and it’s the one the industry has been selling. The harder one is whether the person receiving it was ever going to buy. Most teams can answer that with data they already have, and it costs nothing to start.
