AGENCY GROWTH CLUB
Jason Berkowitz (Break the Web) on Why Nobody Has a Proven AEO Playbook Yet
Guest: Jason Berkowitz
Guest: Jason Berkowitz
Jason Berkowitz, founder of Break the Web, isn’t chasing every new AEO or GEO acronym, he’s watching for actual data, and says nobody, including him, actually has it yet.
On this episode of the Agency Growth Club, he breaks down why ChatGPT’s internal knowledge base hasn’t updated since April 2024 (meaning visibility in Google still drives most of what surfaces in AI answers), the custom GPT he built specifically to challenge his own thinking instead of agreeing with him, and why “hire fast, fire faster” is advice he says he still has to relearn.
Pulled from the transcript’s real per-line timestamps, matched to where each topic starts.
Jason’s clearest use of AI at Break the Web is straightforward workflow productivity, cutting down individual team members’ time on recurring tasks, whether through AI directly or automations that incorporate it. Tasks that used to take an hour or two now take roughly 10 to 15 minutes, freeing up real capacity across the team.
Rather than mandating AI adoption, Jason focused on showing individual team members what was personally in it for them, framed around getting time back for their own lives, not just agency efficiency. He’s candid that SEO work can be tedious, and the pitch was straightforward: the annoying parts of the job can be cut down significantly, freeing people up for more interesting work like professional growth. The broader mindset he tries to instill: constantly ask whether a recurring task can be automated or sped up, even if the honest answer is sometimes no.
Actionable advice:
Jason built a custom GPT trained specifically on his own leadership style and various frameworks, explicitly instructed not to be a “yes-man” and to tell him directly when he’s wrong. He uses it to reframe situations, particularly when navigating client or team issues, stripping out his own biases and emotions from the equation. The tool is continuously refined, every time he catches it agreeing with him when it shouldn’t, he adds new instructions to correct that pattern.
“It’s meant to be a counter to where I’m thinking, and let me know if I’m doing things right.”
Actionable advice:
Beyond the leadership GPT, Jason uses deep research tools, both Gemini’s and ChatGPT’s, to help identify leads and prospective businesses for outbound campaigns. He notes Gemini’s deep research feature specifically performed better than he expected.
This section is Josh’s own observation, not Jason’s or Break the Web’s. Josh described living in a small coastal town in Portugal where the average salary is roughly $800 a month, cash is still the norm, and essentially nobody is aware of AI at all, a reality check he uses to caution against assuming the pace of AI-driven change feels the same everywhere it’s actually being discussed online.
Jason says Break the Web’s overall SEO strategy has stayed relatively consistent, the main shift has been moving search content marketing slightly further down the funnel, since AI Overviews are absorbing clicks from super top-of-funnel informational searches. He also notes that many of Break the Web’s DTC and B2C clients’ actual customers are everyday consumers, largely unaware of AI or Gemini outside of seeing them in TV commercials, a contrast to the “technology bubble” agencies themselves operate inside. Until there’s a broader shift in consumer behavior, Jason says the agency is staying focused on measurable, trackable work rather than chasing every new acronym.
Jason’s clearest, most repeated position across podcast appearances: nobody has a proven case study showing that a specific action reliably produces a specific AEO/GEO result. He points to a concrete reason why, ChatGPT’s internal knowledge base hasn’t been updated since April 2024, meaning a meaningful share of what it surfaces comes from live, real-time search, which is more expensive to run than answering from its own training data. Being visible in Google, he argues, whether through your own site, digital PR, or brand mentions, remains the most reliable way to get surfaced by these tools.
“We’re marketers and we’re supposed to be data driven. Let data drive us, when nobody really has data.”
Jason notes Break the Web’s own site is already getting real traffic from both ChatGPT and Perplexity, an early signal worth watching even without a full causal framework yet.
Jason’s been in SEO since 2010, long enough to have lived through the first Panda update and the “SEO is dead” panic that followed it, plus the voice-search scare that never materialized. His approach with clients worried about AI is to get ahead of the conversation proactively, offering to jump on a call with leadership to align on where both sides stand, rather than waiting for the fear to build unaddressed. He’s clear that no client at this point seems genuinely afraid of losing their own job over it, the more common framing he hears is that the best marketers will simply be the ones who use AI most effectively, similar to how the strongest freelance writers have shifted into high-impact editing roles rather than being replaced outright.
Break the Web has operated fully remote from the very beginning. Jason describes actively resisting the “social vanity” pull toward having a prestigious office in a major metro area like New York, a trap he says he could have easily fallen into. His structure relies on reasonable time zone overlap across a globally distributed team and genuine trust that people will follow through on what they say they’ll do, rather than in-person oversight.
This section is Josh’s own story about Search for Hire, not Jason’s or Break the Web’s. Josh described nearly renting office space after a team member got excited about a touchscreen unlimited coffee machine in the lobby, despite the space itself having no windows, a reminder of how easy it is to get pulled toward office “amenities” for the wrong reasons.
Jason’s standing hiring advice, one he admits he still has to consciously work at, is “hire fast, fire faster,” specifically around cultural alignment rather than any individual’s skill level. Sometimes a mutual fit issue only becomes clear after someone’s already in the role. His second piece of advice is to avoid impulsive hiring, particularly when someone leaves unexpectedly and there’s pressure to fill the gap immediately, rather than properly qualifying and vetting for both current fit and alignment with future growth plans.
Actionable advice:
Jason ties his most expensive lesson directly back to hiring and talent. As a self-described introverted, technical SEO rather than a natural business leader or communicator, running the actual business side hasn’t come naturally, and he’s taken routes that seemed beneficial for himself, clients, or the team that turned out badly. The recurring theme, in his words, comes down to getting better at finding people who genuinely want to work with the agency in a mutual way, not just filling a seat.
Jason is direct that he doesn’t see personal brand activity as a significant direct lead source. Its real value shows up when a prospect is doing due diligence, checking LinkedIn, finding videos and podcast appearances, and using that as validation that Break the Web is a credible agency to continue working with. He acknowledges the hardest part is simply starting, since putting yourself out there feels vulnerable, and uses AI voice-to-text specifically to solve that: he records unstructured rants and lets AI organize them into coherent posts, since he considers himself a technical operator rather than a natural creative writer. He also frames personal brand as valuable independent of the agency itself, useful for employees job-hunting, and for founders who don’t want their own professional identity fully dependent on the business surviving or being sold.
Actionable advice:
Jason references a LinkedIn post from Klientboost’s Jonathan Dane describing a gamified system for getting the team posting, points for shares and posts, a leaderboard, and a monthly bonus tied to results like inbound DMs or leads. Jason finds it genuinely clever, but is candid about the scale difference: even if only a quarter of a large team buys in, that’s still meaningful volume for Dane, whereas a smaller agency like Break the Web is still figuring out how to get team members simply sharing their own thoughts at all, without it needing to promote the agency or SEO directly.
Both Jason and Josh land on the same underlying point here, personal posting works best when it’s not mandated or explicitly promotional. Jason’s framing: it doesn’t need to be about marketing or SEO at all, “could be a picture of your dog,” and people tend to overthink individual posts trying to make them perfect rather than simply starting.
This section is Josh’s own reflection on Search for Hire’s team, not Jason’s or Break the Web’s. Josh described actively encouraging his own team to post, contrasting it with competitors whose teams remain anonymous and reliant on cold calling, while Search for Hire’s leads come in largely inbound. He gave two specific examples from his own team, encouraging his marketing specialist Shreya to share her Comic Con appearance in South Africa, and noting that a well-known industry name (former Jellyfish SEO director Eric Hoover) engaged with that post organically. He also described advising a friend starting her own agency to publish a much larger volume of LinkedIn posts (roughly 100) before judging what’s working, rather than overthinking the first 20.
Jason expects the next two to three years of SEO to look relatively similar to today, without major new innovations, since the underlying fundamentals have stayed largely consistent for at least seven to ten years even as platforms and search result formats shift. He’s more genuinely curious about the ten-year horizon, once younger, AI-native generations move further into professional careers and their behavioral data starts to show up more clearly.
Asked what advice he’d give agency owners currently struggling, Jason doesn’t hesitate: delegate. He pushes back directly on the instinct to believe you’re the only one who can do SEO, finance, or outreach well, arguing that wearing too many hats at once means, in his words, “you’re kind of halfassing just about everything.” He frames letting go of control as a genuine skill to build, not just a mindset shift.
“You can’t do everything the best… delegate would be my first recommendation. And that’s an art too.”
Mainly for workflow productivity, cutting recurring tasks that used to take an hour or two down to 10–15 minutes, plus a custom GPT trained on his own leadership style specifically instructed to challenge his thinking rather than agree with him, and deep research tools for outbound lead prospecting.
Jason deliberately avoided the pull toward a prestigious physical office, building the agency instead around reasonable time zone overlap across a globally distributed team and genuine trust rather than in-person oversight.
According to Jason Berkowitz, no, nobody currently has a proven case study. He points out that ChatGPT’s internal knowledge base hasn’t updated since April 2024, meaning much of what surfaces relies on real-time Google visibility rather than any dedicated “AEO” tactic.
Not fundamentally, content has shifted slightly further down the funnel since AI Overviews are absorbing top-of-funnel informational clicks, but the core strategy remains largely the same, with the site already seeing real traffic from ChatGPT and Perplexity.
“Hire fast, fire faster” for cultural alignment specifically, paired with a warning against impulsive hiring under pressure when someone leaves unexpectedly, since that’s where most cultural-fit mistakes happen.
Jason says not directly, he uses his personal brand primarily as a validation tool prospects encounter while doing due diligence (checking LinkedIn, videos, podcast appearances), rather than as a direct source of new leads.
Delegate. He argues that trying to personally handle SEO, finance, and outreach at once means under-delivering across all of them, and that letting go of control is a skill worth deliberately developing.
With 95% of roles filled within 25 days, our approach is built for pace and backed by a 1 year guarantee. Get access to the systems and frameworks that power elite SFH teams.
