The AI Edge

The Briefing

Google Marketing Live dropped today (May 20) — and it's the biggest platform shift of the year. Google announced Gemini-powered dashboards inside Google Ads, meaning you can generate full performance reports using plain-English prompts. They also expanded AI Max to Shopping campaigns, and Dynamic Search Ads will auto-upgrade to AI Max starting September. If you haven't looked at your campaign structure lately, the platform is about to move it for you.

Klaviyo officially partnered with Anthropic to put Claude inside their email platform. Claude can now write email copy, suggest improvements, and personalise messaging based on your customer data — all without leaving Klaviyo. Delivery mechanisms are getting smarter. If your email list is in Klaviyo, this is worth turning on this week.

Snapchat launched AI Sponsored Snaps — brands can now run ads that answer questions inside DMs. A user sees an ad, asks a question, and a brand AI agent responds directly in chat. This is a new delivery layer — not just reach, but real-time conversation at scale. Watch this format. It's early, but the direction is clear.

Gartner released data this week that should reset how urgently you think about AI adoption. Right now, AI handles 16% of marketing work. By 2028, marketing leaders expect that to hit 36% — more than double. The operators who figure this out in the next 12 months will have an enormous structural cost advantage over those who don't.

The coaching industry is projected to hit $5.8 billion in 2026 — and the fastest-growing segment isn't 1:1 coaching. It's cohort-based and subscription-based delivery. The market is rewarding programs that borrow from how Netflix, universities, and Duolingo deliver their products. If you're still running your coaching business the same way you were three years ago, you're competing in a different market than the one that's actually growing.

The Build

This week I didn't build a campaign.

I did something more uncomfortable. I sat down and actually looked at the numbers.

Conversion rates. Ad spend. Funnel performance at every stage. Email open rates, click rates, sequence drop-offs. All of it, laid out in one place. Not to celebrate what was working. To find the honest picture — and use it to decide what happens next.

Then I went somewhere most people avoid: I looked outside my industry.

I spent time researching delivery mechanisms across completely different markets. Not to steal tactics. To ask a harder question: if the smartest operators in other industries were running my coaching business, how would they be delivering it?

Here's what stood out.

Netflix figured out something coaches haven't. The product isn't the content — it's the ecosystem. You don't subscribe to one show. You subscribe to always having something worth watching. The algorithm knows what to surface next before you do. The content is bingeable, structured in seasons, layered in difficulty.

Most coaching programs are the opposite: one big thing, delivered once, finished. Netflix would call that a movie. They're in the series business.

Universities cracked peer accountability decades ago. The cohort model — moving a group of people through a program together, on a shared timeline — exists because isolated learners quit.

When you're in a cohort, you don't just learn from the instructor. You learn from the person next to you who is three weeks ahead. You stay because people are watching. Universities didn't invent cohorts to be innovative. They invented them because they work.

Most coaches still run rolling enrollment, which means every student starts alone.

Duolingo built an addiction engine disguised as education. Streaks. Badges. Leaderboards. A free entry point that creates the habit before asking for money. Micro-commitments — five minutes a day — instead of a 90-minute weekly call that half your students skip.

The genius isn't the language learning. It's the behavioral design. Duolingo has a retention rate that most coaching programs would trade their entire content library for.

Peloton made the instructor the brand. People don't subscribe to Peloton for the bike. They come back for Alex Toussaint. For Robin Arzón. The content is live AND on-demand — so you never miss a session, but there's still urgency to show up live because community is happening in real time. And the leaderboard creates accountability between strangers.

Social proof doesn't just happen in testimonials. It happens daily, inside the product.

MasterClass taught the market that packaging is part of the product. Production quality signals authority before a single word is spoken. When Gordon Ramsay teaches cooking, the lighting, the editing, the sound design — all of it says: this knowledge is worth your time. Most coaches deliver their best thinking on Zoom calls with a bad background and a laggy connection.

The information might be brilliant. The packaging says otherwise.

I took this framework and ran the prompt on a client's business. Not my own — a client. A high-retention coaching program with 90%+ stick rate. On the surface, the retention problem looks solved.

It wasn't. The AI reframed the problem before it said anything else.

"Retention and completion lands differently here. With 90%+ members staying at this high price point and commitment level, the actual revenue leaks aren't where a generic coaching audit would look.

They're three specific places: entry program to flagship conversion sits around 50% — every 10% improvement is $144K–$288K in additional ARR depending on seat volume. Flagship to Top-Level tier conversion is where 70–80% of members exit at the 24-month mark — the biggest leak in the whole ladder. And engagement velocity inside the flagship: even with strong retention, the difference between an energised member and a lukewarm one shows up in referrals, testimonials, and Top-Level uptake, not in the headline number."

Then it went through each framework against those three specific problems.

Netflix: The members platform is a library. Netflix is a system. A library says "find what you need." A system says "here's what you should watch next, based on where you are." There's no autoplay logic, no "you're three months in, here's your next module," no personalised pathways based on which lever the business is pulling right now.

The quick fix: add a "Continue Where You Left Off" and "Recommended For Your Current 90-Day Plan" surface on the dashboard. No new content. Just re-surface what's already there based on each member's onboarding plan and quarterly focus. That single change transforms perception from "course I bought" to "platform that knows me."

The bigger play: re-cut the content library as episodic seasons — Pricing Season, Team Season, Marketing Season, Exit Season — released on a calendar like a TV drop. Members consume in sequence rather than treating the library as a reference shelf they never open.

University cohort model: The entry program is already cohort-based, so credit there. The flagship isn't. It's rolling enrollment — no semesters, no graduating class, no shared "year of."

Members joining today have nothing in common with members at month 18. There's no Class of 2026. The 24-month journey has no visible milestones beyond quarterly events, and no formal graduation moment.

The quick fix: stamp a cohort identity onto everyone joining in the same quarter, even with rolling intake. "Q1 2026 Intake" gets a private channel, a starting-line acknowledgement, a named moment at the next quarterly event. Costs nothing. Creates instant peer identity. And put a real graduation ceremony at the 24-month mark — public to all members.

The Top-Level question stops being "do you want to keep paying" and becomes "do you want to walk the stage." That reframe alone changes the conversion math significantly.

Duolingo: The smallest unit of commitment is roughly one hour per week on the plan, plus fortnightly coaching calls. Between calls, the system goes quiet for 13 days. No daily touch, no nudge, no signal that someone has drifted.

The quick fix isn't daily check-ins — that's the wrong frequency for a professional audience. It's a 90-second weekly check-in: one number, one win, one blocker, logged in the CRM. Triggers a coach response within 48 hours. No new content. Forces the methodology into a weekly habit between fortnights and gives coaches a heat-map of who's disengaging before the 90-day review surfaces it.

The bigger structural play: a live dashboard pulling key metrics automatically from management software. Members see their numbers without logging anything. Less Duolingo-gamified, more Whoop-ring data visibility. Aspirational, not childish.

One important caveat the AI flagged: public revenue leaderboards will repel this audience. They're peers, not competitors. Comparative benchmarking yes — "you're in the top quartile for revenue growth this quarter" — public rankings no.

Peloton: Peloton built instructor cults. Cody Rigsby is a brand. The lead coach here is clearly the primary voice, but the extended coaching team — several coaches with distinct expertise — aren't built as instructor-brands with their own followings, signature methods, or "I always book her sessions" loyalty.

The quick fix: a monthly live hour, all members in one room, lead coach in the chair, one member's business challenge hot-seated live, Q&A, recorded for replay. Different from the fortnightly small-group coaching. Different from quarterly events. Becomes the anchor moment of the month — the "ride together" beat.

The bigger play: build each coach in the extended team into a proper instructor-brand. Each gets a signature content series, a personal methodology, a defined area of authority. Members start saying "I'm more of a [coach name] person." Reduces single-point-of-dependency risk, increases coach loyalty, and gives the Top-Level tier a reason to stay connected beyond the lead coach relationship.

MasterClass: MasterClass charges premium prices because the production communicates "this is the highest level" before a single lesson plays. Cinematic lighting, b-roll, scored music, immaculate interfaces. Most coaching program recordings — even great ones — are functional rather than cinematic. Functional is fine for retention. Aspirational is what makes members brag about the program and what makes the Top-Level hesitant to leave.

The quick fix: re-shoot 5–10 hero modules — the signature frameworks that define the methodology — in proper studio with cinematic production and the lead coach at the centre. These become the showcase pieces. Everything else stays as is. Transforms perception of the entire library.

The bigger play: a full MasterClass-style series per coach, structured as the welcome artifact for new members. "Here's your library, day one." Pairs with any authority-positioning push as the visible anchor for the whole methodology.

Then it ranked the three highest-leverage moves based on the actual numbers.

Number one: build a deliberate entry program to flagship bridge in weeks 11 and 12. This is the biggest visible revenue leak. Lifting conversion from 50% to 70% on 30 seats per year is approximately $430K in additional ARR.

The mechanism: a formal graduation moment at week 11, the lead coach live in the final session personally inviting each member, an unlock of the flagship library at week 11 so members experience the upgrade before deciding, and a seven-day enrolment window that closes at week 12.

Cohort effect — "most of your cohort is continuing" — combined with an instructor moment and a Netflix-style preview unlock. Three frameworks firing at once.

Number two: insert a 24-month graduation ceremony tied to a flagship semester structure. Right now, leaving the flagship at month 24 is a quiet drop-off. Making it a public graduation at a quarterly event — stage walk, named acknowledgement, formal Top-Level invitation — reframes the question from "do I keep paying" to "do I leave the family."

The AI estimated lifting Top-Level conversion from the current 20–30% into the 40–50% range over two to three cycles. Also adds visible progression markers inside the 24 months — Foundations, Systems, Scaling, Legacy — which gives members a sense of forward movement between quarterly events and reduces the motivation dip that typically hits in months 18 to 22.

Number three: restructure the members platform as Netflix, not a library. The members area is the most under-leveraged asset in the program.

Re-organise into "Continue Where You Left Off," recommended next module based on the member's 90-day plan, and 5–10 hero modules re-shot in MasterClass production. No change to the underlying content. Entire change in perception.

Reduces the late-program question of "what am I still paying for" and gives the Top-Level tier a reason to stay subscribed even when coaching contact naturally decreases — because the library itself still feels premium.

The session took about 20 minutes. The output identified more revenue than most campaigns I've run this year.

What worked: The AI correctly identified that this business's problem wasn't retention — it was conversion between tiers, which a generic "improve completion rate" prompt would have missed entirely. The reframe at the start of the session was the most valuable output. Everything else followed from that. The ranked recommendations with actual revenue estimates — not generic advice — made the output immediately actionable rather than theoretical.

What didn't work: The Duolingo section needed a follow-up prompt to get specific. The first pass suggested gamification broadly — streaks, badges — without accounting for the professional audience or the fact that public competition would actively damage community trust in this market.

One follow-up prompt fixed it: "This audience is high-earning professionals in a peer-trust environment, not students. Redesign the Duolingo recommendations with that filter applied." The second pass was far more useful and flagged the leaderboard risk unprompted.

The Play

If you believe one path wins — if you found a model that works and stopped asking whether something better exists — this is for you.

The cost of staying on one path isn't obvious until it's expensive. It shows up in your CPAs creeping up because your offer looks like everyone else's. It shows up in your positioning flattening because the market moved and you didn't. It shows up in your Top-Level conversion sitting at 25% when it could be 45%, not because the program isn't good but because the exit mechanism was never designed.

Innovation isn't a bonus. It's the price of staying in the market at the level you want to operate.

Run this prompt. Fill in your actual numbers — even rough ones work. The more specific you are about your own delivery model and metrics, the more the output stops sounding like advice and starts sounding like a plan.

I run a high-ticket coaching program for [describe your market and offer].

Here is my current delivery model: [describe your format — 1:1 calls, group coaching, course, community, etc., frequency, price point, how clients access content].

Here are my current metrics: [paste your conversion rate, completion rate, refund rate, retention rate, Top-Level or continuity rate — whatever you have. Rough numbers are fine].

I want you to pressure-test my delivery model against five frameworks used by high-retention consumer products:
1. Netflix (content ecosystem, binge structure, personalization)
2. University cohort model (peer accountability, structured progression, timed enrollment)
3. Duolingo (micro-commitments, gamification, habit formation, freemium entry)
4. Peloton (live + on-demand, instructor as brand, community accountability)
5. MasterClass (production as authority signal, aspirational positioning, premium packaging)

Before you start the framework analysis, identify where my actual revenue leaks are — not generic retention problems, but the specific conversion points in my ladder where money is leaving.

For each framework, tell me:
- What this model does that my current delivery doesn't
- One specific thing I could adopt without rebuilding everything
- One thing that would require a bigger structural change

Then give me a ranked list of the 3 highest-leverage changes I could make — with an estimated revenue impact for each where you can calculate it — based on my specific numbers and market.

One important note before you run it: if you have a professional or high-trust audience — people who would be repelled by public competition or gamification — add this line at the end of the prompt: "My audience are [describe them]. Flag any recommendations that would damage trust or community dynamics in this market before I implement them." That follow-up saved the Duolingo section from being actively harmful advice in the client session above.

The operators winning right now aren't the ones with the best content. They're the ones who borrowed the best delivery mechanisms and applied them first.

Try this today. Fifteen minutes with this prompt will surface more revenue than most campaigns. The question it raises will be worth more than the answer it gives.