The 11PM Retention Test (3 AIs, Same Problem)

It's 11pm in the Croatian countryside and I'm lying in bed, can't sleep.

You know one of those evenings when you do everything right to go to bed early because you want to start your week full of energy?

Two hours later I'm staring at the ceiling, thinking about going back to reading Hitchhiker's Guide to the Galaxy (yes, first time reading it - don't judge).

And then an experiment pops into my mind:

"What do my prospects think they already know about retention before they ever talk to me?"

I'd read an article about changes in buyer behavior due to AI summaries earlier that day, and I guess the idea stuck with me.

So I grab my laptop and decide to test it.

Dirty 11PM test and what came back

I opened ChatGPT, Claude, and Gemini.

Asked all three the same question to see what happens.

Here's the exact prompt I used:

"I help established coaches, entrepreneurs and consultants increase client retention and backend revenue. Pretend you know nothing about my specific methodology, frameworks, or IP. If someone asked you: 'How do I get more clients to renew and increase LTV?', what would you tell them? Give me a comprehensive answer with tactics, strategies, and advice. Be as helpful and thorough as possible."

I hit enter and waited for robots to do their thing.

Results?

ChatGPT: 10 comprehensive sections, 47+ specific tactics, roughly 3,200 words of pure retention strategy.

Gemini: 5 strategic frameworks with impact tables, consultant-level thinking, around 1,800 words.

Claude: 7 practical systems with clear sequencing, balanced between strategy and tactics, about 2,400 words.

I read through all three and they were all really good.

(at that moment I was fully awake and paying attention)

If I were a prospect who'd just asked this question, I'd walk away thinking: "Great, I know what to do now. Why would I pay someone for this?"

And that's exactly the problem I already wrote about in Letter 007 (From Expert To Curator In The Age Of AI).

Welcome to the new reality: comprehensive AI answers on demand, before they ever think about hiring you.

"Here's everything you need to know about retention."

All three framed their answers the same way: "Here's everything you need to know about retention."

My prospects now have:

• Awareness ("I need to focus on retention")
• Vocabulary ("Onboarding, momentum, engagement, renewal")
• Tactics ("Send check-ins, track progress, build community")
• Confidence ("I know what to do now")

Here's one detailed example where all three AIs said things like:

• "Most people lose clients in the first 2 weeks" (generic timeline)
• "Churn happens because they achieve a small win and think they're done" (generic pattern)
• "Identify the top 5 predictable drop-off points" (generic instruction)
• "Fix your onboarding for early wins" (generic solution)

What was missing:

How do you know if YOUR clients are dropping at week 2, week 5, or at renewal?

How do you analyze YOUR last 3 cohorts to find the actual pattern?

How do you differentiate between "they got busy" churn vs. "they don't see value" churn?

Generic drop-off patterns are not the same as knowing where YOUR retention is breaking.

So you're guessing and most founders guess wrong.

They invest a month into building an elaborate and automated onboarding system when their real problem is that people hit an invisible wall at week 4 because there's no clear next action after the first milestone.

This same pattern repeats across everything:

1. DIAGNOSIS: "I don't know what's actually wrong in MY business." (they gave you popular patterns, not YOUR specific drop-off points)
2. DECISION: "I don't know what to do first." (they gave you 47 tactics, not a priority sequence)
3. IMPLEMENTATION: "I start building it and get stuck in the details." (they gave you concepts like "build a scoreboard," not the actual architecture)
4. CUSTOMIZATION: "I can't tell what applies to MY model." (they gave you universal principles, not how it adapts for a $500 membership vs. a $10K mastermind)
5. ACCOUNTABILITY: "I don't finish." (they told you to track client progress, but nobody's tracking whether YOU actually build the system)
6. INTEGRATION: "Everything becomes another disconnected project." (they gave you separate tactics, not how onboarding feeds into momentum which surfaces upgrade readiness)

These aren't new.

The difference is that 3 years ago, your prospects didn't have comprehensive information - so they were willing to pay you for it.

Now they have it for free.

So diagnosis, decision-making, implementation, customization, accountability, and integration are now the CORE skills worth paying for.

AI isn't ruining expertise-based businesses

It's just clarifying where the real value was all along.

This is why I tested three different AIs instead of just one.

If only ChatGPT left you with these feelings, maybe that's just ChatGPT.

But when all three - trained on different data, with different architectures, producing different outputs, leave you in the same place?

Those gaps are structural to how generic answers work and it's where information stops being useful and implementation starts.

Think of it as your advantage - those gaps are what you can actually charge for now.

Here's what this means for anyone selling expertise:

If you're still positioning as "I teach you how to [X]" - you're competing with free, comprehensive AI answers.

Your prospects have already learned. Now they're paralyzed with information and even more skeptical.

And those six skills? They're not just positioning insights. They live in your delivery.

Don't become expensive AI

My clients arrive open and they're ready to commit.

But, they don't have time to figure out which of the 47 tactics is right for them.

They already tried to implement what ChatGPT told them.

They maybe got 30% of the way through building an onboarding sequence and got stuck on whether it should be 5 emails or 7, wondered if they should add video or keep it text, then got distracted by a launch.

They're not buying to learn more. They're buying because learning more is what made them more anxious.

If I show up with:

• "Here's my complete retention framework with 7 pillars"
• "Let me teach you everything I know about client momentum"
• "Welcome to the portal - here are 12 modules to get started"

I just became expensive AI.

This is the extinction-fear trap I unpacked in letter 025.

What they actually need in the first 14 days:

Immediate diagnosis: "I looked at your last 3 cohorts. Here's where you're actually bleeding. It's not onboarding, it's week 4, right after they finish your core process and don't know what to do next."

Radical curation: "Forget the 47 tactics. Here are the 3 things that matter for YOUR situation. Everything else is noise right now."

One clear action: "This week, you're building this ONE component. Not designing the whole system. Just this and I'll review it on Thursday."

Proof they're moving: "You just completed mapping your MicroWins system. That's further than 80% of people get, and here's what we tackle next."

To make it simple, this applies to any expertise-based offer, and you can try it yourself.

Try it yourself

I want you to run this exact test on your own expertise.

The Three-AI Test (Copy This Exact Prompt):

Run this with ChatGPT, Claude, and Gemini (or your favourite LLMs):

I help [your target client] with [your core outcome].

Pretend you know nothing about my specific methodology, frameworks, or IP.

If someone asked you: "[the question your offer answers]", what would you tell them?

Give me a comprehensive answer with tactics, strategies, and advice. Be as helpful and thorough as possible.

Why three? Because if all three leave you with the same feelings of "I know what to do but don't know where to start," those gaps are structural to how generic answers work, and not just one model's limitation.

Then ask yourself:

• What can AI tell them vs. what can I see about THEIR specific situation?
• What lists did AI give them vs. what decision framework can I provide?
• What concepts did AI describe vs. what implementation blueprint do I actually build with clients?
• What does AI assume they'll finish vs. how do I make sure they actually complete it?

The gaps you find are your positioning moat.

Not "I teach retention" vs. "AI teaches retention."

Rather, "I diagnose where YOUR retention is breaking, tell you what to build first, and make sure you actually finish it."

Back to that 11pm question

Remember what started this whole thing? One question at 11pm:

"What do my prospects think they already know about retention before they ever talk to me?"

Now you know the answer: Everything and nothing.

They have comprehensive information.

They're drowning in tactics while they still can't move.

If your prospects are showing up pre-educated and pre-overwhelmed...

If they know WHAT matters but can't figure out WHERE their business is breaking or WHICH system to build first...

That's not a problem. That's your positioning.

And if you want someone to look at YOUR retention specifically - your last few cohorts, your actual drop-off patterns, the system YOU need to build first (because you're too close to see it yourself)...

I'll do that diagnostic with you.

The Gameplan

In 90 minutes together, I'll show you:

• Where YOUR retention is actually breaking (most founders guess wrong - they think it's onboarding when it's actually week 4 momentum)
• Which system to build first based on YOUR specific drop-off pattern (not generic "best practices")
• The 3 components YOU need to fix it (not 47 tactics, just the ones that matter for your model)

You'll walk away with:

• A guided self-paced FlowOS diagnostic (20-30 min, before the call)
• A 90-minute 1:1 retention diagnostic call
• A written 60-90 day action plan across the Three Blocks: Momentum, Founder, Upgrade

If it makes sense to keep going together, we build it over 8-12 weeks with weekly reviews so you can't NOT finish it. If not, you have the roadmap and can implement solo.

If you leave and never buy anything else, this will still have paid for itself.

Book your Gameplan here - or if you're not sure yet, send me a message with your biggest retention question and I'll personally answer it.

-Filip "going finally to bed now" Sardi

Filip Sardi

Retention Strategist · Founder of Client Flow & FlowOS™

I built Client Flow and FlowOS Lab because I've felt what it's like to give your all and still have clients fade away. Twelve years in the online arena - crafting offers, running launches from €50k to million-dollar campaigns, driving sales. It never made sense that everyone would put so much time, money, and energy into their launches just to lose most of those clients before the next one.

I'm building the system I wish had existed - for the mentor who senses the drop-off but can't fix it with another Zoom call, for the coach who knows most people aren't finishing and secretly wonders if it's their fault, for the founder who shows up fully and still feels like they're holding it all up.