#53: Why 90% of Ad Ideas Should Fail and How AI Helps You Find The 10%
Most agency owners I talk to feel the same tension right now.
On one hand, AI has made it easier than ever to generate ads that look incredible. On the other hand, performance still feels unpredictable, CACs are volatile, and scale remains frustratingly hard to control.
The uncomfortable truth?
Ad quality is no longer the moat.
When everyone can produce “Super Bowl level” creative on demand, differentiation shifts away from production and toward strategy, taste, and decision making.
That’s why I brought Brett Friedman back on the podcast. He runs one of the most forward thinking creative strategy agencies I’ve seen, and more importantly, he’s using AI in a way that augments judgment instead of replacing it.
What follows is a breakdown of how AI is actually being used at the cutting edge of agency scaling, without falling into the trap of over automation.
The Core Misunderstanding About AI in Creative Work
Most agencies are using AI backwards.
They’re trying to automate entire workflows end-to-end:
Idea → brief → creative → launch → optimization
All without meaningful human intervention.
The result?
Massive output and mediocre outcomes.
The real leverage comes from understanding one simple principle:
“AI is best at expanding possibilities. Humans are best at choosing directions.”
Brett articulated this clearly: the biggest unlock isn’t speed, it’s perspective.
Large language models don’t just know “marketing.” They understand:
Academic disciplines
Hobbies
Occupations
Psychological frameworks
Cultural metaphors
When you deliberately force AI to think from outside the advertiser’s default worldview, you unlock ideas no brainstorming room would ever surface.
That’s where creative advantage now lives.
Using AI to Generate Angles Humans Would Never Think Of
One of the most powerful (and underused) AI capabilities is role based perspective shifting.
Instead of prompting:
“Act like a performance marketer”
You prompt:
“Analyze this ad account like an exercise physiologist”
“Explain this performance drop like a systems engineer”
“Ideate creative angles like a behavioral economist”
The output still references ad data but through a completely foreign lens.
Why this matters:
It forces new metaphors
It reframes performance problems
It surfaces angles competitors aren’t testing
This isn’t about novelty for novelty’s sake.
It’s about escaping the invisible assumptions every experienced marketer accumulates.
AI helps break pattern lock in.
Why Messaging Still Beats Visuals (Even in an AI World)
There’s a lot of hype around AI generated visuals, and yes it’s deserved.
Today, you can generate:
High fantasy concepts
Impossible studio shots
Rich visual metaphors at near zero marginal cost.
But here’s the counterintuitive truth Brett emphasized:
“Messaging still does the heavy lifting.”
AI didn’t flip that hierarchy.
It removed the bottleneck that used to prevent visuals from matching strong messaging.
Previously:
Great message → compromised visual
Now:Great message → perfectly aligned visual
That alignment increases performance but only if the underlying angle is sound.
When every brand can produce stunning creative, the winner is the brand that:
Chooses the right idea to test
Targets the right audience
Sequences learning correctly over time
Which brings us to strategy.
The Power Law That Governs Every Ad Account
Every serious media buyer eventually learns this lesson:
10–20% of ads drive 80–90% of performance.
That power law changes how scaling should work.
You don’t scale by endlessly iterating on winners.
You scale by manufacturing more chances to find winners.
The strategy looks like this:
Audit everything that has ever run
Identify true outliers (not vanity metrics)
Extract insights but don’t overfit them
Launch net-new concepts aggressively
Accept that most will fail
Repeat faster than competitors
AI accelerates steps 3–5, but it does not replace the judgment required in steps 1, 2, and 6.
That’s where agencies still earn their keep.
Step by Step: Using AI for Net New Creative Exploration
Here’s a simplified version of the workflow Brett outlined.
1. Expand the Idea Universe
Feed AI:
Brand context
Product reality
Differentiators
Historical performance
Competitive landscape
Then ask it to:
Generate angles across value prop frameworks
Map ideas to distinct audience identities
Rank concepts by novelty vs. risk
2. Filter for Strategic Fit
This is human work.
You remove:
Ideas that conflict with brand reality
Angles competitors already dominate
Concepts that don’t map to a scalable audience
3. Test Broadly, Not Safely
Early-stage testing should bias toward:
New messaging
New audiences
New conceptual frames
Exploitation comes later.
4. Let the Power Law Work
Launch volume with intent:
Expect failure
Look for signal, not certainty
Promote winners aggressively
AI makes this process cheaper.
Strategy makes it profitable.
Why Fully Automated “AI Agents” Fall Short
There’s a reason I remain skeptical of fully autonomous AI agents for client strategy.
Creative strategy isn’t just logic it’s:
Taste
Context
Judgment
Risk tolerance
Brand intuition
AI can assist within steps.
It struggles when asked to own the entire process.
The best performing agencies I see use AI like this:
Human → AI → Human → AI → Human
Each step has an owner.
Each decision has accountability.
That structure preserves:
Client trust
Strategic coherence
Cultural alignment
And it avoids the biggest risk of automation: unowned outcomes.
The Underrated Skill That Will Define the Next Decade: Taste
Here’s the hard truth no AI vendor wants to tell you:
“AI does not replace taste. It amplifies it.”
If your taste is bad, AI helps you produce bad ideas faster.
If your taste is good, AI gives you leverage.
Taste is built by:
Studying real performance data
Reviewing actual winners (not Twitter screenshots)
Understanding why something worked
Repeating that process across many brands
This is why agencies remain powerful training grounds.
You see more data, more failures, more patterns faster.
AI shortens the loop, but experience still defines the ceiling.
What This Means for Agency Scaling
The agencies that win the next phase won’t be the ones with:
The most automations
The flashiest tools
The loudest AI claims
They’ll be the ones that combine:
Clear strategy
High creative volume
Strong taste
Disciplined learning loops
AI is no longer optional but it’s also not the hero.
It’s the force multiplier for agencies that already know what they’re doing.
And that’s the real opportunity hiding beneath the hype.
If you want to go deeper, you can run the full version at agencyuplift.co/mini.
Even if you never book a call, the clarity alone is worth it.
