Impossible-to-Possible AI Opportunity Scan
Find valuable AI opportunities by testing previously impossible outcomes
- Difficulty
- Moderate
- Time to result
- ~weeks to results
- Steps
- 6
- Confidence
- 96%
This framework starts with an important customer or business problem rather than an available AI feature. The team identifies outcomes it has historically treated as impossible, then evaluates whether recent technical advances have made any of them feasible. Ideas that merely transfer an existing experience into a new medium are deprioritized because they rarely create a durable advantage. The strongest candidate is the one that unlocks a valuable behavior, result, or business model customers could not access before. Because AI capabilities improve quickly, rejected possibilities are revisited on a regular cadence. The output is a focused portfolio of experiments aimed at new value creation rather than a scattered collection of fashionable AI add-ons.
Origin
Extracted from Marketing Against The Grain through a comparison between early iPhone apps and the businesses that used mobile technology to make previously impossible experiences possible.
Core principles
- 01New technology creates the most value when it enables genuinely new outcomes.
- 02Following the first wave of consensus often produces low-value imitation.
- 03Customer problems should lead technology adoption, not the technology itself.
- 04Previously impossible outcomes should be reassessed as capabilities evolve.
How to run it
- 1
Anchor on a real problem
Choose a meaningful customer or operational problem before considering specific AI tools. Define who experiences it and why existing solutions remain inadequate.
Pro tip Prioritize problems with measurable economic or behavioral consequences.
Watch out Starting with a fashionable model encourages technology-first ideas with weak demand.
- 2
List assumed impossibilities
Document outcomes the team has dismissed as technically, economically, or operationally impossible. Include assumptions about speed, personalization, scale, and cost.
Pro tip Ask what customers would request if current constraints disappeared.
Watch out Do not limit the list to improvements on the current workflow.
- 3
Recheck the constraints
Compare each assumed impossibility with current AI capabilities. Determine whether AI removes the core constraint or merely makes one existing task easier.
Pro tip Test capabilities directly because the technology evolves faster than organizational assumptions.
Watch out A faster version of an old process is not necessarily a new opportunity.
- 4
Reject consensus copies
Deprioritize ideas that imitate the most common response to the technology. Look instead for combinations of capabilities that enable a distinct result.
Pro tip Use first-principles reasoning to describe what becomes newly possible for the user.
Watch out Being early does not compensate for offering low-value functionality.
- 5
Prototype the new outcome
Build the smallest experiment that demonstrates the previously impossible result. Measure the customer outcome rather than the amount of AI used.
Pro tip Design the prototype around one decisive proof point.
Watch out Do not mistake an impressive demonstration for repeatable customer value.
- 6
Revisit possibilities regularly
Establish a cadence for reviewing rejected ideas as models, costs, and supporting tools improve. Promote an idea when its binding constraint disappears.
Pro tip Keep a lightweight opportunity register with assumptions and review dates.
Watch out A conclusion reached about AI several months ago may already be obsolete.
In the wild
Rather than merely porting a taxi website to a phone, a company combines mobile internet, GPS, and a marketplace to coordinate riders and drivers in real time. The design begins with an outcome that was previously difficult: summoning and tracking transportation from almost anywhere.
→ The technology enables a new transportation experience instead of a lower-value mobile copy.
A marketing team assumes a 20% click-through rate is unattainable. It uses evolving AI capabilities to explore new targeting, creative, and optimization possibilities, then tests whether the perceived performance barrier still applies.
→ A formerly fixed assumption becomes a testable innovation target.
Common mistakes
Porting the old experience
Teams reproduce an existing website or workflow in a new technology without delivering a meaningfully new customer outcome.
Following the initial herd
Consensus adoption feels safe but frequently concentrates businesses around the same low-value implementation.
Treating impossibility as permanent
Teams fail to revisit old constraints even though AI capabilities and economics are changing rapidly.
Is it for you?
Best for
It is best for product, marketing, and strategy teams exploring opportunities created by rapidly evolving AI capabilities.
Not ideal for
It is not ideal for teams seeking only minor efficiency improvements in a fixed, well-understood workflow.
From the transcript
“And I think you have to use the first principle of what am I unlocking for people that is actually new, and possible that they…”
“Don't follow the herd when it comes to AI.”
“Lesson than two is think about what is impossible for the problems that you're trying to solve. Like what did you think just wasn't possible?…”
From the episode
6 Marketing Opportunities That A.I. Unlocks For Businesses