The Messy Middle: Product Conviction & AI's Impact
Scott Belsky
Chief Strategy Officer, Adobe (Former CPO)
2025
Framework
The Conviction Question
"If I knew then what I know now, would I do this again?" This is the honest question every struggling founder must ask themselves.
Higher conviction after validation = stay & solve
Lower conviction than day one = quit & pivot
Your conviction today matters more than your initial conviction
In startups, conviction loss is a signal to listen to
Product Sense
The First Mile & Psychology of Users
30
seconds to activation
3x
more time on first mile
∞
customer cohorts
Your users are lazy, vain, and selfish in the first 30 seconds
They want speed, status (look good), and immediate success
No tours, no learning curves, no friction allowed at entry
The first mile never ends—each customer cohort needs a new onboarding
Empathy with context beats assumptions every time
The insightShoulder-to-shoulder customer time reveals the context around your product: their meetings, other tools, their lives. That context is where empathy lives.
Consumer Strategy
Why Consumer Products Actually Fail
Network effects matter: Uber tapped excess capacity. Pinterest unlocked a new consumer insight (collecting interests, not self-portraiture).
Underlying mechanics: Great consumer products have defensible structural advantages, not just clever UX tricks.
Platform dependency: Most new consumer products are expensive R&D labs for platforms that already have distribution, ad networks, and scale.
The lasting test: Does your product create new behavior or just reskin old ones? Durability comes from behavior change.
The Pinterest lesson
Collecting your interests is psychologically different than self-presentation. This insight created network effects (pins driving traffic back to sources, which drove adoption of Pin buttons). Lasting power from deep psychology.
The magic trick
Every great consumer product (Tesla, Spotify) does something users didn't expect. Build for surprise and delight, not meeting expectations.
Playbook
The Golden Gut Framework
Preserve time to explore full surface area of design possibilities
Narrow down to the best 2-3 options via iteration
Get feedback, refine further, then present to team
Use AI as surface-area expander, not idea generator (it gives 5 options; you use none, but it sparks what you'd never want)
Design upstream in strategy, not downstream in execution
The craftGreat product work is about exploring more possibilities faster. AI amplifies this—not by writing your script, but by forcing you to articulate the full context and seeing what surfaces.
Contrarian
Scott's Counterintuitive Takes
✗New consumer products fail because the market's saturatedINSTEAD →✓ They fail because they lack structural defensibility (network effects, behavior change, unique insight into psychology).
✗AI will reduce the need for humans in creative workINSTEAD →✓ AI expands the surface area humans can explore, liberating us from grunt work to do more ingenuity and higher-order thinking.
✗Product leaders should obsess over features and specsINSTEAD →✓ Obsess over empathy and psychology. The first mile, onboarding, and human emotion drive adoption—not feature count.
✗CPOs are independent experts who should own product aloneINSTEAD →✓ Break functional silos. Bring design, engineering, and customer data into strategy. Collapse the stack, not separate it.