Notes — Scott Belsky on the First Mile, Doing Half, and the Messy Middle
Four questions [Adler frame]
Q1 — What is this about? A practitioner’s account of the product design instincts Belsky has refined over two decades — at Behance and inside Adobe at massive scale (Photoshop’s transition from hundreds of dollars to $10/month subscriptions). The conversation covers the opening experience (First Mile), the discipline of subtraction (Do Half), the psychology of persistence (the Messy Middle), and AI’s effect on the experience economy.
Q2 — How is it argued? Primarily through Behance case studies: the iterative killing of features (Tip Exchange, Groups, colour controls) and the observed effect on core-action frequency. Supported by observations from Adobe’s scale. Belsky argues empirically: he learned these things the hard way, not from theory.
Q3 — Is it true? The Do Half argument is well-evidenced within Behance’s case. The claim generalises plausibly — there is substantial product research supporting that fewer features increase core-action completion — but the mechanism (concentration of attention) differs across product types. The First Mile framework is descriptively accurate for most consumer products; the lazy-vain-selfish characterisation is perhaps reductive but practically useful.
Q4 — What of it? The Do Half principle is immediately actionable for any product team with a defined core action. The conviction test is a useful heuristic for founder decisions about persistence. The AI and experience economy take is an interesting hypothesis but not yet empirically grounded.
Glossary
First Mile — The first thirty seconds of a new user’s experience with a product. The design target is the lazy-vain-selfish user, not the sophisticated long-term user.
Do Half — The principle of building only half the planned features, offering half the options, and targeting half the market, in order to concentrate usage on the core action.
Messy Middle — The long, uncertain stretch between starting and finishing a venture, characterised by doubt and the absence of narrative momentum.
Conviction test — A question for founders wondering whether to continue: knowing everything you now know, do you have more or less conviction in the solution than you did at the start?
Brick wall — A failure mode so severe it permanently damages user trust or the business. Must be eliminated rather than tolerated.
The First Mile in practice
New users are not the product’s sophisticated long-tenured users. They arrive with no accumulated understanding, no loyalty, and no patience. Belsky’s characterisation:
- Lazy: wants immediate value; will not invest effort to discover features.
- Vain: wants to look capable and feel successful immediately.
- Selfish: cares only about their own narrow need, not the product’s intended use cases.
The implication: the first-mile experience must remove all friction between arrival and the first moment of value, surface the user’s capabilities before asking them to learn anything, and serve their specific need before expanding to adjacent ones.
The asymmetry: teams spend most of their product-building time serving users who are already past the First Mile. This creates systematic under-investment in the experience that determines whether new users ever reach the depth the team is optimising.
The Behance subtraction experiments
Behance’s core metric was project publications (the action that drove traffic and platform value). The team systematically removed features and observed the effect:
- Killing the Tip Exchange → project publications increased.
- Killing Groups → project publications increased again.
- Removing portfolio colour controls → 24 hours of complaints, then silence; portfolios looked cleaner; publications increased.
Pattern: every feature that diverted user attention away from publishing reduced the frequency of the core action. Removing the diversion concentrated attention.
Generalisation: ‘If you make the whole product about one thing, everyone does that one thing at 10× the velocity.‘
The conviction test
When founders are paralysed by the messy middle question — should I keep going? — Belsky’s method: ask not ‘how tired am I?’ but ‘how much conviction do I have in the solution, knowing everything I know now?’
- More conviction: the founder has evidence that confirms the original thesis even if the execution has been difficult. This is a signal to persist.
- Less conviction: the founder has learned something that undermines the original thesis. This is a signal to stop or pivot.
The test works because it separates the emotional difficulty of the middle (present in all ventures) from the epistemic signal about whether the original bet was right.
AI and the experience economy
Belsky’s take: AI generates the centre — the average, the optimised, the expected. Humans will shift their creative and economic value to the non-scalable edge — live experience, craft, presence, high-touch relationships. The experience economy (things that require a human to be genuinely there, or a hand to have genuinely made it) will be protected from AI commoditisation and will grow in value.