I locked the visual direction, then unlocked it.
Influence is a social network where founders and builders can meet creators who already reach their audience. I settled the product direction early. A week later, I reopened the visual direction and changed it completely.
- Role
- Solo: product, design, build
- Stack
- Next.js, Supabase
- Decisions logged
- 30
- Artifacts
- 24
I locked the visual direction, then unlocked it.
Influence is a social network where founders and builders can meet creators who already reach their audience. I settled the product direction early. A week later, I reopened the visual direction and changed it completely.
A meeting place, not a marketplace.
The wedge is genuinely open: every competitor is transactional, a campaign tool. None is a social graph where founders and micro-creators hold living identities and meet on purpose. That distinction was locked as a decision on day one, because it determines everything downstream. A marketplace optimises for completed transactions. A meeting place optimises for conversations that start.
Solo: product, design, build
Solo: product, design, build
Four navigation items, argued against a dashboard.
The v0 interface proposal was built as a working comparison rather than a static mock, so the discovery-led structure could be tested against the monolithic dashboard it replaced. Switch between nav states in the artifact below to see the argument.
What the work cannot claim yet.
No users, so the positioning is an argument rather than a validated bet. The verification approach is designed and specified but its abuse resistance is untested against anyone motivated to game it. The visual reversal cost a week, and the record shows the decision, not evidence that the second direction performs better than the first.
A meeting place, not a marketplace.
The wedge is genuinely open: every competitor is transactional, a campaign tool. None is a social graph where founders and micro-creators hold living identities and meet on purpose. That distinction was locked as a decision on day one, because it determines everything downstream. A marketplace optimises for completed transactions. A meeting place optimises for conversations that start.
- 01Discovery-led, not feed-led. A cold-start network has nothing to feed anyone.
- 02Role is self-chosen and can be both, because founders create and creators build.
- 03Presence is available or busy, not "open to" a list of transaction types.
No shadow contacts, even though the metric would like them
The activation metric drives the build, and it still does not get this. Growth would be easier with placeholder profiles for people who never signed up. It was rejected: a meeting place made of people who are not there is a directory wearing a costume, and the first real conversation would expose it. The activation metric was defined to drive the build without licensing that shortcut.
Verification without a paid aggregator
A bio token, automated and free. Creator verification usually means paying an aggregator or checking by hand. For v0 both were rejected: manual does not scale past the founder, and paid is a fixed cost before there is a product. The chosen route asks the creator to place a token in their own bio, which the system can verify automatically at no cost. Auth landed on Google plus email magic link, explicitly not social login on the platform being verified.
Four navigation items, argued against a dashboard.
The v0 interface proposal was built as a working comparison rather than a static mock, so the discovery-led structure could be tested against the monolithic dashboard it replaced. Switch between nav states in the artifact below to see the argument.
Dark was locked. Then it was wrong.
The visual direction was locked to a dark treatment on 31 May. On 5 June it was reopened and relocked to an airy light direction, superseding the earlier lock outright. Reopening a locked decision is normally a discipline failure, so the entry records it as a supersession with the reason rather than quietly editing the original. A meeting place should not feel like a terminal.
Audit the interface, then close every finding
A remediation map, and a closure ledger against it. Rather than an audit that produces a document nobody actions, the findings were turned into a remediation map and then a closure ledger tracking which were actually fixed. The audit is interactive: filter by verdict to see what drove the redesign.
What the work cannot claim yet.
No users, so the positioning is an argument rather than a validated bet. The verification approach is designed and specified but its abuse resistance is untested against anyone motivated to game it. The visual reversal cost a week, and the record shows the decision, not evidence that the second direction performs better than the first.
Reopening a locked decision, on the record.
The decision log preserves the reversal and its reason. I did not rewrite the earlier entry to make the process look cleaner than it was.
Pre-launch. 24 artifacts, no users yet.