Driftwood

Prospects were misunderstanding the product, and it was costing the founder clients

People rarely know when they have misunderstood something — they are busy, so they dismiss it and move on. Repositioning the product around feedback from real pitches won the founder three more clients.

PositioningWeb designShipped in code
The shipped driftwood.sh homepage — 'Don't send your leads slop' above a playing product clip and a customer quote

Driftwood is an AI sales agent for demo-led outbound: it researches each prospect, builds them a working custom demo, and sends it from the founder's own account. A genuinely different product — with a website that made it sound like everything else.

The founder had run 20+ sales conversations and landed one client. The problem wasn't the product. It was a single misconception the site never corrected.

Role
Strategy, design & code
Timeline
18 hours total
Type
Shipped client work
Live
driftwood.sh
1 → 4
clients, after the reframe — three closed during message testing, before the new site even shipped
20+ : 1
conversations to clients under the old positioning
18 hrs
total, including meetings, to research, design, code, and ship

The misconception

People thought Driftwood made one demo for everyone

Prospects who saw the old site assumed Driftwood generated one generic demo of your product and blasted it out. That reading made it just another AI outbound tool — easy to dismiss in a market drowning in AI slop.

The reality is the opposite: Driftwood builds a new custom demo, tailored to every single client it contacts. That's the whole value. And it was the one message visitors consistently didn't take away.

The old driftwood.sh hero: a headline about shipping a custom demo beside a dense screenshot of the product dashboard

The hero it replaced. Visitors could not tell what the product actually did — and the dashboard sitting next to the headline was too busy, and too unexplained, to carry a hero.

The old site's 'Don't send out AI slop' section — a generic cold email on the left, the message Driftwood sent on the right

The old AI-slop comparison. The argument was right and the proof was real, but it read as raw screenshots rather than a polished product, and the value took too long to land.

Words before pixels

We tested the story on real prospects before touching the design

Instead of redesigning on instinct, we took new language and framing into in-person conversations with potential clients and watched where understanding clicked — and where it broke.

Once the framing corrected the misconception, the founder landed three more clients — before a single pixel of the new site existed. That was the green light: the message was validated with revenue, and the site's only job was to deliver it.

The pivotal moment

Show the moment the product becomes obvious

Every explanation of Driftwood eventually hits one pivotal beat: the agent has already researched the prospect and built them their own working demo. Text kept burying that moment, so I built it as motion — you watch the demo get made for a specific company, not described in the abstract.

It's the first idea the new homepage gets across, before you've read a paragraph.

The new driftwood.sh homepage, scrolling from the hero through the product story

The shipped homepage. The hero shows the outcome; the pivotal moment plays out on screen instead of being explained.

Architecture

An information architecture that sells

The old page order answered the founder's questions, not the prospect's. I rebuilt the IA around the objections people actually raised in the test conversations, in the order they raised them: what is this → is it slop → prove it's custom → show me the product → let me book.

Each section earns the next scroll. Nothing on the page exists because it 'should be there' — it exists because a real prospect needed it to say yes.

Proof

Case studies you can skim — that still feel special

The degree of customization is Driftwood's proof, so I designed a case study section that shows it: real demos the agent built for real companies, presented so a visitor gets the point in seconds but can go as deep as they want.

Skimmable was the constraint. If a prospect has to read to feel the customization, the section has failed — the visual variety of the demos does the arguing.

Scrolling through the case studies section — each card a different custom demo Driftwood built

Every card is a different demo the agent built. The variety is the argument.

The dashboard

The dashboard section, without the essay

The old 'explore the dashboard' section was word-heavy, with screenshots that had no visual focus — visitors couldn't distill the value quickly, so they skipped it.

I replaced it with an interactive, highlightable dashboard: hover a feature and it lights up with a one-line explanation. Same information, but you explore it instead of reading it. The section went from the most-skipped part of the page to the part that shows the product actually working.

The interactive dashboard section — hovering different regions highlights each feature with a short caption

Hover to highlight. Each region of the real dashboard explains itself in one line.

Reflection

Positioning is a design deliverable

The highest-leverage design work on this project happened before any interface existed: finding the one sentence prospects were getting wrong and fixing it in conversation, with evidence. The website is that sentence, made visual and shipped.

Given 18 hours end-to-end — meetings included — I coded and shipped it myself. It's live at driftwood.sh.

Next project

Spotify moved into mixing before anyone else — listeners were not following