Index
Hose Water · design review · step 0 · Monday Kickoff

Before anyone draws anything: the brief, in your words.

Three designs have been rejected on taste, with no agreed target to measure them against — so this session builds the target first. Nothing here asks whether you like something. Every question below is a named elicitation instrument, chosen because it produces something a designer can actually build to.

Prepared by Cairo · run live in the meeting, not async

How to run this, in order. Pick who you are in the black bar at the top — once, and every answer on the page records as you. Then work down: each criterion carries its own verdict, its own note, and its own edit, so nothing is decided in a bundle and nothing waits until the bottom of the page. Where you and Stacy answer the same instrument, the page shows both answers side by side — the disagreements are the useful part, and they are cheaper to find now than in a comp.

Step 1 · The pre-mortem — two minutes, before anything else

Instrument: the pre-mortem (Gary Klein, Harvard Business Review, 2007). Asking "any concerns?" gets a shrug, because objecting to a plan people are invested in has a social cost. Declaring the thing already dead removes that cost. The finding underneath it — imagining an outcome has happened improves people's ability to name its causes by about 30% — is Mitchell, Russo & Pennington (1989). It runs first on purpose: once criteria have been discussed, everyone is anchored to them.

Write yours before reading the other's — independent first, then compare, is part of the method.

Step 2 · The five criteria — edit them until they are yours

Instrument: criteria ratification with client rewriting. A criterion you reworded is one you actually hold; one you merely nodded at is one I will misread later. Edit freely — the original is always one click away.

C1

Recognition

A Gen X visitor sees their own childhood in five seconds.

Their specific one — hose water, Jarts, the streetlight curfew — not a generic "retro theme" any Etsy competitor could wear.

Drafted from the four audits · yours to rewrite

C2

Voice

The dry deadpan is the loudest design element on the page.

The audits agree the writing is the brand's best asset. Nothing cutesy, bouncy, or decorative competes with it.

Drafted from the voice work · yours to rewrite

C3

Worth the card

It reads like a real brand at $23.33.

Trustworthy enough to hand a card number to, and visibly a cut above the copycat field selling the same joke for $7.

Drafted from the market finding · yours to rewrite

C4

The group chat test

At least one moment is worth screenshotting to a sibling.

"Which kid were you?" is built to be sent. If nothing on the page would get forwarded, reach isn't going to fix itself.

Drafted from the growth work · yours to rewrite

C5

Identity — the one call I can't make for the business

Keep one deliberate thread of the existing equity, or break cleanly.

The current site's equity is the navy and the roundel. Every version so far half-kept them, and half-measures are what died. Either the new design carries exactly one thread on purpose — or it starts from zero.

The fork · answer it below

Step 3 · Spend ten points — what actually matters most

Instrument: cumulative voting, also called the $100 test (Leffingwell & Widrig, Managing Software Requirements, 2003). Five criteria all marked important is the same as none of them being important. A fixed budget forces the ranking a designer needs when two criteria pull against each other — and they will.

Step 4 · Pin the feel — six axes

Instrument: the semantic differential (Osgood, Suci & Tannenbaum, The Measurement of Meaning, 1957 — seven points, as in the original); NN/g on its research use. "Make it feel more premium" is unbuildable; a mark at 6 on a Premium–Everyday axis is a spec. Both marks land on the same line, so a two-step gap becomes a conversation now instead of a rejected comp later. Mark where the new site should sit. The fuller form of this instrument also rates today's site on each axis, giving a from→to vector — that doubles the marks, so it is deliberately left out of a first live session.

Step 5 · The words — five it must earn, five it must never

Instrument: desirability testing with product-reaction words (Benedek & Miner, Microsoft, 2002); NN/g's walkthrough. People react fluently but generate language badly under pressure, which is exactly where "I don't like it" comes from — a fixed set removes the generating. The original deck is about 60% positive and 40% negative on purpose, so both halves are here: an all-flattering list cannot tell you what to avoid. Words you and Stacy both pick are the shared target, highlighted as you go. The method's real payload is the explain-each-pick step — the note field underneath.

Now the other half of the deck — the words that would mean we missed.