Why a page is a set of dials, not a checklist.

The thinking behind the method, in about five minutes. No jargon, four diagrams, and one idea that does most of the work.

Critique gets stuck at two altitudes

Design feedback tends to arrive as either "it doesn't feel trustworthy" or "make the button bigger". The first is true but unusable. The second is usable but unarguable.

What is missing is the middle: a way to get from a strategy you can name to a change you can make, without the leap being a matter of taste. That middle is what this method is. It works by narrowing, and each step narrows for a different reason.

Each step narrows by a different rule. The audit removes themes your page is not betting on, the angle choice removes the ones you are not acting on now, and reading the levers removes the ones already sitting where they should. Forty-eight dials is unusable; four is a morning's work.

A theme is a bet, not a checklist

There are twelve strategies a page can lean on. Trust, Ease, Validation, Momentum and eight more. A page does not need them all, and a page that attempts them all is a page shouting.

So the first question is not "how good is this page" but what is it betting on. Naming the bet turns absences into decisions. A checkout with no Delight has not failed at Delight; it has spent its attention elsewhere, probably correctly.

The useful finding is rarely a theme done badly. It is a theme done faintly, where the machinery already exists and nobody has turned it up.

A lever is a spectrum, not a switch

Inside every theme are three to five levers: dimensions you can actually move. Each runs between two positions that both work, which is what separates this from a checklist.

Neither end is correct. A checklist can only ask whether a thing is present; a spectrum asks where between two workable positions your page sits, which is the question a designer actually faces. The shaded band is where the library stops agreeing with you.

You take two readings on the same axis. Where the page sits, and where it should sit given what the page is for. The distance between them is the finding, and it has a direction and a size, which a single observation never does.

Both ends are real, so there is no score

Fifteen levers carry a counter-principle: a named effect that argues against pushing further. Crank urgency to the ceiling and Reactance says the pressure will be detected and will cost you trust. Max out social proof and the Singularity Effect says one named person would have beaten your big number.

This is also why the tool refuses to rate your page. A score would be the most popular feature and it would break the method, because every user would optimise toward the active pole on all forty-eight dials, which is precisely the failure the counter-principles exist to prevent.

A rating tells you that you are losing. A reading tells you which way to move and what it will cost.

Principles are not owned by levers

The psychology sits underneath the levers, and the same mechanism turns up in many places. This is the part that looks untidy on paper and is correct in practice.

One mechanism, ten levers, six themes. Reactance is not owned by any of them, which is why the layer is a mesh and not a tree. On the three dashed edges it argues the other way: the same effect that makes a locked default annoying is what makes manufactured urgency backfire.

Nothing is forced into a single home. If an effect acts on two dimensions it is filed under both, and the mechanisms that recur most often are the ones worth learning first.

How the library was checked

The levers were written first, from design practice. The psychology was mapped onto them afterwards, and the mapping was then used to audit the levers rather than just decorate them.

The test that validated the library: two levers attracting an identical set of principles are one lever wearing two names. It turns a judgement call into an observation, which is why it can be re-run by anyone revising the library later.

It worked in reverse too. Principles that fitted no lever pointed at dimensions that were missing, which is how anchoring and reciprocity got in. Both are among the strongest effects in the set, and the first draft had nowhere to put either.

Where the psychology comes from

Four sources. Nielsen Norman Group's Psychology for UX study guide, Jon Yablonski's Laws of UX, Cialdini's seven principles of persuasion, and MINDSPACE from the UK Behavioural Insights Team.

Two things are deliberately excluded. Practitioner biases, which act on the researcher rather than the person using the interface, and there are a lot of them in the popular bias card decks. And design heuristics with no mechanism behind them, which sound like psychology and explain nothing.

Things a designer does, like progressive disclosure or autosave, are in the library but as methods rather than principles. A principle explains why moving a lever works. A method is the move. Keeping them in separate layers is what stops the whole thing collapsing into a list of tips.

Every layer is a first pass, including the twelve themes.