Nine UX Principles That Actually Move Conversion

UX advice tends to arrive as a list of things that are obviously true and impossible to argue with. Reduce friction. Be consistent. Respect the user. All correct, none actionable.

These nine are specific. Each one is something we've shipped and measured, and two of them are included because they didn't work.

1. Show cost before you ask for anything

The largest single drop-off we've measured in a checkout was at the step where shipping cost first appeared. Fifth step of five, after the address form. People had committed emotionally and then found out the total, and a lot of them left annoyed.

Moving shipping cost onto the cart page, before any personal detail is requested, lifted checkout completion 31% on that store. The same principle applies to services: a price range on the page beats a price revealed on the call.

People don't leave because something is expensive. They leave because they feel a number was withheld.

2. One column in forms, labels above fields

Two-column forms look tidier and read worse. The eye has to decide whether to go right or down at every row, and on mobile the layout collapses into an order nobody designed.

Labels belong above the input. Placeholder text as the only label vanishes when typing starts, which means anyone who gets interrupted mid-form has to clear the field to remember what it wanted. It also fails screen readers outright.

This is the least controversial item on the list and it's still wrong on maybe half the forms we audit.

3. Real availability beats a callback promise

"Request an appointment and we'll call you within two working days" is a form pretending to be a booking. For a dental group we connected the site to their practice management system so patients see genuine slots per clinician. Online bookings went from zero to about 62% of new patients, and reception call volume dropped 34%.

Where an integration genuinely isn't available, be specific instead of vague. "We'll call you before 5pm tomorrow" is a promise. "Shortly" is a shrug.

4. Filter on the attributes people actually decide with

An outdoor retailer we worked with had 3,400 products filterable by price, brand, and size. Their customers choose technical kit on waterproof rating, fill power, and pack weight — none of which existed as data. It was all written into description paragraphs.

Four of the project's ten weeks went into restructuring product attributes. Dull work. Search-to-purchase rate nearly tripled afterwards, and search users went from converting worse than non-search users to converting better.

The lesson generalises: a filter is only useful if it matches the criterion the buyer already has in their head.

5. Compute contrast, don't eyeball it

Light grey text on white looks refined on a calibrated monitor in a dark room. It's unreadable on a phone in daylight, and it fails WCAG.

We check every text-on-background pair numerically before it enters a palette. The threshold is 4.5:1 for body text at normal size, 3:1 for large text. One clinic site we took over had body copy at 2.9:1, which for a patient base skewing older was actively excluding the audience the site existed to serve.

Accessibility gets framed as a compliance cost. It's mostly just legibility, and legibility converts.

6. Design the thumb path, not the cursor path

Most of our clients see 70–85% mobile traffic and most sites are still designed at desktop width first, then checked on a phone at the end.

Practical version: primary actions in the lower two-thirds of the screen where a thumb reaches, tap targets at least 44px, and no important control in the top corners on a large phone. For a restaurant we put menu, hours, call, and directions within one tap from every page — those four things are 90% of what anyone wants at 6:40pm.

7. Reserve space for anything that loads late

Layout shift is the most irritating bug on the web and it's entirely preventable. An image without dimensions, an ad slot, a cookie banner that pushes content down, a font swap that reflows a paragraph — each one moves the thing someone was about to tap.

Set explicit width and height on images, reserve the banner's height, and use font-display: swap with a matched fallback metric. Cumulative Layout Shift under 0.1 is the target and it's achievable on any site if you care.

8. What didn't work: urgency timers

We tested a countdown timer on a promotional landing page — the standard "offer ends in 04:59" pattern. Conversion moved 0.4% against a control, well inside noise, and support tickets asking whether the deal was real went up.

Manufactured scarcity works in specific contexts, mostly where the scarcity is true. Genuine stock levels ("3 left in your size") did help on a retail project. A timer that resets when you reload teaches people not to believe your site.

9. What didn't work: the exit-intent popup

We were asked to add one, said we'd measure it, and ran it for six weeks. It collected email addresses at a decent rate. It also correlated with a measurable drop in return visits, and the list it built converted at roughly a fifth the rate of the site's normal enquiry path.

Net effect after ninety days was negative and we removed it. Popups that interrupt a first visit buy a cheap metric with an expensive one.

The pattern underneath

Seven of these work and two don't, and the split isn't random. The ones that work remove something standing between a person and a decision — a hidden cost, a missing field, an unreadable line of text. The ones that fail try to manufacture a feeling the visitor doesn't have.

That's the whole test. When you're deciding between two design options, ask which one removes an obstacle and which one adds pressure. The first one wins more often than the second, and it keeps winning after the novelty wears off.

If you'd like us to run this list against your own site, our UX audit does exactly that — usually two weeks, with a prioritised list at the end.