Skip to content

Methodology

Every default cites a source.

We studied 130,000+ hours of e-commerce UX research, then audited ten reference sites that actually convert (Tartine, Sweetgreen, Levain, Sakara, Trade Coffee, Daily Harvest, and others) so you don't have to. Every choice in the storefront has a citation behind it. If we get it wrong, point at the citation and we'll update the platform.

What we let you customize

Brand surface — the parts that make you you, where the research actually allows freedom.

Three mood presets

Artisan (Tartine / Blue Bottle), Modern (Sweetgreen / Sakara), or Studio (Daniel et Daniel). Same components and flow underneath, different aesthetic vocabulary.

Brand colors

Pick any palette. WCAG AA contrast is enforced, so you can't accidentally make text unreadable.

Typography pair

Curated combinations of heading + body fonts. Each pairing is tested for legibility at all the sizes we use.

Photography

Hero, product photos, about photo. Self-uploaded. We render with optimal compression and modern formats automatically.

Copy on every section

Hero headline, about story, FAQ entries, testimonials, delivery zone descriptions, support contact.

Optional sections

Drops, journal, wholesale page, catering page. Toggle them on or off based on what your business actually offers.

What we keep fixed (and why)

The parts we don't let you change. Each one cites the research that backs the choice. Disagree with one? Tell us and we'll either show you the data or update the platform.

Cart and checkout pattern

Sites that deviate from established checkout patterns lose 20-50% of orders to abandonment. We use the exact field order and layout that 130,000+ hours of testing identified as optimal.

Source: Baymard, Checkout Usability Report

Cutoff banner placement

Delivery clarity is the #1 conversion factor for first-time food orders. The cutoff strip lives directly under the hero so visitors see it within the first 1-2 seconds of landing.

Source: Baymard, Delivery & Shipping UX

Product card layout

Image / name / short description / price / dietary tags is the layout that converts best across every food site we audited. Reordering or removing fields measurably hurts conversion.

Source: Baymard, Product List Usability

Three-step process visualization

Per Miller's Law (7±2 working memory items), three is the maximum number of process steps a first-time visitor reliably remembers. We map cleanly to: order → we make → you receive.

Source: NN/g, Working Memory and Web Usability

Mobile thumb-zone interactive elements

Most users hold phones one-handed. Cart access and checkout buttons are bottom-anchored on mobile so they fall in the natural thumb arc. Tap targets are at least 44px per WCAG 2.2 AA.

Source: NN/g, Thumb Zones for Mobile Touchscreens

Three quotes maximum on testimonials

Three is the sweet spot. Fewer feels thin, more feels desperate. We also drop anonymous testimonials because research shows they have negative trust impact (worse than no testimonials at all).

Source: Baymard, Customer Reviews Section

What we will not do

Things customers sometimes ask for that we deliberately ship without. Each one has a citation explaining why.

Auto-playing hero video

Increases bounce by 23% on mobile and shifts Largest Contentful Paint past 4 seconds, well outside the Web Vitals threshold.

Source: HTTP Archive, 2024 Web Almanac (Media chapter)

Cookie banner that blocks the page

We use a non-blocking bottom strip with reject-all visible at the same level as accept. Per CNIL guidance, anything else is non-compliant under GDPR.

Source: CNIL guidance on cookies

Live chat that opens automatically

Conversion drops 14% when chat opens unprompted. Users find it invasive. Our chat is opt-in only, behind a small floating button.

Source: NN/g, Live Chat Usability

Spinning carousels

Real users almost never click past slide 1. Slides 2 onwards have less than 0.5% click-through rate. The visual real estate is wasted.

Source: NN/g, Auto-Forwarding Carousels

Living document

When the research changes, we change the platform. Reviewers reject any change to the storefront defaults that doesn't also update the methodology document. The full source-of-truth version is public on GitHub — read it, fork it, suggest edits.