` and inherits that section's `padding: 48px 56px 0` as its containing block, which is what makes its width `min(1250px, viewport - 112)`. Here the band is its own Elementor section, so that section has to play the same part — and only that part. It sets padding and nothing else: no background (Elementor's document wrapper already carries A2's `--lz-paper`, exactly as v4's `.v4-page` does under its own transparent `.v4-hero`), and no Elementor Style control, because Elementor emits a set control as `.elementor- .elementor-element-` at (0,3,0) and would outrank every ported rule above. THE SHORTHAND IS LOAD-BEARING. lendzi-theme/style.css:1061-1064 sets `section { padding: 5rem 0; width: 100% }` on every bare

, and Elementor's Section::get_html_tag() returns `section`. With `html { font-size: 62.5% }` that is 50px top and bottom. A `padding-inline` longhand would leave both in place and float the band in a 100px void. css/v4-hero.css never neutralises this rule explicitly — it defeats it incidentally, for the hero only, via A4's own `padding` shorthand — so this is the first place on this ticket where it has to be done on purpose. Elementor and elementor-kit-7 set no section padding of their own (grepped), so there is nothing else here to zero. `box-sizing` is re-declared rather than left to A2's `.lz-v4 *`, which reaches this section only while `.lz-v4` is an ANCESTOR of it. On a page that renders the strip with the hero switched off, the scope class this component adds to the document wrapper is what provides that — but the strip's own shortcode root sits INSIDE the section, so if that filter ever stops applying, `width: 100%` plus 112px of padding computes as a content box and overflows by exactly 112px at every desktop width, silently. One line. ========================================================================== */ .lz-v4 .v4-trust-band { box-sizing: border-box; padding: 0 56px; } @media (max-width: 900px) { .lz-v4 .v4-trust-band { padding: 0 20px; } } /* ========================================================================== B4. THE ELEMENTOR BRIDGE The same problem css/v4-hero.css B5 solves, one level shallower. Elementor emits six boxes between the section and the band: section > .elementor-container > .elementor-column > .elementor-widget-wrap > .elementor-widget > .elementor-widget-container > [data-lz-v4-trust-strip] `display: contents` removes all six, leaving the rendered box tree as `section > .v4-hero__trust` — one parent, which is what v4 has. Every ported rule in Section A then applies with no arithmetic and no offset. The container is included DELIBERATELY, and removal is why this needs no specificity escalation. Two rules would otherwise own that box: elementor-kit-7 caps `.elementor-section-boxed > .elementor-container` at 1250px, 1024px at <= 1024, and 480px at <= 544. That last one would render the mobile band 480px wide inset 12px from the left instead of full-bleed — measured, and invisible without measuring. Elementor frontend.css makes `.elementor-container` a wrapping flex container, so the band becomes a flex item and shrinks to fit: measured 35px wide at 1440px. A box that is not generated has neither a max-width nor a flex-basis, so both fights are won by removal. css/v4-hero.css B5 needed a (0,4,0) compound to outrank the kit's caps because its container has to remain a real box (it IS the hero's grid); this one does not. The section also carries Elementor's `layout: full_width`, which is belt and braces rather than the mechanism: verified in elementor/includes/elements/section.php:246-258 to be a `prefix_class` control with NO `selectors` key, so it renders as the class `elementor-section-full_width` and generates no CSS — it creates no (0,3,0) rule, which is why setting it does not violate the no-Style-controls rule above. It fixes only the 480px cap on its own. Its real value is the editor, where the guard below turns `display: contents` off. Guarded on `elementor-editor-active` for B5's reason: `display: contents` removes the boxes Elementor draws its column handles on. Elementor adds that class to the preview body and removes it in Preview mode, so the frontend and Preview both get the bridge. Test the editor AND its preview toggle. `display: contents` changes boxes only, never selector matching — `>` still walks the element tree — so lendzi-theme's `.entry-content > *:not(...)` 58rem cap still matches `.elementor` and not the band. ========================================================================== */ body:not(.elementor-editor-active) .lz-v4 .v4-trust-band > .elementor-container, body:not(.elementor-editor-active) .lz-v4 .v4-trust-band > .elementor-container > .elementor-column, body:not(.elementor-editor-active) .lz-v4 .v4-trust-band > .elementor-container > .elementor-column > .elementor-widget-wrap, body:not(.elementor-editor-active) .lz-v4 .v4-trust-band > .elementor-container > .elementor-column > .elementor-widget-wrap > .elementor-widget, body:not(.elementor-editor-active) .lz-v4 .v4-trust-band > .elementor-container > .elementor-column > .elementor-widget-wrap > .elementor-widget > .elementor-widget-container, body:not(.elementor-editor-active) .lz-v4 .v4-trust-band [data-lz-v4-trust-strip] { display: contents; } /* ========================================================================== B5. WITHDRAWING css/v4-hero.css B11 B11 gives `.lz-v4 .v4-hero` 48px / 24px / 12px of bottom padding at (0,2,0), for one stated reason: the trust band did not exist, so with nothing after it the card sat flush against the next section's edge. Its docblock asks whoever adds the band to come back here. The band now supplies that space itself, with v4's own `margin: 32px auto 0` collapsing through the band section's zero top padding — measured 32px above 900px and 18px at or below it, matching the reference. Left in place the two would STACK: 80px desktop against the reference's 32px. (0,3,0), one class above B11, so this does not depend on enqueue order. v4's own shorthands already declare 0 here, so this restores the reference rather than diverging from it. The coupling to this file is the mechanism, not an accident: this stylesheet is enqueued only on a page that actually renders the strip, so with the strip switched off B11 stands unopposed and the hero keeps its closing space — no `:has()`, and no test on where the band sits. If the strip is on a page but NOT immediately after the hero, the hero closes flush; accepted, and asserted against by docs/scrum-521-04/overflow-sweep.js's adjacency check. ========================================================================== */ .lz-v4.lz-v4 .v4-hero { padding-bottom: 0; } /* ========================================================================== B6. THE HOST TRANSITIONS EVERY , INCLUDING THIS ONE A measured leak, and the one this component's computed-style diff was worth writing for. Something in the host stack declares: a, path { transition: all 0.15s linear } at (0,0,1), which reaches `.v4-hero__trust-badge-link`. v4 declares no transition on that element, so this is a straight fidelity difference — but `transition: all` also makes it a behavioural one, because `font-size` is an animatable property. HOW IT PRESENTED, because it is worth knowing what this looks like from the outside. The diff reported the link and the Trustpilot image with a font-size of 13.5547px against the reference's 12px, at 900px and 600px only. The value is not reproducible: a fresh load at 900px reads 12px, a resize from 1440 to 900 reads 13.7773px, and a second visit to 900px in the same session reads 12.6653px. It was the 14px -> 12px step at the <= 900px breakpoint, animating over 150ms, sampled two frames in. Not a leak in the port's font sizes — the `.v4-hero__trust-stats` rule reports a clean 12px throughout — and not something a screenshot or a settled measurement would ever show. Scoped to the badge link rather than to `.lz-v4 a`, and kept out of B2, for the same reason B2 names `.v4-pill` and `.lz-btn` individually: B2 is byte-shared with two other stylesheets and must stay that way. `initial` rather than `none`, and the difference is not pedantry. Both are behaviourally inert, but `none` computes `transition-property: none` where a page with no rule at all computes the initial `all` with a 0s duration — measured. Section B exists to restore the reference's computed values, so it has to land on the reference's values and not merely on a working page. With `none` the diff reported 20 residual transition-* differences; with `initial` it reports zero. ========================================================================== */ .lz-v4 .v4-hero__trust-badge-link { transition: initial; }