Conversion evidence you can inspect

Modern hero with CTA

Source: tests/benchmark-fixtures/01-modern-hero.json, a static internal HTML/CSS fixture with a heading, two text items, an image reference, a linked button and container structure. The checked source checkpoint is 49641d99e16393a405bb97f1882aa970c696bd78; the converter engine is 0.5.0.

The defined fixture retains 5 of 5 measurable items and 1 of 1 structural IDs. It maps 3 dedicated native widgets and 2 generic native Text Editor widgets, with 0 raw-content fallbacks and 1 separate scoped CSS support widget. The source image points at example.com/hero.webp; this measures reference retention, not a real loaded image or a verified visual match.

Eight recorded warning entries cover URL resolution, image media, inherited fonts, residual CSS, layout/theme differences, containers and import verification. Manual preparation includes replacing the placeholder image, resolving the /start link and reviewing fonts, CSS and breakpoints. This fixture is not a verified AI-tool export.

What has and has not been demonstrated

The generated tree exposes heading, image and button widget settings and native containers. A separate earlier acceptance fixture was imported into WordPress 7.1.2 / Elementor 4.3.2 and modified through the Elementor document API, with saved data reloaded successfully. Those API edits are documented technical evidence.

Native heading control editing, media-library replacement, button label/link/style changes, visual layout review and save/reload through the browser editor remain pending. No Elementor screenshot or video is substituted with a converter preview. Mobile visual parity has not been scored, and responsive controls must be reviewed after import.

Read the benchmark correctly

Across the defined 10-fixture internal benchmark, 51 measurable content items were retained and 50 of 51 mapped into native Elementor content: 28 dedicated widgets, 22 generic native widgets and 1 raw-content fallback. The native content mapping rate is 98.04%. All 21 structural IDs were retained. Ten scoped CSS support widgets are measured separately.

Content retention asks whether expected content survived. Native mapping asks which representation it uses. Visual fidelity asks whether the imported page looks right, and is not scored here. These are different measurements. The rate is not universal visual or conversion accuracy.

The aggregate report records 77 warnings and 7 expected review signals that were not reported. These gaps remain visible; inspect unsupported interactive and advanced styling cases even when warnings are absent. Download the full report to see each fixture and the measurement boundaries.