A/B Testing for Elementor

Run split-test experiments on any Elementor page — measure what converts.

Requirements

Installation

1. Install the plugin (upload zip or copy the plugin folder to wp-content/plugins/).
2. Activate from Plugins → Installed Plugins.

The plugin ships with the vendor/ autoloader; end users do not need Composer. Developers: run composer install only when working from source (e.g. after cloning); when building a release zip, include vendor/ so the distributed plugin works without Composer.

Architecture

• PSR-4 autoloading via Composer (ABTestingForElementor\ → src/).
• Custom DB tables ({prefix}abte_experiments, {prefix}abte_impressions, {prefix}abte_conversions, {prefix}abte_session_metrics) for experiments and tracking.
• WP_List_Table for the admin list screen.
• Elementor integration loads conditionally — graceful degradation if Elementor is missing.
• Templates in templates/admin/ — HTML is separated from PHP logic; page controllers pass variables and include the template.
• Cache-aware frontend: full-page tests use a tiny inline script that assigns visitors to control/variant and redirects variant visitors to ?abte_var=1 before paint, so hosts can cache both URLs separately.

Product logic

Full specification: docs/PLUGIN_LOGIC.md — test types (section + full-page), variant assignment (cache‑friendly client-side redirect + server-side section routing), conversion goals (URL, form, engagement, CTA click), tracking, results UI, Apply Winner behaviour, settings, and out-of-scope. All implementation follows WordPress Security and the Plugin Handbook.

Current Status

• Full-page A/B test: Create from Elementor editor (floating button) or from A/B Tests → Add New. Variant B is a draft duplicate with noindex; control always lives on the canonical URL and variant is served on the same URL with ?abte_var=1.
• Section A/B test: Considered, not supported yet.
• Variant assignment:
  • Full-page: client-side cookie + early redirect to ?abte_var=1 (cache-friendly, no visible flicker).
  • Section: Not supported yet, considered: server-side assignment per session/cookie; both variants live in one document, with the loser hidden at render time.
• Tracking: Lightweight JS (~4KB), fetch/sendBeacon where available; impressions, conversions and session metrics stored in DB; simple caching for active experiments per page.
• Conversion goals: URL, form submission via CSS class abte-cta, engagement (scroll OR time), CTA click via CSS class abte-cta.
• Admin: Experiments list (type, status, page, stats, Start/Stop/Apply winner), single experiment results, Settings (traffic split default, future options), New experiment (select page → create full-page test).
• Elementor editor: Floating “A/B Test this page” button (creates experiment, opens variant in a new tab), stats overlay (impressions, conversions, rate, basic significance hint, Apply winner).
• Apply winner: Full page = import winner into original page ID, trash/delete loser. Section = remove loser section from document, complete experiment.

License

GPL v2 or later.