[WordPress] 外掛分享： Nofliq Server-Side A/B Testing

內容簡介

Nofliq Server-Side A/B Testing 是一款針對 WordPress 的伺服器端 A/B 測試外掛，能有效消除頁面閃爍問題。它在伺服器上決定顯示的變體，確保訪客從第一個字節開始就能看到正確的內容，提升測試準確性。

【主要功能】
• 伺服器端 PHP 變體決策，無需頁面渲染
• 消除閃爍與原始內容顯示
• 可獨立運作或與 AB Tasty Flagship 整合
• 支援 Redis 快取以加速重複訪問
• 內部轉換追蹤與報告功能
• 實驗初始為暫停狀態，方便驗證配置

外掛標籤

開發者團隊

原文外掛簡介

Server-side A/B testing for WordPress that eliminates flicker.
Most A/B testing tools decide which variant to show in the browser, using JavaScript that runs after the page loads. This causes the “flash of original content” (FOOC): the visitor briefly sees the original version before the variant replaces it. It looks broken and it skews results.
This plugin takes a different approach. It decides the variant in PHP, on the server, before the HTML is sent to the browser, and exposes the decision to your theme so the correct variant is rendered from the first byte. The visitor only ever sees the assigned variant, with no flash and no layout shift.
Two decision engines
The plugin has two modes, chosen in the settings:

Local (no account needed): the plugin decides variants itself, on your server, with deterministic bucketing. No data leaves your site. Ideal for evaluation or for running simple experiments without any third-party service.
Flagship: the plugin uses AB Tasty Flagship as the decision engine for remote targeting, segmentation and the Flagship dashboard. This mode sends data to AB Tasty’s servers (see External services below) and requires a Flagship account.

Flagship is the default because it adds targeting, segmentation and centralized reporting that a local engine cannot provide. Local mode is fully functional on its own.
How rendering works
This plugin decides and exposes the variant; your theme renders it. The decision is made available to the page as window.abTestData, and your theme (or a small snippet) reads it to render the assigned variant server-side via the abtf_runner() helper. This is the standard server-side pattern used by tools like LaunchDarkly, Optimizely and Split: the plugin decides, your code applies. A minimal integration example is provided in the FAQ.
Key features

Server-side variant decision in PHP, before the page is rendered.
No flicker and no flash of original content.
Works standalone (Local mode) or with AB Tasty Flagship.
Optional Redis caching for fast repeat visits at scale.
Internal conversion tracking and reporting inside the WordPress admin.
Experiments start paused so you can verify configuration before going live.

External services
This plugin can connect to AB Tasty Flagship, a third-party experimentation service, only when the Decision Engine is set to “Flagship” in the plugin settings. In “Local” mode the plugin makes no external requests and no data leaves your site.
When Flagship mode is active, the plugin contacts the following AB Tasty endpoints:
1. Flagship Decision API – https://decision.flagship.io

What it is: AB Tasty’s decision server, which returns the variant a visitor should see for a given campaign.
What is sent and when: when a visitor loads a page with an active experiment, the plugin sends the visitor ID and the campaign (flag) key to obtain the assigned variation. It also posts an “activate” hit to https://decision.flagship.io/v2/activate (visitor ID and the variation identifiers) the first time a visitor is assigned, so the visitor is counted in Flagship’s reporting.

2. Flagship Collect API – https://events.flagship.io

What it is: AB Tasty’s event collection endpoint.
What is sent and when: when a visitor triggers a tracked goal (e.g. a click on a configured element), the plugin sends the visitor ID, the event name, the variant, and the page URL so the conversion is recorded in Flagship.

The visitor ID is either a fingerprint hash (SHA-256 of IP + User-Agent + Accept-Language) or an ID provided by your analytics tool, depending on the configured Visitor ID Provider.
Use of AB Tasty Flagship is subject to AB Tasty’s terms and privacy policy:

Terms of Use: https://www.abtasty.com/terms-of-use/
Privacy Policy: https://www.abtasty.com/privacy-policy/