> For the complete documentation index, see [llms.txt](https://thecontentforge.gitbook.io/thecontentforge-docs/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://thecontentforge.gitbook.io/thecontentforge-docs/feature-guides/experiments.md). # Experiments Experiments is the content testing layer. It ties content variants to scheduled and published posts, pulls real metrics back from the same analytics that power your Dashboard, and turns the result into a saved learning you can fold back into your [Brand Voice](/thecontentforge-docs/feature-guides/brand-voice.md). The page is no longer a standalone form: it's a workflow that connects [Content Forge](/thecontentforge-docs/feature-guides/content-studio.md), [Publishing](/thecontentforge-docs/feature-guides/publishing.md), and Brand Voice into one loop. Open it from **Intelligence > Content Optimization > Experiments**. ### The dashboard Opening Experiments shows four stat cards and a list of recent experiments: * **Active** - experiments with status `active` * **Waiting for results** - active experiments that don't have results yet * **Completed** - closed experiments * **Learnings** - experiments with a saved learning Below the stats, the **Active experiments** usage bar shows how many active experiments you're using against your plan's cap. Only `active` experiments count - drafts, completed, cancelled, and archived experiments are free. When there are no experiments yet, the empty state shows eight starter templates you can click into. ### Starter templates Each template prefills the hypothesis, the success metric, and a starting Variant A / Variant B strategy. You can change anything afterwards. | Template | Tests | | ---------------------------- | ------------------------------------------ | | Hook test | Two different opening lines, same body | | CTA test | Same content, different call-to-action | | Posting time test | Same content, different times of day | | Short vs long format | Concise vs expanded | | Educational vs urgent | Teach-mode vs alert-mode tone | | Question vs statement ending | Same body, different ending | | Competitor-inspired angle | Adapt a high-performing competitor pattern | | Visual style test | Same caption, different image direction | ### Building an experiment The builder opens as a modal with a 5-step progress bar. 1. **Define the hypothesis** - title, hypothesis, optional description. If you didn't pick a template from the dashboard, there's a "Start from a template" expander here. 2. **Choose platform & content type** - X / Instagram / Facebook, and short post / thread / image / video. 3. **Create variants** - up to four variants (A required, B required, C and D optional). Variant A is the control. On B / C / D, the **Generate** button drafts an AI alternate based on Variant A. 4. **Pick a success metric** - one primary metric and any number of secondary metrics. Options include engagement rate, impressions, replies, likes, shares, clicks, saves, video views, watch time, and follower growth. 5. **Save as draft or attach posts** - save as a draft, or activate immediately. If you turn on **Attach to schedule**, you can map scheduled posts on your Calendar to specific variants in one shot. Activating an experiment counts toward your plan's active-experiment cap. If you're at the cap, save as a draft instead, or complete or archive another experiment first. ### Entry points from other surfaces You don't have to start from the Experiments page. * **From** [**Content Forge**](/thecontentforge-docs/feature-guides/content-studio.md) - every result card has a **Create experiment from this draft** button. The builder opens with the draft already filled in as Variant A. * **From** [**Competitor Intelligence**](/thecontentforge-docs/feature-guides/competitor-intelligence.md) - clicking **Create Experiment** on any suggestion opens the builder with title, hypothesis context, and Variant B prefilled from the example. A "Prefilled from competitor analysis" banner sits on top so you remember to fill Variant A (your control) and the success metric. * **From the** [**Publish Panel**](/thecontentforge-docs/feature-guides/publishing.md) - see "Attaching posts" below. ### The experiment detail page Clicking an experiment opens its detail page. * **Header** - title, hypothesis, status chip, platform, success metric, created date, and a status-aware action button (Activate / Mark completed / Archive). * **Action bar** - **Attach posts** opens a modal for attaching existing scheduled or published posts to a variant. **Refresh metrics** re-pulls results from your synced post analytics. * **Variants** - side-by-side cards, one per variant. Each card shows the variant name and strategy, attached posts, and the current value of the primary metric. The winning variant (highest value on the primary metric) is highlighted. * **Results table** - full per-variant per-metric breakdown. Cells with no data show a manual entry input so you can type the number in directly when an external metric isn't synced. * **Learning** - appears once the experiment is completed (or as soon as a learning has been started). See "Capturing learnings" below. ### Attaching posts Experiments work by attaching real posts to variants. There are three ways a post becomes attached: * **From the builder** - Step 5 toggles "Attach to schedule" and lets you assign scheduled posts to A / B / C / D in one step. * **From the detail page** - the **Attach posts** button opens a modal listing draft and scheduled posts (filtered to the experiment's platform) so you can pick what belongs to which variant. * **From the** [**Publish Panel**](/thecontentforge-docs/feature-guides/publishing.md) - the panel has a collapsible **Attach to experiment** section. Pick an active or draft experiment and a variant; the link is created when the post saves. You can attach to drafts, scheduled posts, or post-time publishes. Once a post is attached and published, it is auto-linked to the experiment - you don't have to attach again after publishing. ### Metrics and refresh Metrics come from your synced post analytics (the same source as the Dashboard and Patterns) and are mapped to the right field per platform - engagement rate, replies, likes, shares, clicks, saves, impressions, video views, watch time, and follower growth. A post with sparse analytics still gets a usable engagement rate, so a variant never looks dead just because one number has not synced yet. There are two refresh paths: * **Automatic** - metrics for all active experiments are reconciled and aggregated automatically, roughly once an hour. * **Manual** - **Refresh metrics** on the detail page does the same on demand. If a metric doesn't sync (a manually logged post, or a metric the platform doesn't expose), use the **Results table** to type the value in by hand. Manual values stay alongside synced values. ### Capturing learnings When an experiment is completed (or partway through, if you want to start drafting), the **Learning** card appears below the results. It captures: * **Winning variant** - defaulted to whatever variant is currently winning on the primary metric. You can override or pick "No clear winner". * **Summary** - one-sentence takeaway. * **What changed** - what specifically drove the lift (or lack of it). * **Recommended future use** - when and how to apply this next time. * **Add this recommendation to Brand Voice signature moves** - checkbox. When ticked, saving appends the recommendation to your Brand Voice signature moves so future Content Forge generations pick it up automatically. Applying to Brand Voice requires Admin or Owner role; the flag is saved either way. The **Suggest** button at the top of the card drafts all four fields from the experiment's variants and results with AI. It only fills empty fields - anything you've already written is left alone. ### Plan limits Each plan tier has a cap on active experiments. Only active experiments count against the cap - draft, completed, cancelled, and archived experiments don't. When you're at the cap: * The **New experiment** button on the dashboard is disabled with a tooltip * Trying to activate a draft surfaces `Active experiments cap reached for your plan. Save as draft, or complete/cancel another experiment.` To free up a slot, mark another experiment **Completed** or **Archive** it from its detail page. To raise the cap, upgrade your plan on [Pricing](/thecontentforge-docs/reference/pricing.md) or contact support. ### Tips * Test one variable at a time. If you change format AND topic AND time-of-day, you've learned nothing. * Use a starter template for the first experiment of a kind, then duplicate the shape for follow-ups. * When in doubt about a metric, leave it on engagement rate - it normalises across reach changes. * When you record a learning, tick **Add to Brand Voice** if the lesson is generalisable. That's how the feedback loop closes. --- # Agent Instructions This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com. ## Querying This Documentation If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question. Perform an HTTP GET request on the current page URL with the `ask` query parameter, and the optional `goal` query parameter: ``` GET https://thecontentforge.gitbook.io/thecontentforge-docs/feature-guides/experiments.md?ask=&goal= ``` `ask` is the immediate question: it should be specific, self-contained, and written in natural language. `goal` is optional and describes the broader end goal you are ultimately trying to accomplish on behalf of the user. GitBook uses it to tailor the answer towards what is most useful for that goal. The response will contain a direct answer to the question and relevant excerpts and sources from the documentation. Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.