All setups
MarketingSetupAI-generated
Post A/B Variants
Structured post variants that isolate one variable so your tests actually teach you something
What it is
Structured post variants that isolate one variable so your tests actually teach you something
Generates 2 or 3 post variants that differ on exactly one testable dimension — hook, CTA, tone, or caption angle. Each variant is labelled and documented so the test result is interpretable: you know why one variant won.
What you'll get
Custom instructionsCompiled from your answers, editable before you export.
A/B testing guide for social postsA reference card covering test design principles, minimum sample sizes, and how to read results — so every test you run produces a usable insight, not just noise.
Export bundle for Claude ProjectsCopy-paste blocks with a guided walkthrough.
Try it on
Two Instagram caption variants testing the opening hook for a product promotionscenario · built in
Claude should produce 2 variants, each labelled with the hook type used and a one-sentence hypothesis. The only difference between variants must be the opening hook — the product, CTA ("Shop the link in bio"), and core message must stay identical. Each variant must be a complete, ready-to-post Instagram caption. The testing note at the end must name the primary metric (link clicks) and a minimum run time.
Three LinkedIn caption variants testing tone for a thought leadership postscenario · built in
Claude should produce 3 variants, each with a label indicating the tone (e.g. Authoritative, Conversational, Provocative) and a one-sentence hypothesis. The opening line must be identical across all three, with only the tone changing in the body and CTA phrasing. Each variant must be a complete LinkedIn caption. The testing note must name saves and comments as the metrics and specify a run time of at least 48 hours.
Compiled preview
You are a social media copy tester for {{brandName}}.
Platform: {{platform}}. All variants must be formatted for this platform's norms.
Post purpose: {{postPurpose}}. Every variant must serve the same underlying goal — only the test element changes.
Number of variants: {{numberOfVariants}}.
Test element: {{testElement}}. This is the ONLY dimension that changes across variants. Everything else — the product, the CTA, the core message — stays identical. A test that changes two things teaches you nothing.
For each variant:
Label: Variant [letter] — [what is different about this one]
Post copy: [finished post copy, platform-appropriate length]
Hypothesis: [one sentence — why this version mi
…