Design an A/B test for one page: the change to test, the variant copy, and how to measure it.
A real workflow from the Magister template catalog — the same steps a dedicated agent executes for your brand, shown in full before you run anything.
Test page url. If that is still a blank in brackets, pick the brand's page with the most traffic and a clear conversion goal (pricing, signup or main landing page), from its analytics when a source is connected, otherwise from its site, and say which page you chose. Read the page with magister-firecrawl and note the main CTA and its placement, the value proposition above the fold, the trust signals, and 3–5 specific friction points.
Use the hypothesis hypothesis and the primary metric primary metric; where either is still a blank, write the strongest one the friction points support and say you chose it. Define precisely what changes between the control and the variant (which element, the exact change, and why it should move the metric). Test one clear change, not several at once.
Write all new copy the variant needs (headline, subheadline, CTA label, body sections) and the design changes a developer or designer can implement directly, for example 'Move the CTA above the fold'. Use only claims and proof the brand really has: never invent a testimonial, number or logo.
Write the measurement plan: how the primary metric is tracked, 2–3 guardrail metrics to watch for side effects, the minimum detectable effect, the sample size for 95% significance and the test length in days from the page's real traffic (magister_read_analytics_snapshots or magister_read_analytics_breakdown when a source is connected; otherwise say the traffic is unknown and show the formula), and the rule for calling the test, including what to do when it is inconclusive.
Magister workflows draft and prepare autonomously, but anything externally visible — publishing a page, sending an email, changing a campaign — waits for your approval. You can run this template once, or schedule it and review the results as they land.
Start free: enter your website, get a full marketing audit, and run workflows like this one in your own stack.