A/B test website copy, free beta

Test four versions
of any section.

Pick a section, write up to four versions, and DraftMine splits visitors evenly between them. Each visitor keeps the version they were shown, and the winning letter is reported into the analytics tool you already run.

✓ No redeploy✓ Version history✓ Your analytics× Not an SEO tactic

The situation

Testing a headline should not need a testing platform.

Conventional A/B tooling wants a tag manager, a visual editor, a developer to wire up goals and often a monthly contract before you can try two headlines. For a small marketing site the setup outweighs the question being asked, so most copy decisions get made by argument rather than evidence.

How it works in practice

What you actually get.

Up to four versions per section

Write the variants in the dashboard. Traffic is split evenly between them, and 'Keep this version' ends the test and promotes the winner to everyone.

Consistent for each visitor

Assignment is sticky for the browsing session via a single non-identifying random bucket, so a visitor sees the same version as they move between pages. No cookie, no visitor ID, no cross-site tracking.

Read in your own analytics

Variants are exposed to your own analytics under the dm_variant dimension, so results live in the tool you already use. DraftMine records nothing about your visitors.

Three steps

Getting there.

1

Open a section in the dashboard and add up to three alternative versions.

2

Start the test. Visitors are split evenly and the shown version is exposed to your analytics.

3

Read the results where your other metrics live, then keep the winner in one click.

Known limits, stated plainly

What this does not do.

This is deliberately lightweight. There is no built-in statistics engine, so significance is yours to judge in your analytics tool, and you need that tool already in place (for example a GA4 custom dimension reading dm_variant). There is no audience targeting or per-visitor personalization: the split is even and anonymous by design. Tests run on public editorial sections only.

Common questions

Questions people ask first.

Does DraftMine tell me which version won?

No. It exposes which version each visitor saw to your own analytics under the dm_variant dimension, and you read the outcome against whatever conversion you already measure. DraftMine never receives your analytics data.

Do visitors see the page flicker between versions?

Visitors receive a pre-generated cached manifest and the swap happens as the page loads. No AI model sits in the request path, and if the manifest is missing or DraftMine is unavailable the original text simply stays.

Is this personalization?

No, and deliberately so. Assignment is an even random split with no visitor identity, no profile and no targeting. DraftMine does not collect anything about who is viewing.

Related

Website copy editor

Add one script tag and every text section on your public pages becomes editable from a dashboard.

Your stack

Install guides for Next.js, Webflow, Framer, Shopify, Astro.

Free during beta

See it on your own pages before installing anything.

Preview your site