Convert guide
Loopful Convert lets you run controlled copy experiments on your live pages — without a redesign, without cookies, and without involving a developer every time you want to test a headline.
It works by targeting confirmed anchors on your page and swapping in variant content at runtime. You define the variants, Loopful serves them, collects results, and lets you publish the winner.
What you can test
Convert currently supports five mutation types:
| Type | What it does |
|---|---|
| Text replacement | Swaps headline, body copy, or label text |
| CTA replacement | Changes button labels and links |
| Section replacement | Swaps an entire page section with an alternative |
| Show / hide | Shows or hides elements based on the experiment |
| Section reorder | Changes the order of page sections |
Start with text and CTA replacements. They are the most reliable, produce results quickly, and are easy to reverse. Section mutations are more powerful but require stable, well-targeted anchors.
Setting up your first experiment
1. Install the loader
Go to Convert > Script in your project and copy the loader snippet. Add it to your site’s <head> or deploy via Google Tag Manager.
Confirm runtime readiness in the dashboard before continuing — Convert will show a green status when the loader is active on the live site.
2. Confirm your anchors
Go to Convert > Pages and review the anchors Loopful detected on each page. An anchor is a stable CSS selector that Convert uses to target a specific element.
Before running any experiment, confirm the anchors you plan to use. Unconfirmed or broken anchors produce unreliable results.
3. Create an experiment
From Convert > Experiments, create a new experiment:
- choose the page and anchor
- write your control (current) and variant copy
- set a traffic split (50/50 is the default)
- define what you’re optimising for (clicks, form completions, scroll depth)
4. Watch Results
Once the experiment is running, check Results daily. Convert uses a cookieless measurement model, so no consent banner is required for the experiment itself.
You’ll see conversion rates per variant, confidence levels, and a recommendation when Loopful has enough data to call a winner.
5. Publish the winner
When you’re satisfied with the result, click Publish winner. Convert makes the winning variant permanent and pauses the experiment. You can archive it or continue with a follow-up test.
What makes a good first experiment
Good first experiments are small and fast to read:
- Hero headline: 2–3 short variants
- Primary CTA label: “Book a call” vs “Get a quote” vs “See pricing”
- Support copy under a form: one sentence vs two vs none
Avoid starting with:
- Multiple simultaneous experiments on the same page (they interfere with each other)
- Section replacement before anchor coverage is confirmed
- Pages with very low traffic (you’ll wait too long for significance)
Safety and control
Convert is designed to feel bounded and reversible at every step:
- Kill switch — pause all experiments on a project instantly from the dashboard
- Anchor confirmation — experiments only run on explicitly confirmed selectors
- Winner publication — changes are only permanent when you explicitly publish
- Runtime verification — check that Convert is serving the expected variants on the live site
Convert and Discover together
Discover and Convert share the same project and the same loader script, so there’s nothing extra to install once both are active.
They solve different problems:
- Discover makes your site legible to machines
- Convert makes your site more persuasive to humans
Run them in parallel. A page that’s well-understood by AI and well-optimised for humans is the full loop Loopful is designed to close.