DEFINE THE CHANGE, THE GOAL AND THE DECISION BEFORE THE TEST
Experiments & Recommendations
Experiments & Recommendations covers Visioix title experiments, which test title alternatives against one on-site click goal over a fixed 7–90 day horizon on a WordPress site, and the Recommendations block that shows a general list or, with visitor permission, browser-local category interests. Results, publication and visitor choice remain reviewed steps.
Title Experiments
Test one title change against a defined on-site goal.
Configure title alternatives, one supported click goal and a fixed 7–90 day horizon. Review the target and publication policy before starting the experiment.
The Growth title workflow measures the stated on-site interaction. It does not attribute variant-level organic Search Console CTR to the title alternatives.
The product's separate visitor/website A/B workspaces have their own configured scope. Keep their settings and results separate from the current Growth title test rather than assuming every experiment measures the same goal.
Pro Visitor A/B Testing. Tests a published form's question, headline, CTA, color, layout, trigger or form step against open, completion, conversion or contact rate, using only Visioix visitor events. Each experiment starts as a draft contract with a traffic ceiling and schedule; a verified winner can be applied as a new form version.
Pro Website A/B Testing. Tests page title, meta title or description, CTA, landing content, page layout, product presentation, content variant, internal links or mobile experience against an immutable snapshot of a published page. Results combine attributed Visioix measurements with Site Kit context and statistical confidence, and are kept apart from visitor form experiments.
Experiment Results
Let insufficient evidence remain insufficient.
Review the collected sample, duration and statistical checks before interpreting a difference between variants. An inconclusive result remains a result to report honestly.
Do not choose a winner merely because one variant has a slightly larger current total. Examine the decision conditions and the actual goal being measured.
Current source context, content state and cache behaviour matter when assessing whether the experiment ran as intended.
Publication Controls
Keep publication a reviewed part of the experiment.
Optional winner publication requires the configured evidence and current-content checks. The workflow guards against publishing a title over a conflicting later edit.
Use the supported field-specific restoration route when reversing the experiment's title change. Review the resulting editor and public page.
Excluding experiment targets from full-page/CDN caching is part of the configured setup. A cached page must not erase the intended variant or measurement context.
Recommendations
Provide a useful general list, with optional local interests.
Add the Recommendations block or supported shortcode to the page. A general fallback remains available when the visitor has not permitted interest-based recommendations.
With permission, category interests stay locally in the browser for up to 30 days. The frontend does not make an AI request to generate that visitor's list.
Recommendations operate within this content-block workflow. They do not provide full-homepage layout personalization.
Both editions. The "Visioix Recommendations" block and the [arnvisiox_recommendations count="4"] shortcode are included in Free and Pro. By default the block lists the latest published posts.
Visitor Choice
Allow preference handling to be reversed.
Present the relevant permission choice and retain the supported withdrawal route. Withdrawing permission restores the general recommendation list. Nothing is stored until the visitor agrees; with consent, the interests stay in the browser and a 30-day consent cookie is set. Logged-in users see the general list, and Global Privacy Control turns personalization off.
Keep recommendation preferences separate from stored form/contact responses and from permission-controlled anonymous link measurements. These are different data paths.
Test allow, decline, expiry and withdrawal so the visible experience matches the visitor's choice.
A Practical Experiment
State the title change and one goal, configure the horizon and cache exclusions, then review whether sufficient evidence was collected. Choose publication only when the configured guards and editorial review both support it.
Questions About Experiments
How long does a title experiment run?
A title experiment uses a fixed horizon of 7–90 days. The title alternatives, one supported click goal and the horizon are configured before the test starts.
Does a title experiment measure organic search CTR?
No. The Growth title workflow measures the stated on-site interaction; it does not attribute variant-level organic Search Console CTR to the title alternatives.
Does a test always produce a winner?
No. Insufficient evidence does not force a winner or automatic publication.
Are recommendations generated through AI for every visitor?
No. The described frontend block uses the available list and optional browser-local category interests; it makes no frontend AI request.
How long are recommendation interests kept?
With permission, category interests stay locally in the browser for up to 30 days. Withdrawing permission restores the general recommendation list.
Continue the Review
Reviewed against Visioix Pro 2.0.0 and Free 2.1.0 · Last updated 29 October 2026