How to Setup a Split Test
Creating A/B Page Variants in Humblytics
A/B (or split) testing lets you compare two versions of a page and prove—statistically— which one performs better. Humblytics makes the workflow dead‑simple: pick a goal, drop in two URLs, choose how visitors are assigned, and hit Start Experiment.
Step‑by‑Step Instructions
1. Access the Split Testing Section
Log in to your Humblytics dashboard.
In the left‑hand navigation, click Split Testing.
2. Start a New Experiment
Click Start New Experiment to launch the five‑step wizard.
3. Select a Testing Goal
Choose the metric that will decide the winner. Humblytics groups goals into two buckets:
Increase Form Submissions
Form completion rate
Lead‑gen or sign‑up pages
Increase Click‑through Rate
Link / CTA clicks
Pricing, feature, or blog CTAs
Reach Destination Page
Journey completion
Multi‑step funnels (e.g., "Thank‑you" page)
Increase Session Time
Time on site
Content & SaaS education hubs
Reduce Bounce Rate
Single‑page sessions
Landing pages & blog posts
Tip: pick one primary goal per test to keep analysis clear.
4. Name Your Experiment
Give the test a descriptive title (e.g., “Home Page Hero — Image vs Video”). Good names speed up reporting later.
5. Create & Add Your Variant Pages
Duplicate or build Variant B in your site builder or custom codebase (Webflow, Framer, WordPress, Next.js, plain HTML, etc.). B in your site builder (Webflow, Framer, WordPress, etc.).
Keep the URL slug short and related:
/
,/contact‑us
,/home‑video
, etc.Change one element (headline, hero media, CTA colour) so you can attribute uplift to that variable.
Copy the live URLs for each page.
Back in Humblytics:
Paste the Control Variant URL (usually the existing page).
Paste the Test Variant URL (your new version).
Optional: jot down notes or hypotheses in the text boxes provided.
Control vs Variant Defined • Control (Variant A): The current or default experience that receives 50 % of traffic. • Variant B (or C, D…): The modified page you believe will outperform the control.
6. Choose Visitor Assignment
Allow Visitor Overlap
Assignment happens on every session—a repeat visitor could see different versions on different visits.
High‑traffic sites, short tests, early ideation phases.
No Visitor Overlap
Assignment happens once per user (sticky)—they always see the same version.
Brand‑sensitive flows, long funnels, or when you need behaviour consistency.
7. Apply Optional Restrictions
Geo Restrictions: limit the test to specific countries (e.g., run a EU‑only experiment to account for GDPR banners).
Audience Restrictions: one‑click toggles to exclude Mobile, Desktop, Organic, or Paid traffic.
8. Launch the Experiment
Review your settings. If everything looks good, click Start Experiment. Traffic begins routing instantly—no extra code deployment required.
9. Monitor Results & Declare a Winner
Humblytics updates lift‑percentages and p‑values in real time.
Green check‑marks indicate statistical significance (default 95 % confidence).
Stop the test manually or let Humblytics auto‑end it once significance and minimum‑sample thresholds are met.
Best‑Practice Checklist
Hypothesis‑Driven: write a one‑sentence hypothesis before every test.
One Variable at a Time: avoids ambiguous results.
Run Full Business Cycles: include weekends if your traffic pattern changes.
Segment Post‑Test: slice by device, channel, or geo to uncover hidden wins.
Document Learnings: record what worked and what didn’t for faster future iterations.
Frequently Asked Questions
How do I build variant pages in Webflow, Framer, or custom‑coded sites? • Webflow: Duplicate the page, update content, publish. • Framer: Duplicate the Frame route, edit, publish. • Custom‑coded (React/Next.js, Rails, Laravel, plain HTML): Copy the page template or branch, adjust the element you’re testing, deploy it at a unique path (e.g., /home‑alt).
Keep the Humblytics global script in the <head>
so tracking continues seamlessly.
No extra tracking snippets are needed—Humblytics’ global script captures all variants automatically.
Can I test more than one variant at a time? Yes—just click Add Variant in the URL step. Traffic will be split evenly (e.g., 33 / 33 / 33 %).
What if I need to pause or edit a live test? Navigate to the Split Testing list, click the ••• menu, and choose Pause or Edit. Data collected so far is preserved.
Will the test slow down my site? No. Humblytics routes visitors server‑side after initial request; the 36 kb script is defer‑loaded and never blocks rendering.
Happy testing! Measure, learn, and iterate with confidence using Humblytics.
Last updated
Was this helpful?