How can we help?

The most common cause is a stale sc_token. Each redirect generates a unique token stored for 24 hours. If you test by clicking a CTA on a page opened in a previous session, the token in the URL may have expired.

To verify: hit your redirect URL fresh, land on your page, and immediately click the CTA. Check DevTools → Network tab and confirm the POST to /api/pixel/click uses the token currently in your URL.

Also confirm your snippet is installed on the correct page and that your CTA element is captured by the querySelectorAll selector.

No. Privacy mode does not append an sc_token to the redirect URL, so the snippet has nothing to read and will not fire.

In Privacy mode, visitor counts are tracked via aggregate counters. CTA click tracking and conversion pixel tracking are not available in this mode by design.

A 200 response confirms the server received the request but does not confirm the database was updated. The most likely cause is an expired token.

Check that your test is Running and that the sc_token was generated within the last 24 hours by a fresh redirect.

GA4 data has a 24–48 hour processing delay. Events that happened today will typically appear tomorrow.

Also verify your GA4 property is stored correctly — the dashboard requires the numeric Property ID (e.g. properties/272480276), not the Measurement ID (G-XXXXXX). If you connected GA4 before July 2026, reconnect to update the stored format.

Verify in GA4 → Explore that your conversion event appears with the ?v=a or ?v=b parameter in the landing page path.

Go to GA4 → Admin → Property Settings. The Property ID is a plain number (e.g. 272480276) — not the G-XXXXXX Measurement ID. SplitCheck handles this automatically when you connect via the OAuth flow.

Select “Custom event name...” and type it exactly as it appears in GA4 — lowercase with underscores (e.g. file_download, generate_lead).

This can happen if your account was not fully provisioned during signup. Please contact [email protected] with your email address and we will fix it within one business day.

Your test is paused, in draft, or complete. Go to your test detail page and click Start to resume it.

If your test is complete and you want to keep sending traffic to the winner, use the traffic split slider to set 100% to the winning variant — this enables passthrough mode.

First check that your test is in Running status. In Privacy mode, counts update within a few seconds of the redirect firing. In Standard or Session mode, counts update via a database trigger on each visitor insert.

If counts are stuck at zero after confirmed redirects, contact [email protected].

Standard sets a 30-day cookie to remember which variant a visitor was assigned to. Returning visitors within 30 days see the same variant.

Session sets the same cookie but it expires when the browser closes. Returning visitors in a new session may be assigned a different variant.

Privacy mode sets no cookies and stores no personal data. Each visit is assigned randomly. Designed for GDPR, Quebec Law 25, and PIPEDA compliance — no CTA pixel or conversion pixel available.

Yes. SplitCheck works at the redirect level — your ad or email link points to your SplitCheck redirect URL, and visitors are forwarded to your landing page with ?v=a or ?v=b appended.

The CTA tracking snippet is standard JavaScript and works on any platform that allows custom code. Paste it before the closing body tag.

React and Next.js pages require the sendBeacon call directly in your click handler rather than the generic querySelectorAll snippet, since client-side navigation may fire before the beacon completes.

SplitCheck uses a Bayesian statistical engine. The probability shown is the posterior probability that Variant B's true conversion rate exceeds Variant A's, given the data collected so far.

A probability of 87% means that given your current data, there is an 87% chance B is genuinely better. Once this exceeds your decision threshold (default 95%), SplitCheck recommends declaring a winner.

See our full methodology at splitcheck.io/methodology.

Data quality reflects total visitors across both variants: Low = 20–99, Moderate = 100–499, High = 500+.

SplitCheck shows results at lower sample sizes than traditional tools because our Bayesian engine is designed for SMB traffic volumes. We flag data quality honestly rather than hiding early results.

The traditional Z-test requires a minimum of 100 visitors per variant and typically needs several hundred to reach 95% significance. The Bayesian engine is informative from 20 visitors per variant.

At high sample sizes (500+ per variant), both methods tend to agree. At low sample sizes, the Bayesian probability is the more useful signal.

Yes. Go to Settings → Billing → Manage Subscription. Changes take effect immediately and are prorated.

Your existing tests and data are preserved. If you have more active tests than the free plan allows (1 test), you will need to pause or complete the extras before creating new ones.