Login analytics and A/B testing
Read the login page funnel (from page view to connection), check the per-method breakdown and compare two published design versions with an A/B test.
The Login analytics tab on the hotspot detail page shows how many unique login attempts remain at each step, from the guest opening the login page to getting online. In the A/B tests section of the same tab you compare published design versions.
Not visitor analytics
This screen counts the steps of the login page only. Marketing analytics, campaign dashboards and automation are coming soon; they do not exist today.
Read the funnel
- Open the hotspot on the Hotspots page and switch to the Login analytics tab.
- Pick 7, 30 or 90 days with Time range (default 30).
- The boxes at the top show Views, Connected, Conversion and Failed attempts. Conversion = connected / views.
- The Login funnel table lists the steps in order; the step with the biggest drop is where the problem is.
| Step | Counted when |
|---|---|
| Page viewed | The guest opened the login page through the router |
| Method selected | A method (tab) was selected |
| Code sent | An SMS code was sent |
| Identity verified | Identity verified, router login prepared |
| Sent to router | The browser was sent to the router |
| Connected | The router opened the session and the guest reached the landing page |
| Error | Method or login error (wrong code, policy refusal …) |
Each attempt is counted once per step. The Login method breakdown lists every method the guest tried.
Typical readings
- Method selected high, code sent low: SMS provider or credit problem. See SMS not arriving.
- Verified high, connected low: the router login does not complete (RADIUS, walled garden or certificate). See RADIUS not responding.
- Low views: guests never reach the page; check the router's login page setting.
Set up an A/B test
A variant is a design version of the hotspot that was published at least once. You need at least two publishes before you can run an A/B test.
- Publish two different looks in the designer (for example a different colour or heading).
- In A/B tests enter Experiment name, pick 2–4 Version entries and a Weight (1–100) for each. The first variant (A) is the control.
- Save the draft with Create draft, then press Start. Only one experiment runs per hotspot at a time.
- When the planned period is over, read the result with Report and press Stop.
- The same device sees the same variant, even after a new redirect.
- A variant changes the look only (theme, colours, images, texts). Login methods, legal texts, ad slots and device remembering always come from the published version.
- The Significance column uses a two-proportion z-test: p < 0.05 is Significant; with fewer than 100 views per variant the verdict is Not enough data.
Do not stop early
Checking the result every day and stopping as soon as it looks "significant" produces false positives. Decide the duration or number of views up front and run it that long.
Only draft experiments can be edited or deleted; stopped experiments stay for reporting. Start and stop are written to the audit log.
KVKK
Funnel events contain no MAC, IP, phone number or username; each attempt is counted with a random pseudonymous ID generated just for analytics. Events are deleted after 90 days.