Make sure you are comparing the same traffic
Install both tags on exactly the same pages and let them run over the same period. Historical data collected before the second tool was installed is not a useful parity test. Start with one completed day, using the same reporting timezone and no optional filters.
Check whether one tag waits for consent, is excluded on certain pages or is disabled for logged-in users. Country exclusions can disagree when services use different geolocation data. Use a period wholly inside one month to avoid introducing our monthly visitor-identifier rotation into your first comparison.
Compare the requests, not just the totals
A script blocked by a browser extension cannot send its pageview. A failed network request cannot become a recorded event. Duplicate installation can do the opposite and count the same navigation twice. Missing code on a second template can hide an entire section of a website.
Use a normal browser and perform a small sequence you can recognise. Open the developer tools’ Network panel to check whether both tags send a request for each action. A successful script download alone does not prove the collection request succeeded.
- Load the home page once.
- Open a second page, then refresh it.
- Use browser Back and Forward, including client-side navigation if your site has it.
- Trigger one deliberately configured custom event.
- Repeat with a second browser and compare the recorded activity.
Separate delivery differences from definition differences
Pageviews are the clearest first comparison because they represent received pageview events. Visitor counts additionally depend on how each tool identifies a visitor. Visits depend on grouping rules. Bounce rate and visit duration depend on the event types included in the calculation.
Our Visitors total counts distinct anonymous identifiers in the selected range. Adding visitor counts across daily bars or individual pages does not reproduce that total, because the same identifier may appear in several rows. A one-page visit also contributes zero measured duration: we cannot measure time after its last pageview.
What to expect when comparing with Umami
Visitor Tracking is built on the open-source Umami engine. Using the same measurement logic removes one source of variation, but does not guarantee matching numbers with Umami Cloud. Infrastructure, delivery, software versions, configuration and geolocation can differ.
For manually tracked events, our JavaScript API is window.visitortrack; Umami’s is window.umami. Trigger each for the same action when comparing both tags. For single-page applications, test all navigation paths: our standard tag handles pushState and replaceState, while bare popstate transitions need a suitable integration.
Keep a small comparison log: the period, filters, actions attempted, requests received and resulting totals. If pageviews disagree, investigate the first missing or duplicated event. If pageviews match but visitors differ, investigate identity and grouping rules next.