Why analytics visitor and conversion counts differ
On this page
Two reports can show different numbers without either being broken. They may count different identities, use different time windows, or include different events. In SaaS Pro Max, the standard overview's visitor total sums daily unique visitors; active-user reports count distinct users across a rolling window. Cookieless counts use a separate daily estimate.
Use this guide to choose a metric and reconcile a discrepancy before changing your instrumentation.
Daily visitors are not range-wide unique users#
Imagine one person visits on Monday and Friday, while a second person visits only on Friday. Monday has one daily visitor and Friday has two. The overview total is three daily visitors. With stable identities, the distinct count across that period is two people.
This distinction makes daily overview totals additive. DAU, rolling seven-day WAU and rolling thirty-day MAU use raw-event distinct counts instead. Identity resolution and retained history still affect what the system can recognize.
| Question | Metric or report | Interpretation |
|---|---|---|
| How much daily traffic did we serve? | Overview visitors | Sum of daily unique visitors in the selected range |
| How many distinct users were active? | DAU, WAU or MAU | Distinct identities in the report's rolling window |
| How many visits took place? | Sessions | Visits attributed to the day they started |
| How many pages were viewed? | Pageviews | Events named $pageview |
| Who is active right now? | Live visitors | Distinct visitors with an event in the last five minutes |
The analytics reference owns the full definitions. Compare like-for-like ranges and definitions when checking another tool.
Cookieless visitors use a different identity model#
The separate server-side cookieless endpoint does not create person profiles, sessions, payment links or recordings. Its identifier changes at UTC midnight. A person active across midnight may therefore contribute two daily visitor estimates, including within a rolling 24-hour window.
Shared network addresses or browser changes can undercount or overcount. These estimates are useful for aggregate traffic, but they do not establish a lifetime count of individual people or cross-visit retention. The cookieless reference explains collection, aggregation and limits.
A conversion is not always a unique customer#
A goal counts matching events. One event can match more than one goal. The overall conversion total counts a converting event once; individual goal rows count matches for each goal. Adding every goal's total can therefore exceed the overall conversion total.
A funnel follows a different contract: a visitor enters once on the first matching entry event within the selected range, then must satisfy its step and time-window rules. A person triggering a goal repeatedly can generate several goal conversions without becoming several funnel entrants.
Check whether the report counts events, people, sessions or funnel journeys before dividing one number by another. Funnel diagnostics also distinguish visitors whose next-step window is still open from those whose window expired.
Incomplete retention is not zero retention#
Retention measures whether a cohort returns to perform the configured action. A dashed cell means its observation period is in progress; a blank cell means the period has not started. An observed 0% is different from both.
Summary rates use completed periods weighted by cohort size. For an illustrative completed period, one cohort with 10 returns out of 100 and another with 9 out of 10 produce 19 out of 110, about 17.3%. Averaging their percentages would produce 50%, which gives the small cohort too much weight.
An asterisk marks an entry period only partly covered by the selected dates. First-entry mode searches retained history, so it cannot establish an all-time first visit after older events have expired. See retention definitions for entry modes, timezone boundaries and overlapping segments.
Reconcile a difference in this order#
- Match the application and environment.
- Match the dates, reporting timezone, filters and event names.
- Identify the unit being counted: event, session, daily visitor or distinct user.
- Compare identity and consent models; do not combine cookieless estimates with identified users.
- Check observation windows, retained history, late events and identity merges.
- Check ingestion health and quota before interpreting missing events as missing activity.
For a practical investigation using these definitions, read how to investigate a signup conversion drop.