Power Pages: why user counts do not match

The main rule: The current Power Pages model counts unique active users of each website in a calendar month, separately for authenticated and anonymous access. An anonymous unit is not reliably one physical person. Establish the licensing model, website, period and metric before comparing figures. Documentation and conditions were checked on 12 October 2026.

Establish what each number measures

A public Power Pages question combines anonymous licensing, authentication and session persistence. Another discussion of anonymous user counting raises the browser question. “It counts a cookie” is useful, but does not explain the discrepancy between a dashboard and purchased capacity.

Microsoft’s traffic analytics documentation distinguishes visits from licensing consumption. An MAU chart can use a rolling 30-day window rather than a calendar month. Analytics may also include a visitor whose eligible Power Apps rights mean they do not consume licensed capacity.

Before investigating, record Site ID, Environment ID, exact date range and metric name. “July”, “last 30 days” and “month to date” can be three different selections. A Users number without those details is not an ordering quantity.

Authenticated contacts and anonymous cookies

The Power Platform licensing FAQ identifies an authenticated user by the Dataverse Contact record. Repeated sign-ins by that contact to the same site within a month do not create additional monthly users. Anonymous activity is identified by a cookie: another browser, device or deleted cookie can create another unit.

Consequently, “500 partners use the portal” does not by itself estimate anonymous capacity. One person may visit from a work laptop, a personal phone and a second browser profile. Many repeat visits in one browser, however, need not increase that month’s count.

Do not interpret the sign-in exclusion as removing every anonymous visit that person made earlier in the month. The documented boundary is the same UTC day. Include journeys across midnight and different browsers in your pilot, then confirm the actual result in consumption data.

Authentication protects data as well as defining access

For a partner-only portal, do not rely on JavaScript redirecting visitors to sign-in after a page loads. Check the actual page and table permissions. Power Pages security connects web roles to page permissions and table permissions. Identity configuration and data access need their own checks.

Open the site without signing in, using a clean browser. Seeing a sign-in button is not the test. Try direct URLs for the protected form, list and record detail. Describe the result precisely: “An unauthenticated visitor cannot retrieve the contract list,” rather than “authentication is enabled”.

Do not add authentication merely to change the meter. Base the decision on the site’s purpose and data. A public marketing site and a customer portal containing personal information require different designs.

Do not sum a cumulative chart

The capacity management view includes cumulative daily consumption. Each point includes users already counted on earlier days. Summing those points counts the same units repeatedly. Reported data can also lag by up to 24 hours.

Hypothetical example: A licensing chart shows 100 on day one, 130 on day two and 150 on day three. Consumption to day three is 150, not 380. An equivalent incremental export might show 100, 30 and 20 new units. Those rows sum to 150. Before using SUM, establish whether each row is a balance or an increment.

Microsoft’s export documentation defines Billed Quantity as newly counted accesses for that day in the current month. Sum it with the correct filters to obtain monthly consumption. An export can contain several sites and environments; also separate authenticated and anonymous Meter Subcategory values.

An export covering the preceding 30 days is not automatically a full calendar month. Check the first and last dates in the file and obtain another export if necessary. Resolve missing days before explaining discrepancies between reports.

PAYG does not automatically top up a prepaid pack

Microsoft’s PAYG FAQ says that assigned prepaid Power Pages capacity is ignored and not consumed when pay-as-you-go is enabled for the environment. Do not assume “use the pack first, then meter only the excess”. Enabling the billing plan can change the entire charging model.

Before enabling a plan, list the environment and websites it affects. Check whether unused prepaid capacity can be reassigned elsewhere. Compare prepaid and PAYG using the same monthly active users, rather than comparing all registered accounts against visits to a different part of the site. Do not automatically apply website-meter rules to Power Apps per app: an app and a website are different licensing units.

Under the Power Pages PAYG meters, a person active on several websites is counted for each website. The documentation also lists exclusions, including certain authentication pages, static resources and recognised bots. Microsoft recommends /_services/about for availability monitoring; a monitor visiting an ordinary page with a browser user agent may affect consumption. Inspect what your monitor actually sends.

Turn measurements into an order

Hypothetical example, not a price list: After exclusions, one website has 160 authenticated and 780 anonymous capacity units for a closed month. If you choose prepaid packs of 100 authenticated and 500 anonymous users as described in the product plans, that isolated scenario needs two authenticated and two anonymous packs. It is not 940 identical licences. Check capacity already owned and available before ordering.

If your only evidence is 780 physical visitors in another analytics tool, do not calculate the packs yet. You need Power Pages capacity units. For seasonal use, include the monthly pattern and expected peak: a quiet month may not represent a campaign launch.

A month-end reconciliation procedure

1. Fix the scope. Site ID, Environment ID, current or legacy licensing model, and one calendar month.
2. Separate meters. Keep authenticated and anonymous units distinct. Determine whether each report is cumulative or incremental.
3. Verify identities. Contact for authenticated users, cookie for anonymous users. Test additional devices and journeys across a UTC day boundary.
4. Check exclusions. Eligible rights, sign-in pages, monitoring and authentication flow. Do not manually deduct estimated exclusions without evidence.
5. Reconcile the same period. Allow for reporting delay; retain the export, filters and explanation of the analytics difference.
6. Then choose capacity. Compare prepaid packs, allocation and PAYG for the same workload.