Employee Directory in Large Organizations: Paging, Caching, Totals and Export Limits

Introduction

Loading time depends on the directory size, network, browser and configuration. In a tenant with thousands or tens of thousands of accounts, the way a web part retrieves people starts to matter: how long the first page takes to appear, how fast search and refiners react, how accurate the totals are and how long an Excel export runs.

The solution offers two directory web parts that follow two different loading models. Employee Directory (Entra ID) loads the whole directory into the browser once and then works in memory. Employee Directory sends a query to the server for every search and loads results one page at a time. This article explains both models, the settings that influence them, and how to configure the directory for a large organization.

The two loading models at a glance

Employee Directory (Entra ID): the whole directory in the browser

First load

When the web part opens without cached data, it reads all users from Microsoft Graph in pages of 900 accounts and follows the paging links until the last page has arrived. Only then does the list appear; until that moment the web part shows a loading indicator. The time of this first load therefore grows with the number of accounts in the tenant.

Two settings change what Microsoft Graph returns:

• Exclude blocked users (in Exclude Employees, on by default) is applied on the Microsoft Graph side: disabled accounts are not downloaded at all, which shortens the load.
• The Manager column: when it is shown in the list or on the card (Show in list or Show on card), each user is requested together with their manager. This makes the Microsoft Graph requests heavier and the cached data larger. When the column is hidden in both places, managers are not requested.

All other exclusions are applied in the browser after the data has arrived: guest accounts are always left out, and Exclude Unlicensed Employees, Exclude Employees with blank Department, Exclude Employees with blank Job Title and the predefined filter narrow the list further. They make the list shorter but do not reduce the amount of data downloaded.

Browser cache and refresh

After a successful load, the web part saves the list in the browser's local storage as compressed JSON, under the key Vitextra.EmployeeDirectory.Data, together with the time of the load. On the next visit:

1. The directory uses the cached user list before any background refresh. Photos, presence and other features may still request additional data.
2. If the cached copy is older than 24 hours, the web part reloads the data from Microsoft Graph in the background. The cached list stays on screen while it loads and is replaced when the new data arrives.

Above the results, Last updated: shows the age of the data (for example, "Last updated: 3 hours ago"), and the Refresh link next to it reloads the whole directory from Microsoft Graph on demand. A manual refresh shows the loading indicator, like a first load.

If a background or manual refresh fails while a previously loaded, nonempty list is available, the web part keeps that list and its original update time and shows the warning Couldn't update the list. The saved copy is shown. Technical error details may follow it. After a failed manual refresh, the loading indicator gives way to the previous list. If no previously loaded, nonempty list is available, the web part shows an error instead. Retry Refresh later and check connectivity and the approved Microsoft Graph permissions if the warning persists.

Keep in mind how the cache behaves:

• Browser storage. The cache is shared within the same browser profile and SharePoint origin, not separated by signed-in Microsoft 365 account. Other profiles and devices have separate storage. Clearing browser data removes the cache.
• One copy for all web parts. All Employee Directory (Entra ID) web parts opened in the same browser on the same SharePoint domain share a single cached copy, whatever page they are on.
• Changes in Entra ID appear after a refresh. Changes returned by Microsoft Graph are reflected after a successful background or manual refresh, subject to the configured filters and exclusions.
• Some settings reload the data, others do not. Changing Exclude Unlicensed Employees, Exclude Employees with blank Department, Exclude Employees with blank Job Title, Exclude blocked users or the Custom Columns reloads the directory from Microsoft Graph. Showing the Manager column does not: the cached copy was loaded without managers.

When the browser storage is full

Local storage has a limited capacity shared with other data from the same origin. If saving the compressed directory fails, the loaded list remains usable in memory. A previous cached copy may remain; if no usable copy exists, the next visit requires a full load.

Typical signs: the directory always takes as long to open as on the very first visit, and Last updated: always shows a moment ago. To reduce the amount of data, keep Exclude blocked users on, hide the Manager column if you don't need it, and configure only the Custom Columns that the page really uses.

What works in memory

Text search, refiners, A–Z navigation, My department, sorting, grouping, totals and export process the loaded list in the browser. Their speed depends on the list size and device. Skills search, personal quick filters, role memberships, photos, presence and details sections can still make network requests.

Display items in pages of: does not change what is downloaded. It only limits how many people are rendered at once; Show More Results... below the list shows the next portion of the already loaded data.

Employee Directory: one page at a time from the server

The Employee Directory web part does not keep the directory in the browser. Every time the page opens, a search is submitted, a refiner is selected, or the My department quick filter is switched on, it sends a new query to the data source selected in Data Source and receives only the first page of results:

• User profiles - the web part runs a SharePoint people search. The page size is passed as the number of rows to return, and every following page starts after the rows already loaded. The values of the refinement panel are returned by SharePoint search with every query.
• Entra ID (Azure AD) - the web part sends a Microsoft Graph query for users with the page size as the number of results; Microsoft Graph returns a link to the next page, which the web part uses for the following portion.

When more results exist, Show More Results... appears below the list. It requests the next page from the server and appends it to the list; while the request runs, Search... is shown instead of the button.

Requesting only one page reduces the initial transfer. Each server search or refiner selection still requires a request, whose latency depends on the service and query. Results reflect the current Microsoft Graph response or SharePoint search index.

The My team, People I work with and Favorites quick filters are an exception in both web parts: they show their own short lists of people, are filtered in the browser and are not paged.

Page size settings

The page size is set by the Display items in pages of: slider in the Data Source group. Its range and default differ between the web parts:

There is no separate option to show all items: in Employee Directory (Entra ID), set the slider to 0. In Employee Directory, results are always paged, and Show More Results... loads the rest.

Totals

With Show Totals on (in Appearance, on by default), the toolbar above the results shows Total number of results:. The number means different things in the two models:

• Employee Directory (Entra ID) counts the people in the filtered list in memory - after the exclusions, the predefined filter, the quick filter, the refiners, the A–Z letters and the search text. The count is exact and covers the whole result, not only the people shown on the current page.
• Employee Directory shows the total reported by the server for the current query: the number of results of the SharePoint search, or the count returned by Microsoft Graph. With the Entra ID (Azure AD) source and Search by skills, people found by their skills are added to that count. The total covers all matching people, not only the pages loaded so far.
• For the My team, People I work with and Favorites quick filters, both web parts count the people in that list.

Settings that affect performance

Export to Excel with large result sets

With the Microsoft Excel spreadsheet toggle on in the Export group, the Export to Excel link appears above the results. In both web parts it creates the file employee-directory.xlsx with a single table named Employees in the Table Style Medium 2 style. The table contains the columns shown in the Details view (Show in list) in their configured order, including custom columns, except Image and Status.

What ends up in the file depends on the loading model:

• Employee Directory (Entra ID) exports the complete filtered list in its current sort order - every person who matches the exclusions, the predefined filter, the quick filter, the refiners, the A–Z letters and the search text, not only the people shown on the current page. The data is already in memory, so the file is created without further requests.
• Employee Directory queries the server again for all results of the current query - the search text, the Predefined Query, the selected refiners and the My department quick filter - in pages of 100, and then creates the file. For a large result set this takes a while, because each page of 100 is a separate request.
• With the My team, People I work with or Favorites quick filter, both web parts export the people in that list.

While pages are collected, the export link is hidden. Both Details and Contact show Exporting to Excel and a record counter, for example “Records 1 ... 300 of 1250”. Keep the page open until the download starts.

Recommendations by organization size

These are general recommendations. Actual loading time depends on the service, network, browser and configured columns. Test with a new browser profile or after clearing the site’s browser data to measure a load without a cache.

Symptoms and fixes

Related articles