Bulk selection UX: make the count, scope and action agree

Make selection mode visibly different from browsing

Noteshelf 3 shows Select All and 1 Selected over a notebook grid with selected-state markers. Dropbox shows two selected items, Cancel and an action toolbar in a file list. Both change the surrounding controls so selection is not merely an invisible internal state.

Explain the mode through the count, selection markers and a clear exit. Do not depend on a long press as the only discoverable entry. A labeled Select action can support users who do not know the gesture or cannot perform it. Once the mode is active, keep item opening separate from selection so a tap has one predictable effect.

Exit should clear selection according to a consistent rule without changing the underlying content. If Cancel leaves the list, it should not cancel a previously accepted background operation unless that relationship is explicit. Preserve the current filter and scroll position so entering and leaving multi-select does not disrupt ordinary browsing.

Define what Select All includes

Collect shows a selected item within All Items, where boards and items may share the surrounding workspace. WPS Office shows a selected file in a folder with Select All and batch actions. The same label can therefore refer to different kinds of collections and current locations.

Choose whether Select All means loaded items, visible filtered results, the current folder or the entire collection. Name the scope when ambiguity is likely. In a paginated list, selecting every loaded row is not the same as selecting every matching result. Do not display All selected if only the first page is selected.

For filtered results, explain whether changing the filter preserves hidden selections. A useful selected-count summary can provide a route to review the selected items. Keep the batch payload based on stable identifiers or an explicit server-side query snapshot. Re-evaluating a live filter at execution time can affect items the user never saw or intended to select.

Show actions that make sense for the entire set

Layout offers Share, Duplicate, Move and Delete for a selected project. Documents exposes Copy, Move, Compress, Delete and More across a mixed file workspace. Each toolbar is a capability decision: which operations can be applied to the selected objects, not just which icons fit along the bottom.

Calculate available actions from type, permission and selected count. Rename may be a single-item operation; Merge may require compatible multiple items; export can have format limits. If an action cannot apply to every item, explain the affected subset before running it or disable it with a useful reason. Do not silently skip incompatible items and report complete success.

Keep action labels reachable with large text and assistive input. An overflow menu can hold less frequent commands, but destructive actions should remain clearly named. Recompute capabilities when selection changes without shifting the primary action underneath a finger. The selection count and available actions must describe the same snapshot.

Make count-sensitive operations understandable

PDF Scanner includes Select All and Deselect All over a document list, with Delete, Move, Rename and Share. Plaud shows Selected 2 and commands including Generate, Rename, Merge and Delete. The two-item recording case makes count-dependent capability especially relevant.

Show the current count near the mode exit and repeat it in a consequential confirmation where needed. If Merge needs two recordings, explain what will be produced and whether originals remain. If Rename supports one item, it should not imply that the same command renames a whole selected set without further input.

Do not substitute a generic More menu for explaining batch scope. A person should be able to tell whether Generate starts one output per recording or one combined output. Where the operation incurs a cost or creates many jobs, show that before confirmation. Selection is only preparation; each action still needs its own understandable contract.

Distinguish pages, files and media items

Forma shows a selected page in a numbered page grid with Share, Move, Copy, Split and Delete. Space Cleaner shows selected images with Compress, Copy, Move, Share, Convert and Delete. These selections operate at different levels: a page inside a document is not the same object as a complete file or a photo.

Name the unit in confirmations and result messages. Delete one page, delete one document and remove one local image have different consequences. Keep numbering stable enough for users to identify the chosen page after reordering. For media, use thumbnails alongside counts, but do not rely on thumbnails alone to express selection to assistive users.

When an operation creates new outputs, keep their relationship to the selected originals clear. Split can create several files; conversion can create a different format; compression may replace or duplicate material depending on the product. Explain that policy and let users find the outputs. A count of completed operations is not sufficient if the resulting items disappear into an unknown folder.

Execute the batch against a stable selected set

Store selection by object identifier, not current row position. Sorting, filtering and new content should not retarget an action. Freeze the relevant set when the command is accepted, and retain enough object metadata to explain partial results. Remove unavailable items from the set deliberately, with feedback when it affects the count.

Track batch identity and per-item outcomes. A transfer, export or deletion may succeed for some items and fail for others. Show completed and failed counts, identify recoverable failures and offer retry only for the remaining work. Do not repeat successful destructive operations because the final summary request timed out.

For deletion, name the scope and destination policy. A recoverable trash action and permanent removal are different. Offer Undo only if the system can reliably restore the affected objects within the stated period. Keep permissions and ownership constraints in the action calculation, and revalidate them when the server executes the command.

Test selection while the list changes

Select items, change sorting, apply a filter and receive a new item. Verify that the count and identifiers still represent the same set. Then run a batch containing an unavailable or incompatible item. The interface should explain the partial outcome without falsely claiming every selected object was handled.

Measure scope corrections, deselections before confirmation, unavailable actions, partial failures and successful retries. Do not infer usability from a high batch-action count alone. People may be repeating commands because the first result was unclear. Inspect whether they can find outputs and understand which originals remain.

Ask someone to select a specific subset, use Select All within a filtered result and recover one failed operation. Include keyboard or screen-reader navigation and large text. A selection pattern is ready when users can predict the affected objects and understand the result, not merely when every checkbox can be toggled.

Review the selected set from entry through outcome

Use these checks on a working build, not only on the polished reference state.

• ✓Is selection mode discoverable and distinguishable from ordinary browsing?
• ✓Does Select All clearly refer to visible, filtered, folder or collection scope?
• ✓Does the count match the exact identifiers that the command will affect?
• ✓Are actions appropriate for selected types, permissions and count?
• ✓Are pages, files, recordings and photos named as different units?
• ✓Can partial failures retry only unfinished work without repeating successes?
• ✓Are destructive consequences and any recovery window truthful and explicit?

Connect selection to organization and output

The mobile filter and sort examples help define which items are currently visible. Selection needs a separate rule for hidden or unfiltered items.

Compare mobile sharing and export screens for batch destinations and output identity, rather than ending the flow at the action toolbar.

Use saved-items examples when the selected set is being organized into a collection rather than deleted or exported.