Definition
Device Tagging (Device Tags) is Key-Value metadata used to label device ownership, purpose, location, etc. The same device can have multiple tags (e.g., region: East China, project: Smart Park) and is deeply integrated with Dynamic Group, OTA, configuration deployment, and report statistics.
Tag Management and Usage
Tags can be added manually, imported in bulk, or automatically labeled through rules. Predefinition and constraints of tag keys are supported to facilitate standardization. Tags are used in Dynamic Group dynamic group rules, OTA target selection, Report Analytics report statistics, RBAC data permissions, and more. They complement Device Grouping device grouping: tags focus on flexible multidimensional classification, while grouping focuses on target sets for bulk operations. Changes in tags can trigger updates to dynamic group members.
Integration with DMP
The Device Management Platform supports tag management in device details, bulk operations, APIs, and other entry points. It allows filtering device lists by tags, building Dynamic Group rules, and exporting reports. Tags can be combined with other attributes (product, firmware version) to achieve fine-grained target selection. When integrated with Asset Management, tags can be mapped to asset attributes, supporting cross-system data correlation.
Use Cases
Suitable for managing devices by region, project, customer, environment, and other dimensions; selecting targets by tags during bulk OTA; aggregating by tags in report statistics. Device tags are a foundational capability for multidimensional classification and flexible management of devices.
What key-value tags need before they can filter
A device tag is key-value metadata hung on the device record, used to express facts that a grouping rule and a fixed list are inconvenient to hard-code, such as the project, the install environment, or the responsible crew. The key is the dimension. The value is the value on that dimension. One device may have several tags, so it can appear in different filters at the same time. Tags can take part in a filter only when the spelling and case of the key, and the set of allowed values, have already been agreed, and the source of the write is trustworthy. Free text is useful only when a person is reading it. Once it enters a rule, a difference of one character makes the match fail. A tag describes a classification, not a live measurement. A quantity that changes continuously, such as temperature, should not be made into a tag.
Tags fit when the same set of devices needs to be selected along several orthogonal dimensions, and the dimensions change with the project. If members must stay fixed for a long time and be confirmed one by one, use a static group, rather than hoping nobody changes a tag. Automatic tagging must say what it is based on, for example a project code from onboarding, or a particular import. An empty value needs a definition. The absence of this key may mean unknown, or it may mean a default region. The two interpretations cannot be mixed in the same rule. A tag change should leave a time and a source. Otherwise there is no way to explain why a device suddenly entered the scope of an OTA job.
When one meaning is written as several keys
The most common problem is keys that are not unified. Some write a word for region in one language, some write “region,” and some put the city into “project.” The rule recognizes only one of them. The selection is missing a piece and does not report an error. Value formats split in the same way. The same province sometimes includes a suffix, sometimes does not, and sometimes uses an abbreviation. An extra space from a bulk paste also turns values that look the same into two values. Treating a tag as permission itself, putting “admin group” on a device and not authorizing it in the permission system, makes someone think that tagging the device makes it visible.
Deleting a key and setting the value to empty have to be separate. Deleting by mistake a key that a dynamic group refers to empties the members at once, and a bulk operation being prepared loses its target. If an overwrite import treats the sheet as the full set, old tags not written in the sheet are cleared, and years of classification on site disappear at once. If there is no cap on the number of tags, one device is written with dozens of temporary marks, and the filter screen cannot be used. Counting by tag in a statistic adds a device with several tags more than once, and the total is larger than the device count. The person reading the report must know whether this is a tag count or a device count. Renaming a key is a new dimension. Old rules do not follow automatically. Find the references first, then migrate.
An import that clears old tags missing from the sheet
When tags are imported from a sheet, say first whether the sheet is an incremental addition or a full overwrite. A full overwrite deletes every old key that does not appear in the sheet. If those keys are being used by a filter or a dynamic rule, members suddenly become empty after the import finishes, a bulk operation being prepared loses its target, and the import result still shows success. An incremental addition only adds or updates keys inside the sheet. Old keys stay. Both modes are usable, but the operator must not have to guess from the file name. An empty value and a missing key also have to be separate. An empty value may mean an explicit “none.” A missing key means not yet collected. If a rule treats both as not matching, unknown devices disappear from the statistics. If it treats both as a default region, unknown devices mix into some real region.
Renaming a key is not an edit in place. It is a new dimension replacing an old one. Old rules do not change their wording automatically. Find the references first, then migrate the data. The same meaning written as several names for region, or as a transliteration, is hit by a filter in only one of those forms. The missing devices are not reported as an error. Spaces, suffixes, and abbreviations in values also split one value into several. Before a bulk paste, check against the agreed set of values. Rows outside the set should be rejected and listed, not written as they are. When one device has too many tags, the filter screen cannot be used. A temporary mark should be deleted when it is finished. When counting by tag, say whether this is a tag count or a device count. Otherwise the total is larger than the number of devices in the record.