Skip to content
Tags vs. Data Metrics: Concepts, Differences, and How They Convert
Big Data 19 min read

Tags vs. Data Metrics: Concepts, Differences, and How They Convert

This article defines tags and data metrics, details their three-type classifications (factual/rule/model tags; atomic/derived/derivative metrics), contrasts their use cases (tags for entity grouping, metrics for business monitoring), and illustrates how they interconvert with banking examples like AUM-based private banking tags and high-net-worth customer migration rates.

Smart Sea Tide
Smart Sea Tide
Smart Sea Tide

1. Tag System Concepts

Tags consist of a tag name and tag value, attached to target objects such as users, products, channels, or marketing activities. Originally popularized in the internet sector, tagging has expanded to broader industry domains.

图片
图片

Tag Types

Factual tags – describe objective, static attributes of an entity (e.g., a component is purchased or not, an employee’s gender). They come directly from entity attributes.

Rule tags – result from processing data with predefined business rules (e.g., whether cargo is overweight, whether a product is a hot seller). They combine attributes with judgment rules and are relatively objective and static.

Model tags – generated by algorithms to evaluate or predict entity characteristics (e.g., a consumer’s phone-upgrade potential: high, medium, low). They are subjective and dynamic.

图片
图片

Tag Management

Tag management splits into tag system construction and tagging .

Tag System Construction

Select the target business object based on business needs, referencing the company’s information architecture.

Design tag hierarchy according to complexity.

Define each tag and its values in detail, including tag definition, applicable scope, and generation logic:

Factual tags must align with business object attributes and cannot be added or modified arbitrarily.

Rule tags are designed per business department rules.

Model tags are produced by algorithmic models.

Tagging Implementation

Tagging establishes the relationship between tag values and instance data. It can be applied to a business object, a logical data entity, a physical table, or a single record. A “user” table (where “user” broadly means a person, organization, department, project, etc.) can be added to facilitate lookup and consumption from a user perspective.

Factual tags: system auto-tags based on the relationship between tag values and allowed attribute values.

Rule tags: system auto-tags according to designed tagging logic.

Model tags: system auto-tags via designed algorithmic models.

2. Data Metric System Concepts

Data metrics distill the essence of big data into statistical quantities that stakeholders need daily. Every mature internet company can build its own metric system with proper event tracking. Planning a metric system is a core competency of platform-type data product managers, distinguishing them from other PMs and data analysts.

What Is a Data Metric?

A data metric differs from traditional statistical indicators; it is a summary result derived from data analysis that makes business goals describable, measurable, and decomposable . Common examples include PV and UV.

A metric comprises three elements:

Dimension – the angle from which to measure (e.g., a time period).

Aggregation method – the statistical operation (e.g., sum, count, ratio).

Measure – the concrete target and unit (e.g., minutes).

图片
图片

Example: Total playback duration – dimension = selected time range, aggregation = sum of durations, measure = minutes.

What Is a Metric System?

A metric system systematically organizes metrics by business model, standard attributes, classification, and layering. Different business stages and types demand different partitioning standards. Macroscopically it forms a comprehensive organic whole; microscopically each metric has a specific meaning reflecting a detailed objective fact.

Value of a Data Metric System

图片
图片

Comprehensively supports decision-making – gives management an objective, data-driven view of company development, avoiding “blind men feeling an elephant.” New metrics can be added as new businesses emerge.

Guides business operations – fine-grained metrics reflect real-time user feedback on operations. Example: content editors push book lists; metric system captures user preference for each list, informing whether to continue similar topics.

Drives user growth – behavioral metrics enable deep understanding of user paths and preferences, helping optimize product details, improve retention and conversion. Systematic metrics link multiple dimensions for global analysis, solving the pain point of isolated metrics.

Unifies statistical calibers – ensures consistent definitions and logic, avoiding ambiguous definitions that degrade data quality. Also reduces duplicate computation, cutting log redundancy and server waste.

3. Differences Between Data Metrics and Tags

图片
图片

Content Perspective

Metrics describe objective facts, are usually numerical, and are independent of specific entities (e.g., GDP, CPI, non-performing loan ratio).

Tags are human-defined concepts used to group entities (e.g., long-tail vs. high-net-worth customers; high-risk vs. low-risk products). They combine human judgment with objective data.

Interconversion : Metrics can derive from tags (e.g., high-net-worth customer migration rate uses two tags). Tags can derive from metrics (e.g., private banking tag: AUM >= 5 million; different banks set different thresholds).

Classification Perspective

图片
图片

Metric classification – atomic metrics (base counts like customer count, transaction volume), derived metrics (atomic metric + dimension, e.g., mobile banking customer count), derivative metrics (ratios, etc.).

Tag classification – factual tags (raw attributes like gender, car ownership), rule tags (statistical results with human judgment, e.g., age-group buckets), model tags (abstract group characteristics like “monthly spenders”, “white-collar”, “shopping enthusiasts”).

Usage Scenarios

图片
图片

Metrics are decomposed: a top-level metric (customer count) is broken down by dimensions and qualifiers (mobile banking customer count, 7-day new customer increment) to monitor and evaluate business performance.

Tags are aggregated: they portray group characteristics (customers or products) to enable differentiated strategies – e.g., different marketing activities, eligibility checks, or product recommendations per tag.

Concrete Examples

China Merchants Bank uses AUM (Assets Under Management) as its north-star metric, decomposing it into cash/deposits, investment modules, and other business modules, while also segmenting users into active/inactive, forming a full metric framework that guides business.

Ping An Bank Pocket APP leverages tags for “thousand faces for thousand people” – each user sees personalized content because tags filter and match the right products to the right user segments.

Original Source

Signed-in readers can open the original source through BestHub's protected redirect.

Sign in to view source
Republication Notice

This article has been distilled and summarized from source material, then republished for learning and reference. If you believe it infringes your rights, please contactadmin@besthub.devand we will review it promptly.

user profilingtagsdata metricsdata product managementnorth star metricbanking analyticstag classificationmetric classification
Reader feedback

How this landed with the community

Sign in to like

Rate this article

Was this worth your time?

Sign in to rate
Discussion

0 Comments

Thoughtful readers leave field notes, pushback, and hard-won operational detail here.