Cost Categories

Cost Categories transform raw cloud spending data into meaningful business contexts by aggregating costs across multiple sources (AWS, GCP, Kubernetes clusters, etc.) and organizing them into customizable business dimensions such as departments, teams, projects, or environments.

With Cost Categories, you can:

• Map cloud costs to business units - Create categories like "Teams" or "Departments" to track spending across organizational structures

• Drill down into detailed cost analysis - Examine specific cost buckets (e.g., the "Operations" team within a "Teams" category) to identify spending patterns

• Apply consistent cost attribution - Use the same business contexts across different cloud providers and resource types

• Filter and group in CACM Perspectives - Leverage your categories in reports and dashboards for comprehensive cost analysis

The logic behind cost categories is simple: Create Rules that bring in cost data -> Rules combine to form Cost Bucket -> Cost Buckets combine to form Cost Category

For example: Imagine your company has multiple departments. You want better visibility into marketing spend across various campaigns and cloud platforms. Currently, marketing costs are scattered across parts of AWS resources (for digital marketing), GCP resources (for YouTube marketing), and some kubernetes Clusters (for social media marketing tools):

• Create a Cost Category called "Marketing"

• Add Cost Buckets to this category:

  • "Digital Marketing" - with rules for AWS resources

  • "YouTube Marketing" - with rules for GCP resources

  • "Social Media Marketing" - with rules for Kubernetes clusters

• Result: A comprehensive view of all Marketing costs across multiple cloud platforms

What is the difference between Cost Categories and Perspectives?

Cost Categories label and organize your cloud spend into custom buckets, while Perspectives let you view and analyze that categorized data using Cost Categories as filters or group-by dimensions.

How They Work Together

• Cost Categories define the labels for your costs.

• Perspectives decide how to look at those costs.

You can:

• Use a Cost Category as a filter in a Perspective → see only costs for a specific category value.

• Use it as a group-by dimension → break down total cost into category buckets.

Example: Using Cost Categories with Perspectives

Let's say your organization runs workloads in multiple AWS accounts. Some accounts are for Production, some for Development, and others for Staging.

Without Cost Categories: Creating a Perspective to analyze costs by environment requires creating perspectives with multiple rules and any new account or tag variation means updating the Perspective again.

With Cost Categories: You can define an Environment cost category that standardizes this classification by using three cost buckets (Production, Development, and Staging).

Rule

Cost Bucket

Operand: AWS, Operator: Account, Value: {all the accounts related to production}

Production

Operand: AWS, Operator: Usage Type, Value: {all the usage types related to development} OR Operand: AWS, Operator: Service, Value: {all the services related to development}

Development

Operand: AWS, Operator: Service, Value: {all the services related to staging}

Staging

Once this Cost Category is in place, you can use it in Perspectives to:

• Filter: Filter by any cost bucket.

• Group By: Break down total spend by your cost buckets.

Before You Begin

To create and manage Cost Categories in Harness CACM, you need:

Active CACM Connectors: You must have at least one active cloud connector set up for the cloud providers you want to categorize costs for: Set Up CACM Connectors.

Required Permissions: Your Harness user account must belong to a user group with the following role permissions:

• Cloud & AI Cost Management: Cost Categories: Create/Edit

• Cloud & AI Cost Management: Cost Categories: View

For more details on CACM permissions, see CACM Roles and Permissions.

Creating Cost Categories

Step-by-Step

Interactive Guide

• In your Harness application, go to Cloud & AI Cost Management > Cost Categories > New Cost Category.

Define Cost Bucket(s)

Attribute Shared Costs

Manage Unallocated Costs

Define your Cost Bucket(s)

A cost category is composed of one or more buckets. Each bucket contains filters that collect data from specific sources.

• Enter a descriptive name for the cost bucket (e.g., "Marketing Department")

• Each bucket collects costs from data sources that belong to that department

• Define Rules: Add multiple conditions to a rule using the AND operator (Use to filter data sources that include both criteria)/ OR operator (Use to filter data sources that include one of the criteria)

What are Rules?

Rules help you define which cloud resources to include in your Cost Bucket using a simple "Operand-Operator-Value" structure.

Operand The data category or attribute you want to filter on Select from:

• Common

• Cost Categories

• Cluster

• AWS

• GCP

• Azure

• External Data

Operator The comparison method that defines how to match values Choose from:

• IN: Include only resources that exactly match the specified value

• NOT IN: Include all resources except those that match the specified value

• NULL: Include only resources where this field has no value (excludes resources with this field)

• NOT NULL: Include only resources where this field has a value

• LIKE: Include resources where the field partially matches a pattern (uses regular expressions)

  • For exact pattern matching, use ^pattern$ syntax

Values The specific data points to include or exclude in your perspective Select the specific data points to apply the rule to. These will vary based on your selected Operand and Operator.

Account-level tags are available as operands too. When you select AWS, GCP, or Azure as the operand, account-level tags appear alongside resource-level tags as Label V2 keys: AWS uses the accountTag/ prefix, GCP uses projectLabels/, and Azure uses a plain key (only when tag inheritance is enabled). This lets you build a bucket from a tag applied to an entire account, project, or subscription, without tagging individual resources. Go to Account tags in CACM to understand the key formats and setup for each provider.

After configuring all your cost buckets, click on "Continue"

Important Limitations When Using Cost Categories:

• If a cost category has a shared bucket, you cannot nest it inside another cost category

• A cost category cannot reference itself in its own rules

• You cannot create loops between categories

  • Example: If Category A includes Category B, then Category B cannot include Category A

• You can nest categories up to 5 levels deep

[Optional] Attribute Shared Costs

Shared buckets allow you to distribute common costs across multiple cost buckets in a category. For example, database maintenance costs that benefit multiple departments can be shared among them using a shared bucket.

1. Name your shared bucket (e.g., "Database Costs", "IT Infrastructure")

2. Define filtering rules for the shared bucket (similar to regular cost buckets)

3. Select a sharing strategy:

Strategy

Description

Example

Equal Split

Divides costs equally among all buckets

$100 shared across 4 buckets = $25 each

Proportional Split

Allocates based on each bucket's existing costs

If Bucket A has 70% of direct costs and Bucket B has 30%: for $100 shared cost, Bucket A receives $70, Bucket B receives $30

Fixed Percentage

Distributes according to manually defined percentages

Manually assign 60% to Bucket A, 25% to Bucket B, 15% to Bucket C

1. Click on "Continue".

Manage Unallocated Costs

When you use Cost Categories in a Perspective (as a filter or Group By dimension), some costs may not match any of your defined cost buckets. These are called Unallocated Costs.

In Manage Unallocated Costs, you can choose to show or ignore unallocated costs, and choose a name for how those costs are displayed.

• Show Unallocated Values as (enter a name)

• Ignore Unallocated Values

• Share Default Costs among Cost Buckets (Coming Soon)

Overview Page

The Cost Categories overview page provides a centralized view of all your defined categories. From this dashboard, you can:

• View category details - Click any category to examine its cost buckets, rules, and configurations

• Manage categories - Edit existing categories, clone them to create similar ones, or delete categories you no longer need

• Copy cost buckets - Transfer bucket definitions between categories to maintain consistency across your cost organization structure:

  • Expand the source cost category

  • Select Manage Cost Buckets

  • Select buckets to copy

  • Click Copy and choose target categories

  • Verify with View Details

  Note: If a destination category already has a bucket with the same name, rename it first to avoid conflicts.

Use Cost Categories

Cost Categories can be used across multiple Harness CACM features. The table below shows how each feature supports different Cost Category capabilities:

Note: When using Cost Categories across multiple features, be aware of the different support levels and behaviors to ensure consistent analysis.

In Perspectives

In Dashboards

You can visualize cost categories in your custom dashboard. Cost Categories is available in AWS, GCP, Azure, Unified and External Data Explores

• Update Delay: Changes to cost categories may take up to 24 hours to appear in dashboard data.

• Historical Data Behavior: Cost category changes apply to data from the current month forward via CUR and Billing Exports. Historical data remains unchanged.

• Deletion Behavior: When you delete a cost category, it remains visible in dashboards until the end of the current month. Example: If deleted on January 24th, it will still appear until January 31st.

• Shared Buckets: Important: Shared cost buckets are not included when using cost categories in dashboards.

In Recommendations

Using Cost Categories with Recommendations

You can filter CACM Recommendations using Cost Categories to focus on specific business areas:

To filter recommendations using cost categories:

1. Go to Cloud & AI Cost Management > Recommendations

2. Select your cost category in the Filter panel

3. Select the cost buckets you want to include

4. All the resources included in your selected cost buckets will appear in the filtered recommendations view.

Stamped and Dynamic Cost Categories

Cost Categories let you group and filter cloud spend using custom bucket rules. But when you filter a Perspective by a specific bucket, the results can differ significantly depending on whether your pipeline uses stamped or dynamic mode, even when the underlying cost data is identical.

Stamped mode assigns each cost record to exactly one bucket at ingestion time, using a priority-ordered CASE WHEN evaluation. The bucket assignment is stored as a column on the row so when you filter later, you're filtering against a pre-computed value.

Dynamic mode skips the pre-assignment. Instead, it evaluates the bucket's filter conditions directly at query time against the raw data. No priority logic is applied, it simply checks whether a row satisfies the bucket's rules.

Refer to this to understand the Dynamic Toggle on Perspectives Page: Dynamic Toggle

Example Setup

Cost Category: CC1

Buckets (in priority order):

1. CB1: region = us-east-1

2. CB2: awsAccountId = acc1

Raw Cost Data:

How Stamping Works

At data ingestion, each cost record is assigned to exactly one bucket using a CASE WHEN statement that respects rule priority:

Result after stamping:

The first row matches both rules (region = us-east-1 AND awsAccountId = acc1), but because CB1 has higher priority, it wins. The row is stamped as CB1 and never considered for CB2.

Filtering Behavior: Stamped vs Dynamic

Scenario: Perspective filtered by CC1 = CB2

Stamped Mode — Result:

Note: The us-east-1 | acc1 row ($10) is not in these results — it was stamped as CB1, not CB2. It doesn't appear as Unattributed either; it simply doesn't match the CB2 filter and is excluded entirely.

Dynamic Mode — Result:

Why Are the Results Different?

The difference comes down to what question each mode is answering when you filter by a bucket.

Stamped Mode: "Which rows were assigned to this bucket?"

Stamped mode stores the bucket assignment as a column on each row at ingestion time. When you filter by CB2, the query is:

This asks: "Give me rows where the pre-computed bucket is CB2."

The us-east-1 | acc1 row was stamped as CB1 (because CB1 had higher priority). When you filter for CB2, that row simply doesn't match — it is excluded from results. It does not fall into Unattributed; Unattributed only applies to rows that matched no bucket during stamping.

Dynamic Mode: "Which rows match this bucket's filter conditions?"

Dynamic mode doesn't use the stamped column. Instead, it evaluates the bucket's underlying filter at query time:

This asks: "Give me rows that satisfy CB2's rule definition."

Both rows have awsAccountId = acc1, so both match — regardless of what bucket they would have been assigned to by priority.

The Root Cause: Priority is Lost in Dynamic Filtering

When stamping occurs, the system evaluates all bucket rules together in priority order and assigns each row to exactly one bucket. The priority logic is baked into the stamped value.

When dynamic filtering occurs, the system only evaluates the single bucket's filter you're querying. It has no awareness of other buckets or their priority. It simply checks: "Does this row match the filter conditions for CB2?"

This is why:

• Stamped respects priority → overlapping rows go to the highest-priority bucket only

• Dynamic ignores priority → overlapping rows appear wherever their attributes match

Grouping Behavior

When grouping by CC1, we always use all buckets' CASE WHEN statement with priority order:

Result (same for both modes):

Grouping reconstructs the full priority logic, so results are consistent.

Summary

Key clarification: In Stamped Mode, the $10 row doesn't go to Unattributed — it goes to CB1. Unattributed only applies to rows that matched no bucket at all during stamping.

Examples

FAQs