Organisation, Teams, Tags: pick the dimension

Outcome

By the end of this lesson, you will be able to pick the right showback dimension based on the audience question, describe the underlying data source for each, and handle a question that needs more than one dimension.


Prerequisites	T0, Foundations · M4.1, RBAC · M4.4: Multi-account
Time	9 minutes

1. Concept

Showback means telling each team what it spent.

ZopNight can slice that three ways: by organisation, which is everything added up, by team, which puts a name against each number, and by tag, which lets you cut it however the business thinks about itself.

Choosing the right one before you build the report is what makes the report answer the question somebody actually asked.

DIMENSION       USE WHEN                                  EXAMPLE
──────────────────────────────────────────────────────────────────
Organisation    Org-wide rollup needed                    "What did we spend?"
Teams           Per-team accountability                    "Which teams overspent?"
Tags            Per-attribute breakdown                    "Which cost-center burned $?"

Each dimension has its sweet spot. Each report answers one dimension at a time, so pick the one that answers the question.

Each dimension’s data source

DIMENSION       UNDERLYING TABLE                          AGGREGATION
──────────────────────────────────────────────────────────────────
Organisation    cost_records (raw)                         summed by org
Teams           cost_allocation_daily                      where dimension_
                                                          type='team'
Tags            cost_allocation_daily                      where dimension_
                                                          type='tag'

Internally, ZopNight computes daily allocations into a single cost_allocation_daily table with a dimension_type column. Reports query this table with filters. The three dimensions are not separate data pipelines; they are different filters on the same allocation engine, which is why the team and tag totals reconcile with each other.

Decision tree

"What's our total spend?"                       → Organisation
"How is each team spending?"                     → Teams
"How is each cost_center spending?"              → Tags (key=cost_center)
"How is each environment doing?"                  → Tags (key=environment)
"What's per-team breakdown by tag?"              → Two dimensions: see below
"Which environment is the platform team's
 spend concentrated in?"                          → Two dimensions: see below
"How much did this marketing campaign cost?"     → Tags (key=campaign)
"What did the data team spend on RDS?"           → Teams, filtered by
                                                    resource type

The right dimension follows from the noun in the question. “Team” implies Teams; “environment” or “cost-center” implies Tags; “we” or “org” implies Organisation.

Question-first design

A common mistake is to start with the report layout (“let’s build a per-environment dashboard”) and then figure out what it answers. Question-first works better:

1. Write the question in plain English
2. Identify the noun (team / cost-center / environment / org / etc.)
3. Pick the dimension that has that noun
4. Build the report

When you cannot pick a dimension cleanly, the question is probably two questions in one: split them.

Questions that need two dimensions

The Teams report filters by provider, resource type and period. The Tags report filters by tag key, tag value, provider, resource type and tag source. Neither report takes the other dimension as a filter, so “team by tag” is not one report.

QUESTION                                    WHAT TO DO
──────────────────────────────────────────────────────────────────
"Platform team's spend by cost-center"      Open the platform team's
                                            resources (Teams), group
                                            them by cost-center yourself

"Prod environment spend by team"            Give each team's prod
                                            resources a resource group;
                                            read the groups' costs

Recurring, high-value question              Restructure: one resource
                                            group per team and
                                            cost-center pair

If a two-dimension question comes up every month, build the answer into the structure (resource groups or assignments) rather than reworking it by hand each time. Unit economics is the one place that accepts either a team or a tag as its scope, but still not both.

What ZopNight surfaces by dimension

DIMENSION       PRIMARY SURFACE              SECONDARY VIEWS
──────────────────────────────────────────────────────────────────
Organisation    Cost Reports → Summary        Per-provider, per-account, trend
Teams           Cost Reports → Teams          Per-team trend, vs budget,
                                              top resources within team
Tags            Cost Reports → Tags           Filter by key + value;
                                              top tag values; coverage

Each surface has consistent UX patterns: filters at the top, summary at left, breakdown chart in the middle, drill table at the bottom.

Audience-to-dimension mapping

AUDIENCE                                    DEFAULT DIMENSION
──────────────────────────────────────────────────────────────────
Executive (monthly review)                  Organisation (rollup)
Finance (budget vs actual)                  Organisation + Teams
                                            (with chargeback math)
FinOps Lead                                 All three (depending on the
                                            question of the week)
Engineering Leader                          Teams (their teams' health)
Platform Engineer                           Teams (scoped to their team)
Compliance / Audit                          Tags (compliance-scope tags)
Marketing (campaign cost)                  Tags (campaign key)

The mapping is not strict: a FinOps Lead might present an Organisation rollup to the executive team and a Teams view to the engineering team in the same week. But the default surfaces by audience are stable.

How ZopNight uses the dimensions

The Cost Reports page in ZopNight’s UI is organized around these three dimensions as tabs. Summary is the rollup landing; Teams is the team landing; Tags is the flexible attribute landing. Each report has its own filters (period, provider, resource type, and tag key and value on Tags); the data underneath comes from the same allocation engine.

For a new analyst onboarding, the recommended exploration order is: Summary (orientation) → Teams (where the money goes) → Tags (deeper attribution). Within a quarter, the analyst should be comfortable switching dimensions to answer questions in real-time during meetings.

2. Demo

Three meetings, three different dimensions:

MONDAY: Exec monthly review
  AUDIENCE: CEO, CFO, CTO
  QUESTION:  "Cloud cost trend MoM?"
  DIMENSION: Organisation
  REPORT:   Cost Reports → Summary
  HEADLINE:  "Total $2.1M (-3% MoM); driven by ml-infra rightsizing"
  TIME:      5-minute slot

TUESDAY: Engineering leadership review
  AUDIENCE:  Engineering Leaders + FinOps Lead
  QUESTION: "Which teams are growing fastest?"
  DIMENSION: Teams
  REPORT:    Cost Reports → Teams (trended over 90 days)
  HEADLINE:  "ml-infra +30% YoY, data +8%, platform -5%"
  TIME:      20-minute slot, with discussion of ml-infra growth

WEDNESDAY: Marketing post-campaign review
  AUDIENCE:  Marketing + Finance
  QUESTION:  "Q2 campaign cost?"
  DIMENSION: Tags (key=campaign, value=Q2-launch)
  REPORT:    Cost Reports → Tags → filter
  HEADLINE:  "$14,200 total; Lambda $4,800 (34%), CloudFront $3,400"
  TIME:      10-minute slot, with ROI discussion vs revenue

Three audiences, three dimensions, three reports: all from the same underlying cost data.

3. Hands-on (6 min)

For your team’s monthly cost review, write the questions in order:

QUESTION 1 (typically the headline):
  __________________________________________________________
  Best dimension: __________

QUESTION 2 (where is the cost concentrated):
  __________________________________________________________
  Best dimension: __________

QUESTION 3 (compare to budget or prior period):
  __________________________________________________________
  Best dimension: __________

QUESTION 4 (drill into the anomaly, if any):
  __________________________________________________________
  Best dimension: __________

If you find yourself wanting "all three dimensions for question 2",
the question is probably actually multiple questions: split it.

4. Knowledge check

Q1

“Which team overspent against budget?” Best dimension:

A. The organisation-wide rollup across all of the accounts
B. Tag-based attribution across the whole estate
C. Teams: the noun in the question is “team”
D. One single cloud account, viewed on its own

Show answer

Correct: C. Team-scoped question; Teams dimension. Cost Reports → Teams shows each team’s spend, and a team’s detail page totals the budgets that sit on its assigned resources and groups.

Q2

“Per-environment spend breakdown” requires:

A. The organisation dimension, taken entirely on its own
B. Cloud account isolation, with one account per environment
C. The team dimension, since teams map onto environments
D. Tags with key=environment, group by value (dev/staging/prod)

Show answer

Correct: D. Environment isn’t a team: it’s an attribute that applies across teams. Environment is a tag-driven attribute; Tags is the right dimension. Multiple teams might have prod resources; tag attribution sums them.

Q3

A question requires “platform team’s spend by cost-center”:

A. The Tags report, filtered by both team=platform and cost-center
B. Platform’s resources from Teams, grouped by cost-center yourself
C. The Teams report, set to group by the cost-center tag key
D. It cannot be answered with ZopNight’s data at all

Show answer

Correct: B. Each report answers one dimension: Teams takes no tag filter and Tags takes no team filter. Start from the platform team’s resources and group them by cost-center, or build a resource group per cost-center if the question recurs.

5. Apply

Cost Reports surfaces each dimension as a tab: Summary, Teams, Tags. Each tab answers one dimension; a question that needs two is worked from the team’s resources or from resource groups.

For recurring reports, save the configured filter as a dashboard widget (M4.7).

Glossary terms touched

Showback · Chargeback · Allocation

Outcome

1. Concept

Each dimension’s data source

Decision tree

Question-first design

Questions that need two dimensions

What ZopNight surfaces by dimension

Audience-to-dimension mapping

How ZopNight uses the dimensions

2. Demo

3. Hands-on (6 min)

4. Knowledge check

Q1

Q2

Q3

5. Apply

Related lessons

Glossary terms touched

Start with the bill.