When to switch from Trend to Flow

TREND       Time-series chart. Answers: "Are we going up or down?"
            X-axis: time. Y-axis: cost.
            Use for: direction, growth, regression detection.

FLOW        Cascade diagram (Sankey). Answers: "Where is the money?"
            Multi-column visualization of cost composition.
            Use for: composition, attribution, "where to look" diagnosis.

"Are we going up or down over time?"
"Is our cost stable week-over-week?"
"How did last week's deployment affect cost?"
"Is the schedule producing the expected savings?"
"Did the right-sizing campaign reduce spend?"
"What's the rate of growth (slope) over the period?"

"Where exactly is our $40K going this month?"
"Which team's resources are biggest contributors?"
"Across which providers does our K8s spend?"
"What's the breakdown of cost by resource type?"
"Inside platform-team, which sub-services dominate?"
"Among on-demand spend, which services contribute most?"

TREND BLIND SPOTS:
  Cannot tell you composition
  Aggregate trend masks heterogeneous causes
    (e.g., one team grew 50%, another shrank 30%; total looks flat)

FLOW BLIND SPOTS:
  Cannot tell you direction
  Snapshot of one point in time
    (e.g., the biggest contributor today may not be the fastest-
    growing: Flow doesn't show acceleration)

TREND FIRST:
  Confirm whether direction is concerning
  If concerning: identify the time window of the change

FLOW SECOND:
  For the concerning window, show composition
  Identify which dimension dominates the spend
  Drill to specific resources

OPTIONAL: Back to Trend:
  For the identified component, show its specific trend
  Confirm the component is the driver (not just the largest contributor)

Q: "Our bill is up 12% MoM. Why?"
TREND only:
  Confirms the +12%
  Maybe shows when it happened (week 3 vs week 1)
  Cannot tell you: which team, which service, which resource

THE NEXT STEP must be Flow (or per-team Trend).

Q: "Where is our $80K monthly spend going?"
FLOW only:
  Shows: AWS $40K, GCP $30K, Azure $10K
  Drill: which accounts, services, teams within
  Cannot tell you: whether any of this is growing
    (the biggest static contributor might be steady;
    a small contributor might be growing fast)

THE NEXT STEP for growth analysis is Trend per-component.

SLIDE 1: TREND (org-wide)
  "Last month: $96K. Prior: $84K. +14% MoM."
  Audience: leadership; sets the headline.

SLIDE 2: FLOW (composition this month)
  "Where the $96K goes: AWS $58K, GCP $28K, Azure $10K."
  Audience: leadership + engineering; sets the where.

SLIDE 3: TREND PER TOP CONTRIBUTOR
  "AWS trend: +18% MoM (the +14% is driven by AWS growth)."
  Audience: engineering; identifies the driver.

SLIDE 4: FLOW (drilled into AWS)
  "Inside AWS: prod-platform $24K, prod-ml-infra $20K, ..."
  Audience: engineering; identifies the team / workload.

SLIDE 5: TREND PER DRIVER (ml-infra deeper)
  "ml-infra +30% MoM; driven by new training cluster."
  Audience: engineering + ml team; ready to discuss action.

CONCLUSION:
  Action items, owners, next review date.

SURFACE                   PRIMARY VIEW
──────────────────────────────────────────────────────────────────
Reports → Cost Trend       Trend (with optional Flow toggle inside
                          the same report)

Reports → Cost Flow         Flow (Sankey) with optional Trend toggle

Cost Breakdown card        Side-by-side Trend + Flow with linked
                          time-range and filters

Dashboard widgets          Both views as separate widgets;
                          dashboard can show one or both

QUESTION: "Why is the bill up 18% this month?"

T+0       Open Reports → Cost Trend
T+5 s     Trend chart confirms: $96K vs $81K prior = +18%
          Inflection point: around week 3

T+10 s    Toggle to Flow view (same time range)
T+15 s    Sankey shows:
            AWS:    $58K (+25% from prior)
            GCP:    $28K (+5%)
            Azure:  $10K (-3%)
          AWS is the +18% driver.

T+20 s    Click AWS in the Sankey
          Breadcrumb: AWS
          Drill cascade: Account → Type → Team
            prod-aws-us-east-1: $35K (+35%)
            prod-aws-eu-west:    $15K (+8%)
            (other accounts):    $8K (-5%)
          prod-aws-us-east-1 is the AWS driver.

T+30 s    Click prod-aws-us-east-1
          Breadcrumb: AWS > prod-aws-us-east-1
          Cascade: Type → Team
            EKS:    $14K (+60%)
            RDS:    $8K (+5%)
            EC2:    $7K (+10%)
          EKS is the account driver.

T+40 s    Click EKS
          Drill: which team's EKS?
            team-ml-infra: $11K
            team-platform: $3K
          ml-infra dominates.

T+45 s    Switch back to Trend (now filtered to ml-infra EKS)
          Confirms: ml-infra EKS +60% MoM
          Inflection: week 3: matches the overall pattern

T+1 min   Diagnosis: ml-infra's EKS cluster grew sharply in week 3.
          Likely cause: new ml training workload deployed.

T+1.5 min ACTION:
          Confirm with ml-team lead: planned or anomalous?
          If planned: expected; document for next month's budget
          If anomalous: investigate further; revert if needed

QUESTION:    __________________________________________________________

WHICH VIEW first?
  □ Trend (looking for direction / growth)
  □ Flow (looking for composition / attribution)

PROJECTED next step:
  □ Trend → switch to Flow to find the driver
  □ Flow → switch to Trend per top contributor
  □ Flow only (snapshot composition is enough)
  □ Trend only (direction is enough)

ESTIMATED time to specific answer:    _____ minutes