Reporting Pipeline¶

This page describes how DriftSpend turns raw AWS billing data into a weekly HTML report and static dashboard.

The implementation is intentionally small: two Lambdas, one S3 bucket, optional CloudFront distribution, and a scheduled EventBridge trigger.

1. End‑to‑End Reporting Flow¶

The diagram below shows the high‑level reporting pipeline.

At a glance:

• Collector Lambda (daily)

• Triggered by EventBridge on a cron schedule

• Calls AWS Cost Explorer to fetch the last 7 days of cost and usage
• Normalises the response into core data models

• Writes JSON snapshots to S3 under reports/raw/

• Reporter Lambda (weekly)

• Triggered weekly by EventBridge
• Reads the last 7 days of JSON from S3
• Hydrates the shared data models used by the collector
• Aggregates totals and computes projections
• Renders the HTML email + dashboard via a Jinja2 template
• Sends the email via Amazon SES
• Uploads dashboard HTML back to S3 (optionally fronted by CloudFront)

The core idea is that data models and templates are shared between the Lambdas, so adding new metrics usually means:

• Extend or refine a model (for example, add a new projection field).
• Adjust a single Jinja2 template.
• Re‑deploy; the pipeline behaviour stays the same.

2. Sample Weekly Email Report¶

To make the pipeline concrete, the following screenshots show a real DriftSpend weekly email generated from a small personal AWS account.

2.1 Summary & Top Services¶

Highlights:

• Header bar shows the reporting window and the regions included in the analysis.
• KPI cards summarise total weekly spend, number of services with non‑zero cost, and average daily spend.
• Top services table ranks services by weekly cost and share of total.
• Results are rendered with tabular numerics so columns line up cleanly across clients.

2.2 Projections & Regional Breakdown¶

What this section conveys:

• Projected (30d / 90d) cards use simple linear projections:
• 30‑day = (week_total / 7) × 30
• 90‑day = (week_total / 7) × 90
• The projection range (low / high) comes from the minimum and maximum daily totals in the week.
• The Projected (30d) by service table answers "if this week repeated for a month, where would the money go?".
• Top services by region highlights per‑region spend while keeping the table layout consistent.

All of these sections are driven by the core data models described in the Data & Model page; the reporting pipeline simply wires those models into a schedule and a template.