I've built this exact pipeline more times than I can count for MSP customers who want a recurring cost report, and the first half of it never changes. EventBridge Scheduler kicks off a Lambda, the Lambda pulls data from Cost Explorer or CUR, and you generate a CSV, PDF, or HTML file.
The decision that actually matters is what happens after the file exists. I've seen teams default to "send an email with SES" every time, then spend extra sprints on bounce handling and HTML templates for a report where the customer just wanted a PDF they could click and download.
So here's how I break down the three flows I actually use, and when each one earns its place.
The shared trunk
EventBridge Scheduler
│
▼
Lambda
│
AWS Cost Explorer / CUR
│
Generate CSV / PDF / HTML
This part is identical across all three flows. Where they diverge is the last mile: how the report actually reaches the customer.
Flow 1: Amazon SES
Best for professional customer-facing reports where the email itself needs to look like a deliverable, not just a notification.
Generate CSV / PDF / HTML
│
▼
Amazon SES
│
▼
Customer Email
Pros
- Full HTML formatting, not plain text
- PDF/CSV as actual attachments
- Lowest cost per email at any real volume
- Scales without re-architecting
- Delivery tracking: bounces, complaints, opens if you wire it up
Where it costs you: SES needs a verified sending domain, DKIM/SPF setup, and if you want attachments larger than a few MB you're building MIME messages by hand or using a library for it. None of this is hard, but it's not zero setup either.
Flow 2: SNS + S3 pre-signed URL
Best for simple, low-volume report sharing where you don't need the email itself to look polished.
Generate CSV / PDF
│
▼
Amazon S3
│
Generate Pre-signed URL
│
▼
Amazon SNS
│
▼
Customer Email
(Download using URL)
Pros
- Very little to build — an S3 PutObject call and a presigned URL generator
- Secure, time-limited access without managing auth
- Genuinely the cheapest option
- Good fit when you have one or a handful of customers, not thousands
Where it costs you: no HTML, no attachments, and SNS email is plain text. If a customer expects something that looks like a report rather than a notification with a link, this flow will feel thin.
Flow 3: AWS End User Messaging + S3
Best for customers who live on their phone rather than in their inbox — which, working with GCC clients, is a real segment, not an edge case.
Generate CSV / PDF
│
▼
Amazon S3
│
Generate Pre-signed URL
│
▼
AWS End User Messaging
(SMS / WhatsApp)
│
▼
Customer
Clicks link to download
Pros
- Reaches customers who don't check email regularly
- No email infrastructure needed at all
- Great for instant "your report is ready" notifications
- Same secure, temporary download link as Flow 2
Where it costs you: no in-line report content, obviously — everything lives behind the link. And SMS/WhatsApp messaging has its own registration and compliance overhead per country, which is worth checking before you commit to this path for a specific market.
Comparison at a glance
Feature | SES | SNS + S3 | End User Messaging + S3 |
|---|
Email report | Yes | Yes (plain text + link) | No |
HTML email | Yes | No | No |
Attachments | Yes | No | No |
Download via link | Optional | Yes | Yes |
SMS | No | No | Yes |
WhatsApp | No | No | Yes |
Best for | Professional reports | Simple sharing | Mobile notifications |
Relative cost | Lowest | Lowest | Slightly higher |
How I'd actually pick, for a single customer
- Monthly billing report that needs to look credible → Flow 1 (SES). This is the default for anything a finance stakeholder or account owner is going to open and judge.
- "Here's your report, go grab it" for one or two contacts → Flow 2 (SNS + S3 pre-signed URL). Don't build more than the situation needs.
- Customer is mobile-first and doesn't live in email → Flow 3 (End User Messaging + S3). Common enough in the GCC market that I keep this in the toolkit rather than treating it as exotic.
None of these three is "the right architecture."
The mistake is picking one because it looks the most sophisticated in a diagram rather than because it matches how the customer actually consumes the report. Match the channel to the reader, not the reader to the channel.
If you're building something similar and want to compare notes on the SES domain verification setup or the presigned URL expiry tradeoffs, drop a comment — happy to go deeper on either.
Loading article