Estimate email volume per month (transactional + marketing)

SES budgets depend on one input: emails per month. The trick is that “normal week” and “incident week” behave very differently. This guide gives you a simple model for baseline volume plus a peak/incident factor so you don’t get surprised.

Step 1: split volume by intent (because variability differs)

• Transactional: receipts, signups, password resets, security notifications
• Product notifications: workflow alerts, digests, reminders
• Marketing: campaigns, newsletters, onboarding sequences

Method A: transactional volume from user actions (baseline)

A defendable baseline is built from events you already track.

• Orders: orders/month × emails per order
• Signups: signups/month × emails per signup flow
• Password resets: resets/month × emails per reset
• Notifications: notification events/month × recipients per event

If you don’t have events broken out yet, start with MAU × avg emails per user and refine later.

Method B: marketing volume from campaign calendar

• Campaign sends: recipients per campaign × campaigns/month
• Variants: A/B tests and segment splits multiply sends
• Test sends: internal tests and previews (small, but include if frequent)
• Resends: resend-to-non-openers can add a predictable extra %

Step 2: add the “retry/duplicate” factor (the common spike driver)

Real systems resend. Your goal is to model it explicitly.

• Retry factor: timeouts and transient failures cause resend attempts
• Duplicate factor: bugs or race conditions send multiple emails for one event
• Alert storm factor: incidents trigger high-frequency notifications

Practical approach: keep two scenarios: baseline (e.g., +3–10%) and incident month (e.g., +30–200%) depending on your product.

Step 3: sanity checks (catch bad assumptions fast)

• Does the estimate match your historical “emails/day” shape (weekday vs weekend)?
• Is marketing bursty (campaign days) vs transactional steady?
• Are you double-counting recipients (multiple recipients per event)?
• Do you have a “silent resend” job that retries failed deliveries in bulk?

Worked example (structure)

• Transactional: 200k orders/month × 1 email/order = 200k
• Password resets: 30k/month × 1 = 30k
• Notifications: 100k events/month × 2 recipients = 200k
• Marketing: 2 campaigns/month × 500k recipients = 1,000k
• Baseline subtotal = 1.43M emails/month
• Add +5% retry/duplicate baseline → 1.50M
• Incident month scenario: +50% alerts/retries → 2.15M

Turn volume into cost

Use AWS SES cost calculator with your baseline and incident volume.

How to validate with real data

• Pick a representative week of send logs/metrics and compute average daily sends.
• Validate separately for transactional vs marketing (their curves differ).
• Check resend/duplicate rate by message idempotency keys or provider event logs.

Related guides

Sources

• AWS SES pricing
• SES sending concepts

Related guides

Related calculators

FAQ