At 125,000 emails a month, SendGrid’s published Pro 100K example calculates to $114.95, while Amazon SES Essentials sending calculates to $20. The difference matters, but so does the work your team takes on when it replaces SendGrid’s surrounding email functions with its own systems.
To migrate from SendGrid to Amazon SES safely, inventory each sending stream, prepare the replacement integration and event handling, then move traffic in observable stages. Preserve suppression and unsubscribe rules, recreate and test templates, and keep the SendGrid route available until SES has passed application-level checks for every message type.
Treat this as an email service migration with a specific boundary: move one stream at a time, such as transactional email first, rather than changing transactional and campaign sending together. A successful API response is only the start; the cutover is complete when messages render correctly, recipient rules hold, and the events your application relies on reach the right systems.
What changes when you migrate from SendGrid to Amazon SES
SendGrid has separate Email API and Marketing Campaigns plan categories. Changing the transactional sender does not automatically move campaign audiences, schedules, or unsubscribe preferences. Identify which features are used by each stream before deciding what to replace and what to leave in place. SendGrid describes the categories separately in its plan-change documentation.
A migration inventory should cover more than API calls. For each application and stream, record the current owner, typical volume, message types, sender addresses, template source, recipient rules, and rollback route. Also document which delivery, bounce, complaint, open, or click events change application data or trigger workflows.
Use a checklist like this for each stream:
- Name the application, service owner, message type, and sending volume.
- Record sender and reply-to addresses, template identifiers, and recipient-selection rules.
- List the events consumed by the application and what each event changes.
- Identify the authoritative source for unsubscribes, complaints, hard bounces, and other suppressions.
- Record the current sending route, replacement route, and tested rollback action.
SendGrid says plan upgrades take effect immediately, while downgrades take effect on the first day of the next calendar month. Include that timing if the migration changes your subscription as well as your sender. Its published example prices are estimates and may be subject to taxes; the Email API and Marketing Campaigns categories should not be treated as one combined plan.
Choose a migration boundary and preserve the old route
Avoid changing every use of email in one release. Transactional messages such as receipts, login links, and password resets have different recipient controls and consequences from campaigns. Moving one stream first narrows the possible causes of a regression: rendering, routing, event processing, or recipient policy.
Put provider selection behind an internal interface or configuration flag. An application might call sendTransactional(message) while a controlled setting chooses the current transport. Keep the application-level message contract stable—recipient, sender, subject, HTML and text content, reply address, and application metadata—so the provider adapter can change without a rewrite of every call site.
Before sending to production, test that a route change can be reversed without changing application code. Keep the SendGrid integration available during the rollback window, secure its credentials, and document who can switch routes. Do not send the same transactional message through both services: duplicate password resets and receipts confuse users and make event comparisons harder.
Create a cutover record for each stream with its owner, current route, SES route, test evidence, rollout status, and rollback action. During an incident, a record tied to the failing message stream is more useful than a general migration plan that does not identify the current route.
Prepare the SES account and sending identity
For a new direct Amazon SES account, production access is a practical dependency: request it before planning a live cutover, and allow time for the request process. Until the account has production sending access, a team cannot treat it as a ready replacement for a production SendGrid route. Prepare the sending identity and DNS authentication in parallel so the account request is not the only outstanding task.
Treat DNS as a coordinated change. Save the current records, identify which provider each record serves, and plan overlap so SendGrid remains usable if rollback is needed. Do not replace an SPF record wholesale when other services already send from the domain. Get the precise record names and values from the SES configuration for the account and identity you will use rather than copying a record from another domain.
A DMARC record can be part of the domain’s policy, but it does not authenticate SES by itself. For example, a domain owner might adapt a record in this format:
Host: _dmarc.example.com
Type: TXT
Value: v=DMARC1; p=none; rua=mailto:dmarc@example.com
Replace the example domain and reporting mailbox with values you control. Test the visible From address and inspect received headers in test mailboxes. Confirm the sender display, reply behavior, and authentication results; successful submission only means the service accepted the request, not that the message was accepted or displayed as intended by a recipient’s mail system.
Volanea takes a different account-preparation path: it requires domain verification before sending to real recipients, but its sending account does not require an SES production-access request. That distinction is relevant if you want a managed sending layer rather than operating a direct SES account. Every Volanea email still goes out over Amazon SES.
Recreate templates and verify the application contract
For each SendGrid template, record its identifier, application call site, renderer, and approval owner. Do not assume a template can be copied into another system with identical rendering or behavior. Plan to recreate or adapt it in the replacement path, then compare rendered messages using realistic application data.
Include test cases for missing optional values, long names, non-ASCII text, HTML-sensitive input, and generated links. For password resets, inspect the exact destination URL and check that tracking does not expose or alter a sensitive token. Test text alternatives, attachments, reply handling, and any custom headers the application depends on.
Test rejection and failure behavior too. Preserve a stable application-level distinction between submission acceptance, later delivery, bounce, and complaint. A successful response from the sending API should not be recorded as final delivery, and a provider error should not silently disappear from application logs.
For direct SES, decide which team or service will own template rendering, suppression policy, and event processing if your application currently relies on SendGrid features for those jobs. AWS’s SES pricing page lists sending prices and plan features, but a sending-price calculation is not a complete feature-by-feature replacement plan. Keep rendering tests in the application repository after cutover so later edits to receipts or recovery messages are tested in code review.
Preserve suppressions and rebuild event handling
Suppression data is a recipient-safety control, not just an account export. Before sending, identify the source of truth for hard bounces, complaints, unsubscribes, and other blocks. Export the relevant records from SendGrid and make them available to the system that decides whether a message may be sent. Do not assume the migration itself imports those records or that a new provider has the same suppression behavior.
Keep suppression categories distinguishable. A campaign unsubscribe may not mean the recipient should stop receiving a password reset they requested, while a hard bounce or complaint can require a broader block. Define those rules for each stream and test them against representative records before routing live traffic.
List every event the application consumes—such as delivery, bounce, complaint, open, or click—and document its effect on your database or workflow. Then test the full path from SES through the event destination to the application handler. Map incoming provider events into your own application event model instead of assuming names and payloads match SendGrid.
Make handlers safe to run more than once and tolerant of delayed events. Retain the provider message identifier and recipient in logs, and test that a duplicate event does not repeat a business action. A controlled bounce or complaint test is more useful than assuming event delivery works because a normal test message arrived.
Compare monthly costs at the volume you actually send
Amazon SES Essentials costs $0.16 per 1,000 emails for 0–10M emails per month, $0.14 per 1,000 for 10–100M, and $0.11 per 1,000 above 100M. At 100,000 emails a month, the Essentials calculation is 100,000 ÷ 1,000 × $0.16 = $16/month. At 125,000 emails a month, it is 125,000 ÷ 1,000 × $0.16 = $20/month. AWS lists these rates on its Amazon SES pricing page.
| Option and monthly volume | Calculation or included volume | Monthly price example |
|---|---|---|
| SES Essentials, 100,000 emails | 100,000 ÷ 1,000 × $0.16 | $16/month |
| SendGrid Pro 100K example, 125,000 emails | $89.95 + (125,000 − 100,000) × $0.001 | $114.95/month |
| SES Essentials, 125,000 emails | 125,000 ÷ 1,000 × $0.16 | $20/month |
| Volanea Growth, up to 50,000 emails | Monthly-billed plan with 50,000 emails | $20/month |
| Volanea Scale, up to 200,000 emails | Monthly-billed plan with 200,000 emails | $149/month |
The SendGrid Pro 100K calculation uses its invoice guide’s example of a 100,000-email allowance and a $0.001 overage rate for each additional email. SendGrid labels that rate an example, not a universal current rate, and says paid-plan overage rates vary month to month. Overage charges appear on the invoice for the month after the overage occurred. SendGrid also says Email API Essentials and Pro sending is capped at 10 times the plan allotment starting June 25, 2026; its example sets a 500,000-request monthly cap for Essentials 50K. See its invoice example, overage guidance, and sending-limit explanation.
These figures compare different scopes. SES Essentials is a sending calculation; it does not include the engineering work or any additional services needed for templates, suppression policy, and event processing. Volanea is a managed platform price that includes an email allowance and product functions beyond raw sending. At 125,000 emails a month, the relevant Volanea tier is Scale at $149/month, which includes 200,000 emails; the price is the tier price, not a per-email charge recalculated at 125,000.
Volanea’s pricing page lists monthly billing at $5 for Starter, $20 for Growth, and $149 for Scale. The corresponding allowances are 7,500, 50,000, and 200,000 emails each month. Yearly billing costs 20% less, so distinguish the monthly-billed prices from yearly-billed equivalents when comparing plans. Volanea stores contacts for $0 per contact, and sending pauses when the monthly allowance runs out rather than accruing overage charges.
Roll out with baseline-based stop conditions
Move traffic in stages, but choose stage sizes and hold times from your own traffic pattern rather than applying an arbitrary percentage schedule. A low-volume password-reset stream, for example, may not produce enough messages in a short interval to support a meaningful comparison. A high-volume receipt stream may produce useful evidence sooner, but still needs checks for business outcomes as well as submission success.
Before the first production send, record the normal baseline for provider errors, queue depth, event arrival, and application outcomes. Agree on stop conditions that refer to those baselines and to the impact of a failure. Pause expansion if a critical template is wrong, required events stop arriving, a recipient who should be suppressed is sent mail, or errors rise beyond the team’s agreed tolerance.
At each stage, review:
- Submission success and provider errors by message type.
- Sender, subject, body, and links in representative messages.
- Required delivery, bounce, and complaint events reaching the application.
- Suppression and unsubscribe checks before messages are sent.
- Queue behavior and the outcome that matters to the user, such as successful use of a recovery link.
Assign one person authority to pause the rollout and document the rollback route before traffic moves. If a stop condition occurs, stop expansion, route new messages back to SendGrid, preserve logs and event payloads, and investigate before trying again. Keep the old integration and credentials protected during the agreed rollback window; remove them later through a separate access review.
Where Volanea fits in a SendGrid migration
Direct SES suits teams that want to own the sending integration and the surrounding systems. Volanea is an alternative when the goal is to combine transactional email, campaigns, contacts, and workflows on one platform while sending through Amazon SES. It includes domain setup, templates, campaigns, automations, webhooks, and bounce and complaint handling; the Volanea product overview describes the combined service.
A SendGrid migration still requires deliberate data and content work. Export suppression and unsubscribe records, map their meaning to your application’s rules, and load or otherwise make them available to the authoritative pre-send check before sending to those recipients. Do not assume there is a one-click SendGrid suppression import. Likewise, inventory SendGrid templates and recreate or adapt them for the destination; compare output with realistic data before routing the message type.
Volanea sends from shared Amazon SES infrastructure and does not offer dedicated IPs. If your SendGrid setup uses a dedicated IP, you cannot carry that IP reputation or its precise IP-level sending history into Volanea. Plan the cutover around domain and message-level checks, and assess whether shared infrastructure fits your requirements before scheduling the change. Do not make dedicated-IP warmup or IP continuity a dependency if the destination does not provide a dedicated IP.
Volanea’s monthly-billed Growth plan is $20 for up to 50,000 emails, while Scale is $149 for up to 200,000. That makes Scale the applicable listed tier for a 125,000-email month, rather than multiplying a lower tier’s price by volume. Volanea offers 1,000 free emails per month, unlimited contact storage at no charge, and no overage billing; when credits run out, sending pauses until the next monthly reset or an upgrade.
Every plan includes the full MCP server with 23 AI agent tools. For SMTP testing, the username is volanea and the password is the project’s secret API key. Keys beginning with sk_test_ render and log but never deliver, so a rendered test message with that key is not evidence of a delivered message. Choose direct SES when you want to build and operate the surrounding email systems yourself; choose a managed layer when consolidating those responsibilities is part of the migration goal.
FAQ
When should I start the SendGrid-to-SES cutover?
Start when the sending identity, production access, application adapter, templates, suppression checks, and event path are ready to test—not merely when an SES account exists. Keep the SendGrid route available while you validate representative messages and application outcomes. If you are also changing a SendGrid plan, account for its documented timing: upgrades take effect immediately, while downgrades take effect on the first day of the next calendar month.
What happens if I exceed my SendGrid email allowance?
SendGrid says overage charges appear on the invoice for the month after the overage, and paid-plan overage rates vary month to month. Its Pro 100K invoice example uses $0.001 per extra email, but that is an example rather than a universal rate. Review actual account usage and the active plan’s terms when setting a migration date or a billing alert.
Can I keep SendGrid Marketing Campaigns while moving transactional email?
Yes. SendGrid treats Email API and Marketing Campaigns as separate plan categories, so you can move transactional traffic while retaining campaign schedules and audiences. Keep the campaign unsubscribe and suppression controls in place until you deliberately migrate that program and test its replacement.
What should I choose if I send about 125,000 emails each month?
For direct SES Essentials, the sending calculation is $20/month at that volume, before your team’s surrounding engineering or optional service costs. Volanea Scale is $149/month for up to 200,000 emails, while the SendGrid Pro 100K published example calculates to $114.95 at 125,000 emails. Compare the operating responsibilities alongside those prices. Start free with 1,000 emails a month.