Tutorial•By John Iseghohi•Aug 21, 2026•6 min read

How to Add Email to Your Weekend MVP (Without SendGrid Hell)

Resend for receipts, Loops for onboarding, Beehiiv for the list. The 2026 email split, the DNS you actually need, and what to skip until you have paying users.

A blank envelope on a dark desk caught by a single mint side light, shallow focus, no text

Who This Is For

Your app can create an account.

It cannot send a password reset that lands in the inbox.

You pasted a SendGrid tutorial from 2019, hit a sender-identity maze, and decided email was "phase two." Phase two is how people get stuck in your product with no way back in.

This is the plumbing article. If you do not have a product yet, grab a wedge from startup ideas, ship login, then come back and wire mail before you invite anyone.

Three Jobs, Three Tools

Founders break email by forcing one vendor to do three jobs:

  1. Transactional — magic links, receipts, "your export is ready." Must arrive. Must not look like marketing.
  2. Lifecycle — day-1 onboarding, "you never activated," dunning. Triggered by product events.
  3. Audience — a newsletter people opted into on purpose.

In 2026 the boring split is:

  • Resend for transactional (React Email, npm install resend, free 3,000 emails/month, 100/day cap on free; Pro about $20/month for 50,000).
  • Loops when you want visual onboarding sequences tied to user properties (paid plans commonly start around $49/month; unlimited transactional on paid is the pitch versus Resend's per-send curve).
  • Beehiiv when the list is the product or the waitlist engine (free through 2,500 subscribers on Launch). That is a different surface than receipts.

Do not put password resets on Beehiiv. Do not put a weekly essay on Resend. Mixing them is how you train Gmail that your domain is a newsletter farm, then wonder why receipts die.

If you already run a list-first motion, build the newsletter before the product is the strategy piece. This article is the app-side send.

What To Ship This Weekend

Must have by Sunday

  • One verified sending domain (not @gmail.com, not the Resend onboarding domain in production).
  • Magic link or password reset.
  • One "you signed up" email that names the magic moment (see onboarding in five minutes).

Nice after 10 customers

Skip until someone asks

  • Broadcast newsletters from the product database.
  • Five-step drip with branching.
  • Open-rate dashboards you will not look at.

Step 1: DNS Before Code

Buy or use a subdomain: mail.yourproduct.com or updates.yourproduct.com. Keep marketing on a different subdomain if you run both. Shared reputation is how one promo blast tanks resets.

Add what the provider shows you:

  • SPF
  • DKIM
  • DMARC (p=none is fine on day one; you are collecting data, not punching Yahoo)

Send 20 test emails to Gmail, Outlook, and a catch-all you own. If they land in spam, stop coding features. Fix alignment first. A beautiful app that cannot reset a password is a locked box.

Step 2: Transactional With Resend

Resend is the 2026 default for Next.js founders because the DX matches the stack. React Email templates live next to the app. Webhooks cover delivered / bounced / complained.

Weekend shape:

POST /api/send
  to, subject, react template or html
  idempotency key = userId + template + day

Idempotency matters. Cursor will happily fire welcome email on every page load if you let it.

Free tier: 3,000/month, 100/day. That is enough for a launch. When you cross it, Pro at $20 for 50k is still cheaper than a day of SendGrid archaeology.

Use Resend (or Postmark, if inbox placement is the product) for:

  • Auth
  • Invoices
  • "Your job finished"

Not for: weekly roundups, affiliate blasts, "we miss you" novels.

Step 3: Lifecycle Only When You Have an Event

Loops (and Customer.io, later) earn their keep when you can fire user.activated from the product.

Until that event exists, a Resend template on a cron that checks "signed up, never created a record" is enough. You do not need a visual journey builder for twelve users.

The crossover math circulating in 2026 comparisons: Resend stays cheaper at modest volume; Loops' $49 unlimited-transactional pitch starts to win somewhere north of ~100k sends/month if contact-tier pricing does not eat you. You will not be there this weekend. Start Resend.

Step 4: Beehiiv Is Opt-In, Not Auth

Beehiiv is correct when:

  • You collect emails before the product exists.
  • You publish on a cadence.
  • You want referrals and an ad network later.

It is wrong as your magic-link pipe. Keep form_id and automations on the Beehiiv side for the list (this site's own subscribe flow does that). Keep users.email in your database for product mail. Sync later if you must. Do not make Beehiiv your user table.

The One Template That Matters

Subject: Your login link for Acme

Body, three lines:

  1. You asked to sign in.
  2. Button.
  3. If you did not ask, ignore this.

No "we're building the future of X." Auth email is not a landing page.

Welcome email, four lines:

  1. You are in.
  2. The one action that proves the product (link deep, not to /dashboard).
  3. Reply to this email — it hits you, not a void.
  4. Unsubscribe footer even on transactional-adjacent welcome if it is marketing-flavored. When in doubt, separate types.

Common Ways This Breaks

  • From: noreply@ — people cannot ask for help. Use from you actually read, or reply-to.
  • One domain for promo + resets — split them.
  • Sending from the app server on localhost in prod — you will look like a botnet. Use the vendor API.
  • No bounce handling — a full mailbox is not a lead. Suppress it.

Need a product worth emailing into? Pick a scoped build from startup ideas so the first email has a magic moment to point at.

FAQ

Can I use Gmail SMTP for v1?

For you and three testers, maybe. The day you email 50 strangers, Gmail will rate-limit you and you will look unprofessional. Spend the hour on Resend.

Is SendGrid still fine?

It works. The DX and reputation story in 2026 indie write-ups is why people leave. If you already have it configured and mail lands, do not rewrite it this weekend.

Do I need a dedicated IP?

No. Dedicated IPs are a Scale-plan problem for people sending hundreds of thousands. You are not that person yet.

TL;DR

Verify a domain. Resend for anything the app must deliver. Add Loops when you have real lifecycle events. Put the newsletter on Beehiiv, not on your auth pipe.

Ship the product first — start from startup ideas — then send the one email that gets them to the magic moment.