Customer support decision
Mara
A narrow self-hosted replacement that drafts and sends lifecycle emails plus simple variant testing is realistic for a small team, but reproducing Mara's full polishing (deliverability, reply-guardrails, integrations and operational playbook) would take substantial ongoing effort.
Visit website↗Open-source builds that already do this
Every project below is open source and already does this job today. Fork one, self-host it, or take the parts you need - the build prompt further down assumes an empty file, and this is the shortcut past that. Licences differ; check the one on each card before you ship. All Mara alternatives, with the arithmetic →
What a replacement has to do
- Ingest product and billing events, generate email drafts in a customer's voice, run A/B variants on real sends, allocate more sends to winners and rewrite losers, queue drafts for approval and send via the customer's sending domain, classify replies and draft response candidates.
What it still won’t have
- Built-in sending infrastructure and deliverability expertise (reputation maintenance, dedicated IP)
- Production-hardened reply-classification guardrails and legal/complaint protections
- Polish of integrations (one-click Clerk/Stripe/Supabase connections) and ML-backed playbook
- Operational monitoring, weekly plain-language reporting, and ongoing model tuning
What remains hard
- Product polish and ongoing maintenance
First-year cost
Keep paying
Paying is—cheaper in year one.
On cash alone, building overtakes the subscription at 2 seats.
Money you would actually spend
Time you would spend
—
What you would spend
What we assumed
The verdict above measures whether you could build it. This one is only about money.
Runnable build prompt
Build a minimal self-hosted lifecycle email assistant using Next.js (or Express) + Postgres + Redis for queues + SendGrid (or Postmark) + OpenAI/Anthropic for generation. In-scope: (1) an identify/track HTTP API that accepts signups, billing events and product events; (2) a small web UI to paste 1–3 sample emails and view/approve queued draft variants; (3) LLM-driven generation of 3 variants per program step and basic A/B allocation based on opens/clicks; (4) sending via a customer-owned subdomain with SPF/DKIM setup instructions and an inbox webhook to ingest replies; (5) reply classification into at least 5 classes and draft-reply generation; (6) logging, retries, idempotency, and tests for API endpoints and generation pipelines. Out of scope: advanced deliverability operations (managed dedicated IPs), multi-tenant billing UI, ML model training, and recruitment of human-in-the-loop reviewers. Include error handling, rate-limit handling, and unit + integration tests for core flows.
How we checked
How the score was reached
- Partly verdict base52
- An open-source build was found+5
- 5 cited sources+3
- Price verified on pricing page+3
- 3/3 assessment runs agreed+4
- Evidence score67
The base comes from the verdict. Everything under it is a check that either happened or did not, and each one is a fact frozen in this record rather than a judgement made at render time - so the same evidence always produces the same number.
How scoring works →Cited sources · 5
Every page the run actually retrieved.
- official productHire Mara — product
- official pricingHire Mara — pricing
- official docsHire Mara — docs
- open sourceknadh/listmonk
- open sourcemautic/mautic
Integrity checks
What held up, and what did not.
