Choosing a WhatsApp Driver

Five drivers. Two official (Twilio, Meta Cloud), three unofficial (WAHA, Fonnte, Wablas). The stakes on this choice are higher than any other channel — the wrong pick can get your WhatsApp number banned, which is much harder to recover from than an email provider swap.

Decision flowchart

1. Customer-critical comms, account loss would be catastrophic? → Meta Cloud API or Twilio WhatsApp.
2. Need data on your own infra (compliance / self-hosted stack)? → WAHA.
3. Indonesia market, tight budget, low volume? → Fonnte or Wablas.
4. Global enterprise already using Twilio for SMS? → Twilio WhatsApp (single vendor, single bill).
5. Prototype or internal tool only? → WAHA Core (free).

Official vs unofficial

This is the fundamental split.

If you can't accept the ban risk, you cannot use an unofficial API. No amount of careful rate-limiting fully eliminates it.

Pricing at scale

Unofficial drivers (Fonnte, Wablas) use a flat monthly subscription — typically Rp 25k–100k/month for basic tiers (≈ $1.50–$6/month). That includes essentially unlimited volume within WhatsApp's own limits.

Official drivers bill per conversation. A "conversation" is a 24-hour window opened either by the user messaging you (free in many cases) or by you sending a template (charged).

Rough conversation pricing (varies by country and category — verify at each provider):

At 10,000 utility conversations/month:

• Fonnte/Wablas: Rp 25–100k (~$1.50–$6), ignoring per-message incrementals.
• Meta Cloud: ~$50–$500 depending on country.
• Twilio WhatsApp: ~$100–$600.

Unofficial is dramatically cheaper. Official is dramatically safer. Pick based on which failure mode hurts more.

Indonesia-specific

For Indonesia, the unofficial route is common and culturally well-understood — many production apps use Fonnte or Wablas for low-stakes comms (order updates, promos, internal tools). The ban risk is real but manageable at small volume:

• Keep sends under ~500/day per number.
• Honor opt-outs immediately — a report from even one recipient can trigger a flag.
• Maintain 2–3 backup numbers registered with the same provider; swap if the primary goes down.
• Warm new numbers gradually over the first weeks.
• Don't blast cold contacts — that's the #1 way to get banned.

For anything customer-critical (OTP, order-arrival alerts where failure means support tickets), pay for Twilio or Meta Cloud API. The cost is worth the reliability.

Self-hosted (WAHA)

Use WAHA when:

• You need WhatsApp data to stay on your infra (compliance, legal).
• You want to own the stack, not rent it.
• You're comfortable running Docker in production and have ops capacity.

Don't use WAHA when:

• You can't accept downtime from session disconnects (WhatsApp Web sometimes logs sessions out, requiring re-scan).
• Your team can't absorb the WAHA upgrade cycle.
• The cost of a banned number is high — WAHA uses the same unofficial protocol as Fonnte/Wablas; same ban risk.

Migration paths

Switching WhatsApp providers is more involved than switching email because:

• You may need a different phone number (official APIs require you to register a number with WhatsApp Business).
• Recipients see the change (your contact name may disappear from their chat, or messages may look different).
• Templates have to be re-approved on the new provider.

Plan migrations carefully — ideally during a quiet period, with communication to frequent customer contacts.

Multi-provider pattern

Register primary and backup under different channel names. If the primary starts failing, your OnError handler can switch notifications to the backup:

notifier.RegisterChannel("whatsapp",         fonnte.New(/* primary */))
notifier.RegisterChannel("whatsapp-backup",  wablas.New(/* backup */))

This is a decent reliability lift for unofficial providers. Not needed (and often not wanted) for official ones — their reliability is already high.