What Is Per-Conversation Pricing? Why It Can Make Your Bill Skyrocket

Many chatbot and AI customer service platforms charge based on conversation volume or number of active contacts, not on a fixed capacity. This sounds fair at first glance, but the cost structure penalizes you precisely when your business is busiest. This article explains how it works, why bills spike, and how a fixed-reply pricing model offers a more predictable alternative.

What exactly is per-conversation pricing?

Per-conversation pricing calculates costs based on the number of conversation sessions between your business and your customers. The definition of "one conversation" varies by platform, but it generally follows one of these patterns:

• Per 24-hour window — all messages with a single customer within 24 hours count as one billable conversation. This mirrors the conversation model used by WhatsApp Business API.
• Per session — a new conversation is counted each time a customer starts a chat after a specified idle period.

Beyond per-conversation, there are two volume-based billing variants that work similarly in economic terms:

• Per-MAU (Monthly Active Users) — you pay for every unique customer who interacts within a month.
• Per-MAC (Monthly Active Contacts) — a MAU variant used by some multi-channel platforms.

All three share one common trait: costs rise with the number of customers, not with the capacity you actually need.

Why can per-conversation pricing make your bill skyrocket?

The core problem isn't the per-unit price — it's the unpredictability and the fact that costs are directly correlated with your own business success.

1. Your bill rises exactly when business is booming

When you run a promotion, go viral on social media, or hit a busy season (Eid, major sales events), the number of customers reaching out spikes. With per-conversation or per-MAU billing, every surge translates directly into a higher invoice. Moments that should be profitable end up adding significant cost pressure.

2. Monthly budgets are nearly impossible to predict

Because costs depend on customer behavior — which you can't fully control — you never know exactly what your end-of-month bill will look like. For SMBs with a base of potential customers drawn from ~112 million active WhatsApp users (Statista, 2025), a single successful campaign can explode your contact count within days.

3. Double billing: per-user + per-conversation/MAU

Some platforms combine two dimensions: a per-agent/per-user cost (for your team) on top of a per-conversation or per-MAU cost (for your customers). That means you're paying twice for growth — adding staff and adding customers both push your bill higher.

4. "Active contacts" are often defined very loosely

Under per-active-contact models, a customer who sends just one casual message can still count as a billable contact. You pay full price for interactions that deliver little value.

How is a fixed-reply pricing model different?

Spicelab takes a different approach: customer contacts are unlimited, and what's counted is the number of AI replies included in your plan — not the number of people who chat. You can serve 100 or 10,000 unique customers at the same plan price, as long as your total reply count stays within your quota.

Spicelab's pricing is transparent across three dimensions you can calculate yourself:

1. 1Plan tier — your base reply capacity per month.
2. 2Spark top-up — Rp 500 per additional reply when your quota runs out, not a mysterious billing spike.
3. 3Extra channels — if you need more channels (WhatsApp, Instagram, etc.).

Because replies are capped (not contacts), growing your customer base doesn't automatically raise your costs. Spicelab's AI also never goes fully offline — when your quota runs low, the system shifts into a tiered savings mode rather than going dark, so customers stay served 24/7.

Spicelab pricing

A free 7-day trial is available with no credit card required. Spark top-ups at Rp 500/reply apply across all tiers when quota is exceeded.

AI customer service platform billing models compared

Worth noting: AiSensy is WhatsApp-only and rule-based (no generative AI), primarily targets the Indian market, and has no Indonesian localization — making it a poor direct comparison for AI customer service needs in Indonesia.

When does per-conversation pricing make sense — and when doesn't it?

Per-conversation pricing can feel affordable if your chat volume is low and steady — for example, a niche B2B business with a small number of high-value clients. In that scenario, the conversation count stays small and total costs remain manageable.

However, this model becomes expensive and unpredictable when:

• Your chat volume is high or volatile (retail, F&B, beauty, fashion, clinics).
• You regularly run promotions or campaigns that trigger contact surges.
• You want a fixed monthly budget you can lock in ahead of time.

For Indonesian and Southeast Asian businesses handling thousands of WhatsApp and Instagram chats, a fixed-reply model with unlimited contacts is generally more budget-friendly — because costs don't explode every time you gain a new customer.

A simple cost calculation example

Imagine your F&B business serves 3,000 unique customers in a month, each averaging one conversation.

• Per-MAU/per-conversation model: 3,000 unique customers are counted as billable units. If the effective rate is around Rp 1,000–1,500 per contact/conversation (illustrative — actual rates vary by platform), costs could easily reach several million rupiah and keep climbing as your customer base grows.
• Spicelab fixed-reply model: as long as total AI replies stay within your plan quota (e.g., Pro at 3,000 replies), your cost stays at Rp 1.490.000 — regardless of how many unique customers you serve. If you need more, add Spark top-ups at Rp 500/reply, a number you can calculate in advance.

This illustration captures the core difference: with a fixed-reply model, your costs correlate with AI workload — not with the number of customers you serve.