There are two ways to use WhatsApp for business. Most people who ask about WhatsApp marketing have been using, or have been pitched on, the wrong one.
The WhatsApp Business App is free. You can broadcast to up to 256 contacts at a time, manage conversations, set automated greetings, and do most of what a small business needs. If you're a restaurant, a beauty salon, or a retail shop with a loyal local following, it's the right tool.
The WhatsApp Business API is different. It requires a Meta-verified business account, a third-party Business Solution Provider (BSP), pre-approved message templates, and a monthly send cost that scales with volume. In return, you get broadcast capability to millions, programmatic message triggering, CRM integration, and two-way conversation flows at scale.
The question that determines which you need is not "how many contacts do I have?" It's "how does WhatsApp fit into the economic model of my customer relationships?"
What the API actually costs
Meta changed its pricing model in mid-2023 from a message-based model to a conversation-based one. The current model charges per 24-hour conversation window, with different rates depending on conversation category and country.
In South Africa, a marketing conversation (outbound promotional message that opens a window) costs approximately $0.059 per conversation. A service conversation (opened by the customer) costs approximately $0.016. Utility conversations (transactional: order confirmations, delivery alerts) cost approximately $0.014.
On top of Meta's conversation fees, you pay your BSP. Pricing varies widely: self-serve platforms like Wati charge from £39/month for access, with per-message charges on top. Enterprise BSPs build bespoke commercials. A managed agency layers a service fee over the platform costs.
For a brand sending 10,000 marketing messages per month (assuming most open a new conversation window), the Meta cost alone is approximately $590/month. Add BSP charges and managed service fees, and a properly-run WhatsApp marketing programme costs between £1,500 and £4,000/month depending on volume and service level.
When it earns its cost: the three cases
Case 1: High-ticket, time-sensitive offers
WhatsApp's 98% read rate and 12-35% reply rate are only meaningful if the offer being delivered is worth the cost of the conversation. Brands with average order values above £80, fashion, homewares, electronics, travel, consistently see WhatsApp campaign ROAS exceed email by 2-4× on the same offer, because the urgency and visibility of the channel drives faster conversion decisions.
If your average order value is £20 and your margins are 30%, the maths doesn't work. WhatsApp is a premium channel with premium costs. It belongs on premium offers.
Case 2: Transactional touchpoints with high service cost
Delivery notifications, booking confirmations, appointment reminders. If your support team is fielding inbound "where is my order?" queries at scale, WhatsApp transactional messaging can intercept those contacts before they become tickets, and at a per-message cost that's a fraction of a human support interaction.
One logistics client reduced inbound WhatsApp and email enquiries by 34% after implementing proactive delivery status messages through the API. The conversation cost was £0.014 per notification. The average support contact cost was approximately £2.80. The ROI was clear within 30 days.
Case 3: Abandoned-cart recovery where email is saturated
In categories where email inbox saturation is high and list health is degraded, WhatsApp cart recovery outperforms email recovery consistently. The mechanism is simple: the customer who doesn't open the cart recovery email is not necessarily uninterested, they may have just not seen it. The WhatsApp message is harder to miss.
This only works if the customer has explicitly opted in to WhatsApp communications. Sending WhatsApp cart recovery to contacts who opted in via email is a POPIA and GDPR violation. Build the opt-in into the checkout flow, at point of subscription, or through a dedicated re-permission campaign before you run recovery sequences.
When it doesn't earn its cost
When your list doesn't have verified WhatsApp consent
The most common reason a WhatsApp programme fails to reach ROI-positive in its first six months: the opt-in list is too small because consent was only collected from a small subset of customers. Re-permission campaigns via email typically convert 15-25% of the contactable base. If your email list is 50,000 and 12% opt in to WhatsApp, you're launching with 6,000 contacts. That's a viable start, but it requires realistic volume assumptions in the business case.
When your offer frequency is low
WhatsApp works on frequency-justified value. If you're in a category where customers buy once or twice per year, the maintenance cost of a WhatsApp programme, BSP charges, consent management, template approval overhead, rarely justifies the return. Email, which you can use to stay in contact at lower cost, is usually the more efficient channel for low-frequency purchase categories.
When you don't have operational capacity to respond
WhatsApp is a two-way channel. If you send a promotional message and 12% of recipients reply, which is common, you need the operational capacity to handle those replies within a reasonable window. Brands that open a WhatsApp channel and then fail to respond to customer messages damage trust faster than if they'd never opened the channel at all.
The API doesn't automatically handle replies. You need a chatbot layer for the obvious cases (opt-out, "when will my order arrive"), and a human escalation path for the rest. Build this before you send your first campaign, not after.
The honest answer to "should we use WhatsApp?"
If your average order value is above £60, your email list is healthy, you have a clear consent strategy, and you have the operational capacity to handle two-way conversations, yes. Build it.
If you're asking because a platform sales team told you it's an "untapped channel", be careful. It is untapped, precisely because the businesses that built it badly burned their brand and violated their customers' trust. The channel's power is real. The risk of executing it poorly is equally real.
Run the maths. Build the consent first. Then decide.