Your Website to WhatsApp: Link, App or API?

Almost every Indian business site now has a green circle in the bottom right corner. Most of them were added by a plugin, nobody checked what happens after the click, and the number behind it is somebody's personal phone. That works, in the way a handwritten sign works. It stops working at a very specific point, and it is worth knowing where that point is before you pay anyone for the next step.

There are three separate things people mean when they say "WhatsApp on my website": the link, the app, and the Platform — what everyone still calls the API (Application Programming Interface, a way for software to talk to software). They cost nothing, nothing, and real money respectively. Choosing badly is expensive in both directions.

How the click-to-chat link actually works

There is no integration involved. A WhatsApp button on a website is a plain hyperlink. The address looks like this:

https://wa.me/919811234567?text=Hi%2C%20I%20saw%20your%20pricing%20page

Three things about that string matter, and plugins get them wrong constantly.

• The number is in full international format with no plus sign, no spaces, no dashes and no leading zero. So 91 followed by the ten-digit mobile number. wa.me/+91 98112 34567 will simply fail, and it fails silently — the visitor sees an error page, you see nothing at all.
• The ?text= part pre-fills the message box. It does not send anything. The visitor still has to tap send. Spaces and punctuation have to be percent-encoded, which is what the %2C and %20 are.
• WhatsApp decides where the link opens, not you. On a phone it opens the app. On a desktop it opens WhatsApp Web or the desktop app, and if the visitor has never linked a device it shows a QR code, which loses roughly half of them. Worth remembering before you assume desktop enquiries are low because desktop buyers don't exist.

The single most useful improvement you can make to an existing button costs nothing: change the pre-filled text per page. A visitor on your AMC page arriving with "Hi, I have a question about annual maintenance" already in the box tells you what they want before you say a word. A generic "Hi" tells you nothing and you burn two messages finding out.

The second improvement is to count the clicks. A WhatsApp link is an outbound click — your analytics will not record it as a conversion unless somebody sets up an event for it. Without that you are guessing which pages produce chats. That is a small piece of work on the site itself, not a WhatsApp feature.

What the free app does, and where it stops

The WhatsApp Business app — the green one you download from the Play Store — is genuinely good, and a very large number of businesses never need anything else. It gives you a business profile with address and hours, a greeting message, an away message for after hours, quick replies you trigger with a slash, labels to sort chats, a product catalogue, and a QR code and short link you can put on a card or an invoice.

Here is where it stops, and these limits are structural rather than something a clever configuration gets around:

• One number, one primary phone. You can link a small number of companion devices — a laptop, a tablet — so two or three people can be in the same inbox. But it is one shared inbox with no way to assign a chat, no internal notes, and no record of who replied. Beyond about three people it becomes chaos, and you will find out because two of them answer the same customer differently.
• Broadcast is deliberately hobbled. A broadcast list tops out at 256 recipients, and critically, a recipient only receives it if they have saved your number in their contacts. Most customers have not. So "we'll broadcast our Diwali offer to our database" quietly reaches a fraction of it.
• Nothing connects to anything. Chats stay in the phone. They do not reach your CRM, your accounting software or your order system. Exporting a chat gives you a text file, one conversation at a time. When the employee holding that phone leaves, the enquiry history leaves with them.
• No automated sending. The app cannot send an order-dispatched message by itself. Any tool claiming it can is driving the app through unofficial automation, which is the fastest way to get the number banned.

Where the API becomes necessary

The WhatsApp Business Platform is not an app. There is no interface — it is a channel that your software plugs into, so you also need an inbox or CRM that speaks to it, usually supplied by a Business Solution Provider (a reseller Meta approves to set you up).

It solves exactly four problems, and if you do not have at least one of them, do not buy it:

1. More than two or three people answering one number, needing assignment, notes and visibility over who said what.
2. Messages your system must send by itself — order confirmed, out for delivery, payment due, appointment tomorrow at 11.
3. Reaching people who have not saved your number, at scale, legitimately.
4. Chat history that lives in your business, not on a handset.

The mechanic you must understand before signing anything is the 24-hour window. When a customer messages you, you can reply with anything you like for 24 hours. Once that window closes, you cannot send free-form text. You can only send a template — a message whose exact wording you submitted to Meta in advance and which was approved, with variables for name, order number and so on. Templates fall into categories such as marketing, utility and authentication, and they are charged. Meta has changed the pricing model more than once, including a shift towards charging per message rather than per conversation, so treat any rate card you are shown as something to verify against Meta's current published pricing rather than a fixed truth.

Two more consequences that catch people out. You need documented opt-in before sending template messages, and marketing templates to people who did not ask are how quality ratings fall and numbers get restricted. And a number cannot be on the Business app and the Platform at the same time. Migrating your existing business number means giving up the app on that number, and chat history does not travel across cleanly. Many businesses keep the old number for walk-in conversations and start the Platform on a new one.

What not to buy

This category has more bad products than good ones.

• Bulk WhatsApp sender software, usually sold per-month with a promise of unlimited messages. These drive the ordinary app through automation. The number gets banned, and you lose the number the printed flyers point to.
• Purchased number lists. Illegal opt-in, terrible response, and the fastest route to your account being reported.
• Anyone selling you the green verified tick as a deliverable. Verification is Meta's decision, granted at its discretion against its own criteria. Nobody can promise it.
• A full API setup for a two-person business that gets eight enquiries a day. The free app plus a per-page pre-filled link handles that better and costs nothing.
• A chatbot that tries to answer everything. A menu that routes — sales, service, order status — earns its keep. A bot improvising answers about your pricing will eventually invent one. Keep automation on the repetitive parts and humans on the judgement.
• Any platform you cannot export your contacts and conversations out of. Ask that question before the demo ends.

The part nobody configures

Two small things affect results more than the tier you choose.

First, the widget itself. Many WhatsApp chat plugins load a chunk of third-party JavaScript to draw one green circle. A plain styled link does the same job with no loading cost, and a heavy widget on a page that already struggles is a poor trade.

Second, the away message. If your away message says "Thanks for contacting us, we will get back to you shortly", it is doing nothing. Say the actual hours, and say what someone should do if it is urgent. A customer who knows you reply at 9.30 tomorrow will usually wait. One left guessing messages your competitor at 9.31 tonight.

What to do next

Open your own website on a desktop browser and click your WhatsApp button. If it errors, or shows a QR code with no explanation, fix the number format first — that is a five-minute job that may be costing you enquiries today. Then set different pre-filled text on your three most-visited pages and make sure the clicks are being counted.

Only after those are done should you ask whether you need the Platform. The honest test is the four problems listed above. If none of them describe your week, keep the free app and spend the money on something else. If two or three do, it is worth a proper conversation about which inbox sits on top of it and how it connects to the rest of your systems — tell us what your current setup looks like and we will tell you plainly which of the three you actually need.

Questions we get asked