AI chatbots on WhatsApp give B2B marketing teams a way to answer inbound questions, qualify leads and book meetings inside the messaging app many buyers already use for work. They run on the WhatsApp Business Platform through Meta’s Cloud API or a business solution provider, and they work best when the bot handles the first exchange and a person takes over once intent is clear.
This playbook covers the parts that usually go wrong: opt-in, template approval, qualification logic and the handoff to sales. It assumes you sell to companies, with longer cycles and several people in each deal.
Why does WhatsApp matter for B2B buyers?
In markets such as Turkey, the Gulf, Latin America and much of Europe, WhatsApp is where procurement managers, founders and distributors actually reply. Email gets filtered. Phone calls get declined. A short message with a clear question often gets an answer the same day.
For B2B, the value is speed at the top of the funnel. A visitor clicks a WhatsApp button on a pricing page, asks whether you integrate with SAP, and gets a correct answer in seconds instead of waiting for a form follow-up two days later.
What can an AI chatbot on WhatsApp actually do?
A modern bot combines a language model with your own knowledge base and a set of rules. The model handles free-text questions; the rules decide what the bot is allowed to promise. Typical jobs include:
- Answering product, pricing-range and integration questions from approved documentation
- Asking qualification questions such as company size, region, timeline and current tool
- Booking a call through a calendar link or a native scheduling flow
- Sending case studies, spec sheets or price lists as documents
- Routing the conversation to the right regional sales rep with a summary
What it should not do is negotiate, quote custom prices or make contractual commitments. Write those limits into the system prompt and test them with adversarial questions before launch.
Setting up the WhatsApp Business Platform
You need a Meta Business account, a verified business, a phone number that is not already registered on the consumer app, and access to the Cloud API either directly or through a provider such as Twilio, 360dialog or MessageBird. The provider choice affects pricing, inbox tools and how easily you connect a CRM.
Templates and the 24-hour window
WhatsApp separates user-initiated conversations from business-initiated ones. When a user messages you, you can reply freely for 24 hours. Outside that window, you can only send pre-approved message templates, grouped into categories such as marketing, utility and authentication. Meta reviews templates, and marketing templates are priced differently from utility ones.
For B2B follow-ups, write utility templates where the content genuinely is a service update, such as a meeting reminder or a document you promised. Do not dress marketing content as utility; templates get rejected or reclassified.
Opt-in
You must collect clear opt-in before sending business-initiated messages. On a website, that means a checkbox or explicit text near the phone field stating that the person agrees to receive WhatsApp messages from your company. Store the timestamp and source in your CRM.
Designing the qualification flow
Most B2B bots fail because they ask too much too early. A buyer who asked a simple integration question does not want five form fields in a chat. A better pattern is answer first, then ask one question, then another only if the conversation continues.
- Answer the question the user actually asked, with a source link if possible
- Ask one context question: company size or use case
- Offer the next step that matches the answer (demo, document, pricing page)
- Ask for the remaining details only when booking a call
- Write a short summary to the CRM record: need, size, timeline, objections
Keep the language of the bot matched to the user. If a buyer writes in Turkish, the model should reply in Turkish, and your knowledge base should have Turkish content so the answers are not rough translations.
How do you hand off to a human without losing the lead?
The handoff is the moment that decides whether the bot helped or hurt. Define triggers: the user asks for a person, mentions a budget above a threshold, names a competitor, or the model’s confidence on an answer is low. When a trigger fires, the bot says a colleague will continue, and the conversation moves to a shared inbox.
The sales rep should see the full transcript and the bot’s summary before replying. Tools such as HubSpot, Salesforce and Zoho have WhatsApp connectors or provider integrations that attach the conversation to the contact. If the rep has to scroll a separate app to find context, response times slip.
Which metrics show the bot is working?
Message counts and resolution rates are easy to report and say little about revenue. For B2B, track metrics that connect to pipeline:
- Conversations started from each entry point (ads, website button, QR code at events)
- Share of conversations that reach a qualified state
- Meetings booked and meetings held from WhatsApp-sourced contacts
- Median time from first message to human reply after handoff
- Opportunities and closed revenue with WhatsApp as first touch
- Template block and report rates, which Meta uses to judge quality
Watch the quality rating of your phone number in WhatsApp Manager. A falling rating limits how many business-initiated conversations you can open, and it usually means templates are too frequent or poorly targeted.
How Alien Road applies AI chatbots on WhatsApp
In a typical engagement, our team starts with a discovery week: we read the last few hundred sales emails and support tickets to list the questions buyers actually ask, and we map which ones the bot may answer and which need a person. That list becomes the knowledge base and the guardrail section of the prompt.
Setup covers the Business Platform connection through the client’s chosen provider, template drafting and submission, opt-in wording on forms and the CRM integration. Before launch we run a test set of real and hostile questions in each target language and fix wrong or overconfident answers.
After launch, we report monthly on qualified conversations, meetings and pipeline from WhatsApp-sourced contacts, alongside the transcript review: a sample of conversations read by a person to catch answers that were technically correct but unhelpful. The prompt and knowledge base are updated from that review, not from guesswork.
Common pitfalls to avoid
- Launching without a knowledge base, so the model fills gaps with plausible but wrong claims
- Sending marketing templates to contacts who only opted into service messages
- No handoff path outside office hours, leaving buyers waiting with no expectation set
- Measuring only bot resolution rate, which rewards the bot for keeping people away from sales
Where to start this week
Pick one entry point, such as the WhatsApp button on your pricing page. Collect the twenty most common pre-sale questions and write approved answers for each. Connect a provider sandbox, build a bot that answers only those questions and hands off everything else, and let your sales team use it internally for a week before any buyer sees it.
Keep reading: AI Marketing Agency · AI Marketing