Quick recap for anyone who missed it: starting October 1, Meta starts charging for WhatsApp service messages — every reply inside the 24-hour window, human or bot. Rates go public Sep 1, but the shift itself is confirmed. Here's the part most people pitching this to clients are missing: the 72-hour Click-to-WhatsApp entry stays free. Someone clicks the ad, opens a chat — that window doesn't get touched. After October 1, it's not just free, it's the cheapest way to open a conversation in the entire WhatsApp stack, because literally everything downstream of it starts costing money. That's a genuinely good argument to bring to a client right now: "the door is still free, what happens after the door isn't — let's make sure what happens after actually converts." And since that's exactly the moment to bring clients in — we're running a contest for it. 🏆 $300 / $200 / $100 for the top 3 affiliates by new paying customers, Aug 21 – Sep 21. Min. 2 qualifying customers to place. Full details and your live standing: affiliate.chatfuel.com/my-contests/1358 The pitch and the timing line up: clients need this conversation before October 1 anyway — might as well be the one who gets a check for having it first. Questions on the rules? Drop them below.

Somewhere in your client's chat history, leads are explaining in their own words why they didn't buy, what confused them, what they kept asking, and what would have changed their mind. Nobody reads it. Everybody pays for focus groups instead. Three questions, and the answers are already written The whole thing is three prompts to the co-worker, and it reads the entire history, quoting real messages, never inventing a number: - "what do clients ask most" - "why do deals fail" - "why is AI failing" The first surveys the conversations for recurring questions and comes back with two lists: the ones the bot handled well, and the ones it stumbled on. The second compares lost conversations against won ones and shows you where they diverge, with the quotes attached. The third audits the moments the AI handed off to a human and tells you what it couldn't answer. A question asked forty times "Do you do weekend appointments?" asked forty times is not support noise, it's a hook. The questions every lead asks before buying are exactly what your next ad should answer before the click. Mine: don't turn the question into an insight, turn it into an answer. The translation people get wrong looks like this: - Wrong: "scheduling flexibility is a key concern for this audience." That's a slide. - Right: the ad opens with "Yes, Saturdays too." Use their words. The whole advantage of reading transcripts instead of running a focus group is that you get the phrasing for free, from people who had no idea they were being asked. Your best ad copy has already been written, by strangers, in your client's inbox. The filter before you build the hook Also mine, and it saves you a wasted creative cycle. Before a frequent question becomes a hook, check whether the answer is already in the ad or on the landing page. That splits into two very different problems: - Answer isn't there. Real hook. Put it in the first line and watch the qualified count, not the CTR. - Answer is there and they still ask. Not a hook, a clarity bug. They didn't read it, or they read it and didn't believe it. Fix the creative before you write a new one.

Qualification is the whole point of the funnel. Ads bring everyone. The client's sales team only wants the ones worth a call. So the number of questions the bot asks is the single setting that decides whether that works, and almost every agency sets it too high in the first week. Three, and a reason to add a fourth There's no hard limit. Three is the recommended starting point, and the rule behind it is the sentence worth writing on the wall: every extra question costs completion. Start lean, watch real conversations, and extend where the sales team keeps asking for a missing detail. Which means the fourth question doesn't come from the kickoff call. It comes from a salesperson who has now asked you the same thing four Mondays in a row. If the client's team genuinely can't prioritize a lead without five, ten, or twenty data points, fine, build it. Just make them show you the ten. Every question you add costs a completion at one end and a reader at the other. Sets that already work You don't have to invent these. Two working sets per niche, pick one and ship it: - Real estate. District, budget, timeline. Or: property type, buy-to-live or invest, financing. - Beauty and clinics. Service wanted, preferred date, budget. Or: service wanted, main concern, new or returning client. - Courses and coaching. Goal, current level, timeline. Or: goal, budget, why now. - E-commerce. Product interest, budget, delivery city. Or: product interest, size or variant, gift or for themselves. - Anything else. What they need, budget, timeline. Or: what they need, location, urgency. Look at the second column for a second and you'll notice something the page doesn't spell out: budget is in five of these ten sets, not ten. It's the most expensive question in the list, and half the working sets do the job without it. The test a question has to pass Mine, and it's the only filter I use now. For each question on the list, name the two answers that would make the sales team do something different this afternoon. Not know something different. Do something different: call first, call later, send the price list, drop it.