AI is everywhere right now. In every feed, every webinar, every second ad promise. And almost every coach, consultant or service provider I talk to is trying out something.
But the honest question isn't whether you use AI. It's this: are you making more money because of it?
For most people the answer is no. Not because the tools are bad, but because they get thrown at tasks that were never the problem. In this article I'll show you how we decide, the one place where we use AI and automation consistently, and the three building blocks we have live because of it.
Everyone uses AI, hardly anyone sees it in the bank account
The numbers show how fast this is moving. According to Bitkom's survey on AI adoption in German businesses, roughly one in three companies with 20 or more employees now uses AI, up from one in five a year earlier. Germany's Federal Statistical Office measures more conservatively but sees the same trend: 26 percent of companies used AI technologies in 2025, and 23 percent of small firms with 10 to 49 employees.
In other words: using AI is no longer the differentiator. What you use it for is.
What I see in practice is mostly these four things:
- Generating Instagram captions and hooks
- Creating images for posts and thumbnails
- Rewording and polishing emails
- Summarizing meeting notes
All legitimate. None of it is bad. But if your calendar is empty, a nicer caption won't fill it. And if your calendar is full, an AI image won't save you a single hour of discovery calls.
AI for coaches starts with one question
Before you touch any tool, answer this honestly:
Where are you actually stuck? Not enough inquiries, or not enough time?
Those are two completely different problems, and they need different solutions.
- Not enough inquiries: your bottleneck is visibility and trust. AI only helps here if it gets more qualified people to the point where they want to talk to you.
- Not enough time: your bottleneck is pre-qualification. You're sitting in calls with people who were never going to buy, explaining the same thing for the thirtieth time.
Anything you automate somewhere else changes your revenue by exactly zero. That's plain math: revenue is inquiries times close rate times price. A measure that moves none of those three factors is activity, not growth.
Where AI makes us money: between the click and the booked call
That's why we use automation and AI consistently in exactly one place: between the first click and the booked appointment.
Why there? Because that's where most leads get lost without anyone noticing. The Harvard Business Review study on the short life of online sales leads tested more than 2,000 companies. Firms that responded within an hour were many times more likely to have a qualifying conversation than those that waited a day. Almost a quarter of the companies tested never responded at all.
The study is from 2011, but the principle has only gotten sharper: someone who finds you through an ad has three other tabs open. If there's a manual step between interest and appointment, that lead is gone.
So we build that path to work without us. Three examples, all live.
Example 1: the application funnel
Six questions, and every lead sorts themselves. If they're a fit, they see the calendar immediately. If they're not, they get an honest answer and a better next step.
The trick isn't the technology, it's the questions. Good qualifying questions don't ask for name and industry. They ask about the situation:
- Where are you right now? (For example: how many clients do you currently work with?)
- What have you already tried?
- What's your goal for the next six months?
- How soon do you want to start?
- Who else is involved in the decision?
- What budget have you realistically set aside for a solution?
Technically, it's a switch: depending on the answer, the person goes down a different path. Tools like Typeform call these logic jumps for branching forms, and scheduling tools can now do it natively too. Calendly's help center explains how routing forms send people to different calendars based on their answers.
One caveat: every extra question costs you form completions. The Nielsen Norman Group's core guidelines for usable web forms recommend asking only for what you really need. Six questions work when each one triggers a decision. A question whose answer you never use gets cut.
If your leads come straight from Meta ads, the same principle applies inside the form. Meta offers a "Higher intent" form type with an extra review step, explained in its help article on the different instant form types. Each lead costs a little more, and the call afterwards is noticeably better. How we run that in campaigns is on our Meta Ads page.
Example 2: a calculator instead of a PDF
The classic lead magnet is a PDF. "The 7 mistakes in …", downloadable in exchange for an email address. Most of those PDFs are never opened.
For a company formation consultancy we built a tax calculator instead. Visitors enter a few basic details and see their own number within three minutes. Not a sample calculation, but theirs.
That has three effects:
- The value is immediate. Not a promise for later, but a result now.
- The conversation starts at a different level. The prospect comes into the call with a concrete number and wants to know how to get there.
- The page is content in its own right. A real calculator answers a search query better than a form in front of a PDF. Google's guidance on helpful, people-first content sets exactly that bar: did the visitor get what they came for?
This works just as well for coaches. A calculator that shows how many hours a month a bottleneck costs. Or what one additional client per month adds up to over a year. The math has to be simple and honest. If you use flattering formulas, you lose trust at exactly the moment you're trying to build it.
Example 3: a quiz with instant results
Ten questions, instant result. For a coach we built a "wheel of life" quiz: participants rate ten areas of their life and immediately see where they're stuck.
That puts the problem on the table before the conversation starts. The coach doesn't have to dig for it anymore and can pick up right where the prospect has already recognized themselves.
This is where AI shows its strength most clearly: the result can be written far more individually than three prewritten outcome texts. But the same rule applies: the quiz lives or dies by its questions, not by the model behind it.
What you need to watch legally
As soon as AI and automation interact with real people, obligations kick in. This isn't legal advice, it's the standard we work to (and we operate under EU rules):
- AI has to identify itself. If a chatbot or assistant talks directly to prospects, Article 50 of the EU AI Act has required since August 2026 that people are told they're interacting with AI. One visible sentence at the start is enough.
- No rejection by algorithm alone. The GDPR protects people in Article 22 against solely automated decisions with significant effects. A funnel that pre-sorts and offers a suitable next step is different from a system that rules people out for good. With us, everyone can still get in touch.
- Be transparent about data. If you store answers from a quiz or calculator, Article 13 GDPR on information obligations requires you to explain what for and for how long. That belongs in your privacy policy and as a short note right at the form.
- Tracking only after consent. In Germany, pixels and analytics tools on funnel pages need consent before they load under Section 25 of the TDDDG.
Sounds like box-ticking. But it's exactly what separates serious funnels from the ones that go offline again after three months.
How to find your own starting point
If you want to use AI in your business in a way that pays off, go in this order:
- Name your bottleneck. Not enough inquiries or not enough time? One sentence, one answer.
- Map the path from click to appointment. Every step, every wait, every manual action.
- Mark the spot where leads drop off. Usually it's waiting for a reply, or a call with someone who was never a fit.
- Build one building block right there. Application questions, a calculator or a quiz. One is enough to start.
- Measure. How many appointments per week, how many are a fit, how many buy. After four weeks you'll know whether it works.
Only once that path is running does the nicer caption pay off. Because then it leads somewhere.
The most common mistakes
- AI on content instead of the sales path. More posts do nothing if nobody books a call afterwards.
- Too many tools at once. Three half-configured systems are worse than one that runs cleanly.
- Questions without consequences. Every question in the funnel has to decide something, or it just costs you drop-offs.
- Automation without testing. Click through it yourself before ad spend hits it. How expensive untested automation gets is something I covered in the article on reactivating old leads.
- Everything on one platform. If your whole sales path hangs on a rented channel, you can lose it overnight. That's exactly what happened to us with WhatsApp, and the story is in our article on owning your email list.
Bottom line: use AI where it hurts
AI for coaches isn't a content topic. It's a sales path topic. The question isn't which tool is trending, but where between click and appointment you're losing people today.
Start exactly there. One building block, built properly, measured honestly. It's less spectacular than any AI demo on Instagram, but it's the part that shows up in your bank account.
And if, while building, you realize you're missing the proof on that path, meaning content that builds trust before you're ever on a call yourself: that's exactly what we produce video courses starting at €590 for. Because the best funnel does nothing if nobody knows why they should trust you in particular.
