Heat pumps and BEG funding: how to win these customers via ChatGPT

Why your best customers are drifting off to ChatGPT right now

Today's heat-pump customer is unsettled. They've heard about the heating law, about funding that was cut and reinstated, about neighbours with high bills and neighbours who are thrilled. Before they call you, they want to understand what they're getting into. And they increasingly seek this understanding from ChatGPT, Perplexity or the Google AI overview, no longer in ten open browser tabs.

This fundamentally changes the rules of the game for your SHK business. In the past you fought for position one on Google. Today the customer puts their question to an AI and gets a single, formulated answer with two or three recommended businesses. If you're not among them, you simply don't exist for them. The second click, on which you used to still be visible, often doesn't happen at all any more.

The good part: almost no heating installer in your region has grasped this. Whoever acts now occupies a field that will be contested in two years. Generative Engine Optimization, GEO for short, is the discipline with which you achieve exactly that: that AI systems know, understand and recommend your business.

How an AI even decides whom to recommend

To become visible, you have to understand how the model ticks. An AI doesn't recommend a business because it has the prettiest website or burns the most advertising budget. It recommends the business it finds the clearest, most consistent and most professionally sound information about. It looks for clarity: what does this business do, where, for whom, and with what demonstrable competence.

Concretely, the AI draws its answers from your website, from industry directories, from reviews, from specialist articles and from mentions on other sites. If your page says 'We are your partner for all things bathroom and heating' and nothing else concrete, the AI has nothing to pin the idea on that you're the heat-pump and funding pro. Vague wording is worthless to models.

The more precisely and more often you leave the same clear message in various places across the web, the more confident the model becomes. 'Specialist for air-to-water heat pumps and BEG funding processing in the Rosenheim area' is a signal an AI can process. Sentences exactly like that are your raw material.

The questions your customers really ask the AI

GEO begins with you knowing the real questions. And with heat pumps and funding they're very concrete. 'How much BEG funding do I get for a heat pump in 2026?', 'Is a heat pump worth it in a 1985 old building?', 'Who files the funding application with the KfW, me or the heating installer?', 'Which heat pump suits my old radiators?' These aren't fantasy questions, people type them in every day.

Your job is to be the best answer on the web for each of these questions. Not the most advertising-heavy, but the most honest and helpful. If your page cleanly explains that the basic funding is 30 percent, that an efficiency bonus and an income bonus can be added and where the pitfalls lie, then your page is exactly the fodder from which the AI builds its answer, and names you as the source.

Collect these questions systematically. Ask your team in the office which questions keep coming up on the phone. Every one of them is a potential section on your website and thus a potential hit in an AI answer.

Your website as an answer machine, not a brochure

Most SHK websites are digital glossy brochures: pretty pictures of bathrooms, a contact form, one sentence about 40 years of family tradition. Nice for humans, nearly empty for an AI. Models need text that answers questions, in clear language, with concrete numbers and subheadings phrased like questions.

So build a real guide page per major topic. A page 'Heat pump and BEG funding' with headings like 'How high is the funding in 2026?', 'For which buildings is a heat pump worth it?', 'Who handles the funding application?'. Beneath each, two or three paragraphs in plain German. Models can read out and reproduce exactly this structure superbly.

Currency matters. Funding rates and programmes change. Put the year in the text, update it and keep the details correct. An AI that notices your figures are right and current will draw on your page again more readily. Outdated or wrong details harm you doubly.

Why mentions elsewhere often count more than your own page

A model trusts a statement more when it doesn't only stand on your own page but is also confirmed elsewhere. That's why it's worth its weight in gold when your business appears in industry portals, in the regional trades directory, in the guild list, on manufacturer sites as a specialist partner and in local news with the same clear profile: heat pump, sanitary and heating, your region, funding competence.

Mentions in connection with manufacturers are especially strong. If you're a certified specialist partner for a well-known heat-pump brand and listed in their partner search, that's a trust signal models pick up. The same goes for entries on lists of energy-efficiency experts or regional climate-protection initiatives. Maintain these entries actively and uniformly.

Make sure that name, address and service description read the same everywhere. If you're called 'Heizung Müller GmbH' once, 'Müller Sanitär' another time and 'Bad & Wärme Müller' yet another, your profile frays. Consistency is no detail for machines, but the basis for recognising you as a single, reliable entity at all.

Reviews: your strongest and most honest signal

Customer reviews are enormously important for AI recommendations, because they arise independently of you. But it's not just about the star count. Models read the content. A review that says 'The business handled the entire KfW funding for us and integrated the heat pump cleanly into the old building' is a precise proof of your competence for the AI, more valuable than ten reviews saying only 'All great, happy to come back'.

Steer that actively. Don't just ask satisfied customers for a review, tell them what would be helpful to mention: the heat pump, the funding application, the consultation, the location. Invent nothing, just make the real experiences concrete. A customer who writes what you really did delivers you machine-readable proof.

Also respond to reviews, especially to critical ones. A factual, professionally grounded reply shows people and machines that you've mastered the topic and take responsibility. These dialogues are text that models read along and let flow into their picture of you.

How to make yourself AI-visible in four weeks

Start small and concrete. Week one: collect the twenty most common customer questions about heat pumps and funding and write two honest paragraphs for each. Week two: build a structured guide page from them with question-based headings and current funding rates. That's your content foundation, from which AI systems draw.

Week three: check all your entries across the web and unify name, address, services. Register in relevant directories, manufacturer partner searches and specialist lists. Week four: launch your review campaign and test for yourself by putting the typical customer questions for your region to ChatGPT and Perplexity and seeing who gets named.

This last step is your honest success check. If your business doesn't yet appear, you know where the work lies. If a competitor appears, look at what they do better. GEO isn't a one-off project but a cycle: ask, answer, prove, measure, sharpen.

Staying honest: what GEO can't do

So the expectations are right: GEO doesn't turn a bad business into a good one. If you install heat pumps half-heartedly or let funding applications slide, that will sooner or later become visible in reviews and mentions, and models pick up exactly that. AI visibility amplifies what's real, it doesn't replace it. That's uncomfortable, but fair.

Nor is GEO a trick with which you can outwit the model. Keyword stuffing, invented certificates or bought reviews get exposed and damage your profile permanently. The systems are quickly getting better at distinguishing substance from façade. So bet on real professional competence that you make visible and machine-readable.

The honest core is: if you're genuinely good at heat pumps and BEG funding, then GEO is the way to make sure exactly the customers who are currently asking an AI about it find out. You make visible what you can do anyway. And in a contested, funding-dependent market that's a head start no one takes from you quickly.

The BEG jungle: this is exactly where customers turn to the AI

Hardly any topic preoccupies homeowners like funding. How high is the subsidy? Does the climate-speed bonus still apply? Does the application have to go out before the contract is awarded? People hardly google these questions individually any more. Instead customers ask the AI directly: "Which heating installer near me knows about the BEG funding for heat pumps?" Whoever delivers no clear answer to that on their page simply doesn't appear in the recommendation.

The trick isn't to pile up technical jargon. It's about depicting the typical process honestly and understandably: funding check, application via the KfW, specialist-company declaration, payout. Write a dedicated section that explains exactly this sequence in an informal, direct tone. The AI loves structured, step-by-step explanations, because it can pass them straight on. That way you become the cited source for "heat pump funding process" instead of just one name among many.

Currency matters. Funding rates and bonuses change, and an AI quickly notices when your page is outdated. Set a visible "last checked" date and keep the percentages current. That signals reliability, to the customer and the machine.

A concrete example the AI likes to pass on

People and AI remember stories better than bullet lists. Describe a real case from your everyday work: the Berger family, built in 1998, old gas heating out, air-to-water heat pump in. Name the key figures honestly, the funding rate achieved, the approximate investment sum, the duration from consultation to commissioning. Such concrete worked examples give the AI exactly the material it can present to an unsure prospect.

Make sure to show the process, not just the result. Where did it snag? Why was the hydraulic balancing needed? How did the funding application via the customer account go? These details make you credible and set you apart from pure advertising promises. An AI recognises the difference between real experience and marketing phrases surprisingly well and favours the former in its recommendations.

Common questions you should answer directly

Build a real FAQ block onto your page that picks up the typical heat-pump worries: does the heat pump also work in an old building? Will it get warm enough in winter? Is it worth it without underfloor heating? Do I absolutely have to insulate? Each question gets a short, honest answer in two to three sentences. This question-and-answer format is exactly what AI systems take straight into their recommendations.

Phrase the questions the way your customers really ask them, not like an engineer but like a worried homeowner. "Is a heat pump in an old building nonsense?" gets you further than "Efficiency of heat pumps in unrenovated existing buildings". The AI matches the real user question with your text, and the closer the two are, the more likely you're cited.

Stick to the truth. If a heat pump doesn't make sense in a specific case, say so. This honesty builds trust and protects you from disappointed customers, and modern AI systems now weight balanced, non-sugarcoating sources more highly.