A property company just showed what AI does to margins
You run a service business. Every month you look at three numbers: what a job pays, what it costs to deliver, and what you can afford to spend to win the next customer. Those three numbers decide how fast you grow.
Heimstaden, a European residential real estate operator, just published results that move all three. According to OpenAI, the company behind ChatGPT, Heimstaden chose OpenAI in June 2025 after evaluating several platforms, with the goal of reworking how thousands of homes get managed.
The headline number is blunt. In the Netherlands, the cost of producing a virtually staged property image, meaning a digitally furnished photo of an empty apartment, fell from around €50 to less than €0.10. That is a drop of more than 99%, and Heimstaden estimates the annual savings potential in that market alone at €130,000.
It went well beyond marketing images
One of the bigger wins came from outside the tech team. Julia Erlandsen, an environmental sustainability analyst at Heimstaden with no software development background, used Codex, OpenAI's coding tool, to assess environmental risk across 20,112 property assets. The job was estimated to need five people working for a year. It took two people two months, and the company avoided hiring external consultants to solve it.
Adoption spread fast too. After ChatGPT Enterprise launched in September 2025, half of Heimstaden's employees were using it within two to three months. In Norway, a caretaker who had been sceptical used ChatGPT to assess a neglected apartment and draft a prioritised refurbishment plan. Through all of it, employees remain responsible for reviewing AI output and approving decisions.
My take
AI is best used to improve margin and increase customer lifetime value. That is where I want every owner reading this to focus. What I love about the Heimstaden story is what happens next: as a business improves those two numbers, the budget that makes sense to spend on paid ads and lead acquisition goes way up. It is a win-win.
Here is the math in plain terms. Say a plumbing job currently nets you $250 after labour and materials. If AI trims your quoting, admin and scheduling costs so the same job nets $350, every lead is worth more to you. You can pay more per click than the competitor down the road and still come out ahead.
Lifetime value works the same way. Lifetime value is the total profit one customer brings you over the years, across repeat jobs and referrals. If AI helps you follow up faster and serve customers better, that number climbs, and so does the ceiling on what you can pay to acquire each new customer. My bet is the service businesses that adopt AI for operations first will also be the ones that dominate the ad auctions in their market, because they can justify bids nobody else can.
In my experience as a Google and Meta Partner for over 10 years, the winner of a local ad auction is rarely the best marketer. It is the business with the strongest unit economics, meaning the profit left on each customer after all costs. Heimstaden just showed how fast AI can move that number.
The lesson owners should copy
Heimstaden did not start with the technology. The company broke down its most important business processes and asked where AI could create value, and its leadership lists adoption and imagination as the bottlenecks, with another lesson attached: AI will not fix a broken process, but it can dramatically scale a good one.
That applies directly to a dental practice, an HVAC company or a law firm. If your intake process loses half your leads, AI will lose them faster. Fix the process, then let AI run it at lower cost. The way I see it, a clinic that cuts admin cost per patient has effectively given itself a raise on every appointment, and that raise should flow straight into acquiring more patients.
I expect most of your competitors will treat AI as a toy for writing emails. The owners who treat it as a margin tool will quietly out-earn them, then outbid them on Google and Meta.
How to prepare your business
1. Pick one repetitive back-office task this week and test ChatGPT on it, with a person reviewing every output before it goes anywhere.
2. Work out your real customer lifetime value, the total profit a customer brings over the years, and write it down.
3. Recalculate the maximum you can afford to pay per lead using that lifetime number and your improved margin.
4. Move the savings into lead acquisition rather than letting them disappear into the general budget.
If you want help running that math against your own Google or Meta account, book a call with us.
