The idea
Neighbourhood help already happens in Berlin. It happens on notice boards in building entrances, in Facebook groups organised by postcode, and in the WhatsApp chat somebody started for the Hinterhof. It works, and it does not scale past the people who already know each other.
KiezHelfer is the same exchange with a wider radius. Someone needs a shelf mounted, a laptop looked at, a child collected from Kita, a garden cleared. Someone three streets away can do it. The platform's only job is to make those two people aware of each other.
Where it is now
Live beta, covering twelve Berlin districts with a small but real user base and a growing set of offers across more than ten categories: repairs, tech support, childcare, tutoring, cleaning, moving help, gardening.
Modest numbers, and I would rather say so than dress them up. A marketplace at this stage is not a traffic problem, it is a density problem, and density is local. Five hundred users spread across Germany would be worth less than the ones concentrated in a handful of Kieze.
The two sided problem
Every marketplace has the same opening question, and reading about it is not the same as being in it. Nobody posts a request into an empty platform. Nobody signs up to offer a service where there are no requests.
The lever that has worked is geography. Rather than launching Berlin, the platform is being filled district by district, because a user only needs the platform to be dense in the two kilometres around them. Twelve districts covered is a more useful sentence than a city wide launch with nothing in it.
Decisions I have made and would defend
- Price is agreed between the two people. No fixed rates, no platform pricing. Neighbourhood help has social pricing that an algorithm would get wrong, and a platform that sets the rate turns a favour into a transaction.
- Direct chat before commitment. Both sides talk first. The mismatch that ruins these arrangements is almost always about expectations rather than money, and a conversation surfaces it earlier than any form field.
- Ratings on both sides. The person asking is as much a part of the exchange as the person helping, and one sided ratings quietly make the helper the only one accountable.
- A map view, not just a list. The whole proposition is proximity. Hiding that behind a distance number in a list throws away the one thing that makes the product different.
What it taught me
My day job is infrastructure. Infrastructure has a correct answer: the mailbox migrated or it did not, the profile connects or it does not. You can be right, and being right is the job.
A marketplace has almost no correct answers. Whether ratings should be two sided is not a question with a proof, it is a bet about how people behave that you only settle by shipping it and watching. Learning to make decisions that cannot be verified before release has been the harder and more useful skill, and it has changed how I run the infrastructure side too.