Everything you are told to check is public. None of it is in one place.

You are standing in a flat on a Sunday with about twenty other people. You have fifteen minutes inside, a few days to decide, and you are about to commit several million kronor. Under Swedish practice the duty to investigate the home is yours — the agent in the hallway works for the seller and is required to remind you of that duty. Nobody is required to tell you how to discharge it.

Our position: the buyer's problem is not secrecy. Almost everything you are told to check is already public. The problem is that it is scattered, undated, written in planning language, and due by Tuesday.

What you are actually asked to do

The listing tells you what has been decided. A fee increase that has been resolved appears; one that is obviously coming does not have to. The building's condition is a survey you pay for separately. The association's finances are in an annual report you can read, and several cheap services will read it for you.

None of that covers the question people actually lose money on: what is going to happen around this place? Not the flat — the view from it, the noise outside it, the road past it, the field behind it, the ground under it. That is the part where a listing's silence is not a lie and not an answer either.

Where the answers live

In Sweden this information is genuinely open. It is also spread across the municipality's project pages, its plan register, the national geological survey, the flood portal, the noise map, the transport authority's forecasts and the statistics agency — each with its own format, its own vocabulary and no shared idea of where a place is. Sweden has roughly 290 municipalities, and they do not publish this the same way as each other.

A determined buyer can do it. We know, because we do it by hand, and it takes one to two hours per address with practice and a research agent helping. Between a Sunday viewing and a Tuesday bid, across three addresses you are considering, that is not a thing a person with a job does.

Three things that make it harder than "look it up"

A plan is not a yes or a no. It has a stage, and the stage is the whole meaning. Planuppdrag, samråd, granskning, antagen, laga kraft and överklagad describe six different futures for the same drawing. In one of our own sample reports, a scheme for around sixty homes was adopted by the city in November 2025, appealed and struck down by the land and environment court in June 2026, and returns to the planning committee on 12 November 2026. A buyer who reads "sixty homes planned" and a buyer who reads "struck down, back next month" are looking at the same document and making different decisions. A search that returns the plan without its stage is worse than silence, because it feels like an answer.

A forecast is not a date, and prices are paid as though it were. A new metro line near one of the addresses we sampled is currently expected to open in 2037. The regional review that produced that number also raised the projected cost substantially and stated plainly that no new funding or timetable has been decided. "The metro is coming" is in the listing. "In eleven years, unfunded" is in a PDF. Both are true; only one of them is priced in.

The words that sell the flat are the ones nobody checks. Lugnt område. Sjöutsikt. Nära pendeltåg. Barnvänligt. These are not lies and they are not facts; they are descriptions whose truth depends entirely on what gets built in the next five years, and they are the part of the listing that does the most work on a buyer. Each one is testable against a plan register, a noise map or a timetable. Almost nobody tests them, because testing one takes an afternoon.

So what we are building

One report for one address, in Swedish. The plans within about 500 metres with their stage and their dates; the infrastructure forecasts, each labelled with who made it and when they last said so; flood, soil and noise at the point; the listing's own wording tested against all of it; the questions we would put to the agent; and a section saying what the report could not establish.

It is not a survey, not a valuation, and not a verdict on the association's finances — those are covered, cheaply, by people who do only that. It is the forward view, which is the part nobody assembles.

How you will know whether we were right

From 26 October to 22 November 2026 we are giving this to twenty people who are actually bidding in Stockholm. The first ten reports are free in exchange for a conversation; from the eleventh they cost 199 kr per address, or 499 kr for a month of them. We are publishing the lines now, before the result exists, so that moving one later would be visible:

Continue if at least 3 of the 10 priced reports are paid, and a report takes under two hours by the tenth.
Kill if fewer than 3 are paid, or if Hemnet or Booli launch the same buyer-side check while we are running.
Park if buyers clearly value it and will not pay — and then say so, rather than quietly lowering the price until someone bites.

The honest risk is not that nobody buys. It is that the problem is real, the report is useful, and it still takes three hours to make — in which case the thing works and the business does not. That is the number we are watching hardest.

Sources, stated plainly: this note cites none. The examples are from our own sample reports and the legal shape is described rather than cited, because we have not re-verified every figure ourselves and a number without its provenance is decoration. Nothing here should be relied on for a purchase. — What happens next. The result goes on our receipts page on or shortly after 22 November, against the lines above, whichever way it falls. Sann Bostad is live and reports are free during the test.

Also read: Brief Nº4 — The AI rule that is easy to write down and hard to verify →

All notes