Measure/Water due diligence for investors
Platform · for investors
Water due diligence before the offer goes in
You can underwrite a roof from a photograph and a rent roll from a spreadsheet. The one thing that is genuinely hard to see from a desk is what the ground does at the walls — and it is the thing that turns into a line item after closing.
It is measurable remotely, from a bare-earth model of the ground, at $25 a property or 200 a month on a plan. This page is about what that measurement settles, and — more usefully at underwriting — what it does not.
What arrives
Four quantities per building, each with its own uncertainty.
| Quantity | Unit | The limit on it |
|---|---|---|
| Ground slope, per wall | % over 3 m | Only printed when the direction survives a meter of outline error in sixteen directions. |
| Contributing area, per wall | m² | Sampled at nine points and published as a range, never a decimal point. |
| Modeled flow paths | — | A model of where water goes on the surface as measured, not an observation of water. |
| Standing water | m depth, m² extent | Within 20 m of the footprint, outside every building, 3 cm floor. |
And one thing you will not find anywhere in it: an opinion about the property.
What this is not
This is not a finding that the property has a drainage problem, a wet basement, or any defect. It is a measurement of ground slope at the walls. Whether water has ever reached this structure was not measured and is not claimed.
Why we will not give you a cost, a score, or a probability
A repair cost implies a scope of work. A scope of work implies a defect. We did not measure a defect, so the chain breaks at the first link and we stop there.
The same applies to a risk score. Compressing four measurements and their error bars into one number destroys exactly the information an underwriter needs, and it dresses a measurement up as a prediction.
We tested measured terrain against Chicago's 311 basement-water complaints and it did not separate them. Read carefully, that is mostly a verdict on the proxy: an owner who knows their grading is bad pays a contractor rather than calling the city, and when a third of a block reports water in one hour the cause is the pipe under the street. It is not evidence that grade does not matter. It is one city, and it is the reason we quote no loss rate.
The part that is genuinely hard to replicate
The elevation data underneath this is public. Anyone can download it. What is hard is turning raw elevation data into a defensible number at a specific wall of a specific building, and that is where most of the engineering sits.
The surface is tested before it is trusted
The median step between adjacent elevations has to clear 0.004 m. Below that the data is a coarse source stretched onto a fine grid, and every channel computed from it would be interpolation presented as measurement. The engine declines the address rather than draw on it — a refusal, with the measured step in the payload, not a confidence downgrade.
Hollows are filled to their spill point, and the depth is kept
Skip that step and millimeter-deep pits eat the map. Measured on real ground: 282,572 m² of contributing area in Chicago and 201,312 m² in Pittsburgh drained into interior pits two to four millimeters deep and simply stopped, against largest surviving channels of 1,034 m² and 33,931 m². On synthetic ground, a bowl cut into a 2% slope fills to its spill point at 0.28 m predicted against 0.289 m measured.
The engine argues with its own answer
A published outline is a trace. Shifting a real Pittsburgh footprint north in 0.2 m steps moved one wall from +5.4% to −23.5%, crossing zero inside 1.4 m. So the whole footprint is displaced one meter in each of sixteen directions and every wall is measured again — seventeen positions — and a direction is printed only when all sixteen agree on the sign.
That is why a report on a house sometimes calls seven walls out of twelve. The other five come back undecided with their lean and the count of how many of the sixteen agreed.
The identity problem, which is the one that costs money at volume
Running one house, you would notice a wrong roof on the aerial. Running two hundred a month, nobody is looking.
Identity is decided by graded evidence, strongest first, and the grade that answered ships with the result. Where only proximity is left, that path locates rather than identifies — it is labeled in the payload, and it is never sold as a report.
We publish no headline identification rate — one figure would hide the thing that decides whether a deal number is safe to use, which is what identified the building on that house. Where the underlying records are thin the evidence is thin too. And not every disagreement is a different house: some are our source and somebody else's tracing the same roof and disagreeing about its center.
What it costs to run this at deal pace
- $25 — one property, no subscription, twenty pages.
- $99 a month — Solo, 25 lookups, renders and reports included.
- $399 a month — Pro, 200 lookups, plus development analysis on vacant land.
A refusal is never billed, and if the building is wrong the property is replaced free. We can promise that because the engine measures its own uncertainty and only releases what clears it.
Run it on a deal you already closed
The most informative first test is a property you have owned long enough to know. Put the address in, see which walls the engine calls and which it refuses to call, and compare that against what you found after the first heavy autumn.
Holding rather than trading? Portfolio pricing is annual and per property — talk to us.
Questions investors ask
Can this tell me a house will flood?
No. Nothing here predicts a water event at any structure, and we do not sell a page that does. It measures ground slope at the walls, the area draining to them and the depth a hollow fills to.
Will it give me a repair cost?
No. A cost implies a scope of work, a scope of work implies a defect, and we did not measure a defect. You get meters of wall, square meters of upstream area and centimeters of ponding depth.
How many properties can I run?
Solo is $99 a month for 25 lookups, Pro is $399 for 200, both with renders and reports included. One property with no subscription is $25.
What happens on an address you cannot resolve?
It is refused and never billed. An address that cannot be found is declined outright rather than resolved to something nearby.
Does a bad grade predict a bad deal?
We have not observed that and will not claim it. The Chicago 311 test did not separate them, which mostly indicts the proxy but leaves us with no figure to quote.