I think experience has shown us time and time again that the following forms are misapplications of Occam's Razor:

- business decision X isn't (in part) motivated by the intention to exploit the incidental access to data a product class brings

- data class Y seems worthless/redundant, they wouldn't make a business out of it

Scans of people's home layout definitely wouldn't be worthless, I just don't think that's the main motivator of the acquisition.

It’s not the home layout so much as the habits that data uncovers. Access to peoples homes is the holy grail and especially so in amazons case. Whereas the Alexa gives Amazon audio from inside, and ring gives them video and audio from outside, this would take things a step further. You have the obvious data like in and outs, bedtimes, house schedules (kitchen to living room, living room to bedroom, bedroom to bathroom trips) but it can also get more granular and even more so with LiDAR and image recognition. Imagine a vacuum that knows which brand of shoe you wear and size of feet you are, which can then identify if a new pair shoes up on its next scan (same size? Oh this person bought a new pair of shoes. Different size? Oh this person is potentially in a relationship now.)

These are all key data points which Amazon drools over.

The locations of walls and doors, yes. Your furniture and other stuff, no. Having a "man on the inside" able to tell whether that third bedroom is being used as an office or a nursery could be of value to a company deciding whether to advertise child care products or office supplies. Hell just giving a heads up that a household is in the market for a big ticket item like a couch could be quite valuable.

Yes it's possible to lookup the schematics but completely impractical at Amazon's scale.

If home schematics were actually a thing they cared about (and I'm skeptical of even that premise), then at Amazon's scale, they could make it easy to achieve at their scale.

It is a totally different proposition to go and often manually lookup and harvest building floorplan data from town/city archives than to passively collect the same data floorplan data generated by devices in their ordinary operation.

The former is not scaleable in any reasonable way and every operation is an ADDITIONAL cost, whereas the latter is generated for the products' main purpose and you merely add some storage for the harvested data. Moreover, the generated data will be much more up-to-date (plenty of renovations or updates happen without building permits, legally or not).