The claim that demos are proven to reduce sales lacks context.

The Apogee shareware model was extremely successful in its day but was predicated on certain assumptions about marketing and distribution that are no longer true. A standard scenario is that you'd read about the game in a magazine, play the demo episode on the accompanying disk, order the full version by fax or telephone and then have it within a few days.

Games based on the Apogee model were designed from the ground up to demo well and hence monetize well.

I hear what you're saying, but I feel like their approach was a bit different. A demo is clearly an incomplete piece of a larger product, hopefully setting up the customer to see the steak, but never taste it. The Apogee model really was to give away a complete contained game, and if the customer wanted "more like that" they basically bought additional "episodes" which were more or less just a couple more complete games built with the same engine and style.

i.e. You can play through all of the Doom shareware episode and enjoy the entire experience of playing a complete game.

It's hard to say it didn't work, when it did work very well for many years. Today, games just aren't produced like that I guess. But I guess the equivalent would be say, a Grand Theft Auto Game set in NYC, with what's just the part of the game where you're locked onto one borough/island, and where you complete enough of the game the episode has a logical "conclusion" or "finish". If you want to play more missions and see more story arcs that include other boroughs you can buy episodes 2 and 3. (this is also different than the DLC approach in some important ways).