Well, look at the UK contact tracking system, where they were using an old Excel spreadsheet and it was throwing away anything after row 65535. (Max number you can store in a 16-bit unsigned integer.) All the names they were supposed to follow up just vanished every time they added more.
Computers are literal, and humans are not. Getting one to program the other is asking for trouble.
For example, humans often think 'up to 100' might include the number 100, but to a computer ... nah. Or maybe books with an odd-numbered ASIN are sorted down the bottom of the table when the code serving up the product page is deciding which of 10,000 competing ads to show. Who knows?
In two years of trying I've never had a really successful AMS ad. I've tried with 7-8 different series across the same number of genres, with 99c books, full price books, freebies, you name it.
I suspect the 4 years I spent between 2014-2018 where I published nothing and sold even less might have left a black-hole sized drag on my 'relativity' as an author.
I've never sat down and thrown wads of cash at AMS, trying to kick-start things, because I'm a frugal type and as long as I've got enough to live on I don't really care about anything else.
But on an intellectual level, as a computer programmer with all the custom software, spreadsheets and info I need to make and track ads*, I'm annoyed that I can't get everything to click. I mean, I have eight different series starters I can advertise, with a backlist of 29 full-priced books. If I can't make AMS work, what hope does my friend have when she's starting with one 2.99 book and a budget of about $10/week?
* I'm planning on modifying SalesScanner so it will import the AMS reports and show the ad spend next to each book in the royalties/income tabs. That way I can see if I'm spending too much on something which isn't selling, and not enough on something else which is.