You draft twelve real songs, three at a time. The game then charts your album and counts how many tracks reach number one. Here is what moves that number, biggest effect first.
Each song carries a rating for how big it was. That rating does most of the work. When you are unsure, take the most famous of the three. Doing that every time is worth about one extra number one per album.
Five tracks have a job. Fill them and those songs get a boost. Filling them is worth about another number one.
An album that sticks to two or three genres scores better than one that hops across ten. Spreading those songs over several decades scores better than staying in one. The game header shows both counts while you draft.
Every track carries a little luck, like a real release. Good drafting gets you close, never certain. In testing, a careless album averages about 5 out of 12 and a considered one about 7. Twelve out of twelve happens roughly once in a few hundred games.
Every set of spins has a par: the median score of dozens of random drafts on exactly the same songs. It tells you how hard today's spins were, so a 6 on a tough day can beat an 8 on an easy one. Everyone playing the same daily sees the same par.
Once per game you can redraw the three songs on offer. The replacement comes from the same shuffle for that slot, so everyone who redraws a given track sees the same new three. Share links record which track you redrew, so scores stay checkable.
Expert mode hides the slot hints, the previews and the "fits" markers. Scoring is identical. You just have to know the roles yourself.