Okay... starting from 1992.
-Force Sega of Japan to competently support the Sega CD throughout its life. SoJ game support for the SCD in 1991, 1992, and 1995 was minimal; they only tried in '93 and '94, and even then some major titles ended up on cart -- Outrun 2019, Phantasy Star IV. Things were so bad early on that SoA was forced to include third-party multiplatform games (Sol-Feace is a X68000 port, and Sherlock Holmes: Consulting Detective is also on PC, Turbo CD, and more) and some Genesis ports as the launch pack-in library! Pretty sad. And the next major packin was Sewer Shark, another game which later on got a port (to 3DO). SoJ needed actual games for the system.
-Make up some information for first and third party teams about how to do good sprite scaling on the Sega CD. Few teams managed to figure it out -- Malibu, a couple of SoJ teams (for Beyond the Limit and Sonic CD bonus stages), Core, and not much else.
-Sega Multimedia Studios was a good idea, and their two games are good, but were they worth the significant expense? Also, even IF it was the mid '90s... Wild Woody, conceptually, should have been changed at some point.
-Never release the 32X. It was a huge mistake.
-Somehow get SoJ to understand that by undercutting SoA, they destroyed most of their sales basem, and they could NEVER replace that all with Japanese sales. As much as they hated it, they needed the American market, and needed to focus on trying to stay relevant in it, instead of going for Japan (with the Saturn) and losing when FF7 and DQ7 were announced as PS1 exclusives.
-Don't release the Saturn as it is. Release a more powerful system (the SGI system, the Lockheed system, something else, they had options), either in late '95 or in '96. This would hurt in Japan, but do a lot more good in the West than bad in Japan, overall. Sega went from over 20 million Genesises in the US to not even 2 million Saturns! Absolutely catastrophic, and with better hardware and no 32X the hurt would be reduced.
-Never, EVER hire Bernie Stolar, he helped finish off Sega in the US.
-With a later and better Saturn, the DC wouldn't need to release in 1998, so it could have been more powerful and not quite as dated specs-wise compared to the other systems of the generation. DC maximum polygon counts particularly are limited compared to the GC, Xbox, or PS2.
-Built in DVD drive in the Dreamcast, with DVD playback either included, or via an addon like the original Xbox remote. A LOT of people bought PS2s in 2000-2001 for DVD playback alone! Match this.
If you REALLY want a successful Sega, though, Sega quite simply needed more money. Sega had limited financial resources compared to the other systems, and that is what pushed them out in the end. Even with all these things, Sega probably eventually ends up out simply because of how each generation is even more expensive to stay in than the last one. Eventually Sega's limited finances were going to give out, unless they could manage a series of smash-hit systems and manage to not immediately waste all the money, as they had done with their Genesis money. I don't know if anything could change this. Like, how could Sega get that money? What would they need to do? I guess if they could moneyhat Square and Enix they could win Japan, but Sony could offer much larger moneyhats, and Nintendo too if they believed in such things (which they don't, of course).
As for the West, canning the 32X and releasing a better Saturn would have worked wonders. How about buying exlusivity (or the studio, even?) to Core Design titles? Core and Sega were close on the SCD/Genesis, and if somehow Sega could have gotten Tomb Raider exclusive, that'd have been huge... (though Sony's moneyhats would have been larger, probably, making this tough; Sega would have needed to get them before they got big, I guess. But at that point, how do you know beforehad it'd be worth it? GEt Tomb Raider console exclusive on Saturn, though, and you help Sega in '96. Also of course... a Sonic game. X-Treme, or something from SoJ, if you force Sonic Team to make a Sonic game instead of NiGHTS. Release one, you NEED it for the US, it's not optional.
However, all of that would only help for a while -- later on console development just got so expensive! Microsoft lost over $4 billion on the original Xbox and shrugged it off as the cost of getting into the industry, while Sega was driven out of the industry with only a tiny fraction of that amount of losses. It's easy to imagine Sega doing a lot better than they did in the 5th generation. But becoming "the world's biggest gaming juggernaut"? That would require INCREDIBLE sales of Sega hardware, and I have trouble imagining a scenario where they could have overcome their massive financial disadvantage, particularly when after the mid '90s Sega's style of arcade-focused game gradually faded in popularity.