Ditch the red herrings -- they never help and players will create their own, you don't have to help.
Build the investigation like a five room dungeon. Use one of the more linear shapes so that each clue leads to the next. The three ways to find a clue is important, but also allow for spontaneous play to resolve a clue. You can make each 'room' of the dungeon/mystery a challenge that needs to be overcome to receive the clue. Each clue should point clearly to the next 'room', but not answer the mystery on it's own.
For that last, you have to introduce multiple possible bad guys. For instance, show 10 possibles, three men in red clothing, two men in black clothing, three women in red clothing, two in black. The first clue is a scrap of red clothing and a note about a rendezvous, which is in a back alley where the players are jumped by a gang before finding a satchel. The second clue is a tube of liptstick with a bottle of expensive brandy from an exclusive club. The players have to talk their way in (or buy or sneak) to search the place/talk to the clientelle, etc about which of the three women in red clothes prefers that color lipstick and that brandy. The final clue (at the den) is information on which exact BBEG it is and where to find them, leading to the final confrontation. You can put in a side room anywhere that either earns the players some help in the final confrontation, offers assistance if they get lost (this is a great device if you think the players might go off the rails, let them know there's a one time help, like an oracle that will answer a question or a friendly sage that can offer advice), or, if you insist, is the outcome of a red herring. If you use the herring, it needs to be quickly resolved as false and immediately point back in the right direction.
If you build the mystery so that it unfolds this way, with each clue both offering one crucial but not deterministic part of the puzzle AND also leading directly to the next clue, you give the players something to discuss and mull on but also a clear path through the investigation so they don't bog down. Bogging down is the biggest risk you face in this style adventure.
Now, all of this is advice geared towards your Con game needs -- personally, I build more organic investigations that take longer, but I'm not limited to a four hour slot in my home game. With the constraints you provide, you need to put in some rails. Let the adventure be about how they overcome the challenges to get the clues and not finding lots of clues and then detectiving which are good and which aren't and solving the riddle. That makes for a great TV show, but a bad Con game.
Finally, because you will have that party that just guesses and goes for one of the possible bad guys, you need to either be ready to let the adventure end quickly in success or failure, or consider how to prevent the players from accessing the BBEG without finding most of the clues. In a non-con game, the repercussions of failure usually will suffice, but in a Con game, maybe think of adding pieces of a key or other thing they have to overcome prior to assaulting the BBEG as part of the clues. They have to collect at least 2 of the 3 before they can skip to the end because they just cannot get to the end without those clues. Again, rails aren't necessarily bad for a con game if you allow plenty of leeway in how the actual challenges are overcome.
