Every techno producer I know has gone through the same phase: two or three reverbs stacked on a single bus, all of them sending the same kick tail into the same cathedral, and a mix that sounds like it’s being played from the bottom of a well. The problem is almost never the reverb plugin. It’s where you put it.
Reverb on an insert treats the effect as part of the channel strip. Reverb on a send treats it as a separate space you’re injecting signal into. The two approaches produce completely different mixes, and most bedroom producers who complain about “muddy reverb” are using the first when they should be using the second. Here’s how I think about the split, and where I’ve stopped second-guessing myself.
What an insert reverb actually does
An insert reverb means the dry signal hits the reverb first, then everything goes through the fader together. That has one big consequence: you can’t pull the wet back without also losing the dry.
This is why insert reverb works on exactly two things in electronic music. First, individual drum hits that you want to sound like one object: a snare with a short room baked in, a clap with a tiny plate, a rimshot whose tail is part of the groove. Second, single sound design elements where the reverb isn’t “space” but a component of the timbre. A pad through Valhalla VintageVerb with a long decay is not a pad in a room. It is a pad. There’s no dry version you want to protect.
Insert reverb is also useful for one specific technical trick: when you want the reverb to be squashed together with the source by a bus compressor, so the tail is pumping with the original signal. That only works if they share a fader.
What insert reverb is terrible at is being a shared space. If you stick an identical reverb on four different channels, you now have four different reverbs with four different pre-delays and four different tails that don’t agree with each other. It doesn’t sound like one room. It sounds like four rooms stacked in a hallway.
Why sends are the default for a reason
A send reverb takes a copy of your signal, processes it on a separate bus, and returns it to the master alongside the dry channel. You still hear the original, unmuddied. You can ride the wet level independently. And you can point seven drum channels at the same reverb bus and genuinely have them sit in the same space.
This is the biggest practical argument, and it’s rarely stated clearly: a send gives you one shared reverb character that everything lives in. Your hats and your stab and your vocal chop all decay with the same tail length and the same pre-delay. That’s what “glue” actually means in a mix. It’s not a compressor thing. It’s a shared-space thing.
The second argument is CPU and control. One reverb instance with a decent pre-delay and a high-pass filter costs less than four channel reverbs doing the same job worse.
The third argument is the one most people miss: on a send, you can process the reverb after the reverb. That means EQing the return, sidechaining the return to the kick, or compressing the return to make the tail sit forward without raising the wet level. That’s where the real work happens.
The setup I actually use
Two reverb sends. That’s it. One short, one long. Adding a third usually means you’re compensating for something the first two should already handle.
Send one (short): a room or plate with a decay between 0.6 and 1.2 seconds, pre-delay around 20 to 30 milliseconds. This is for depth. Hats, percussion, snares that need to sit half a metre behind the kick. The pre-delay is doing the heavy lifting here. Without it, the reverb smears the transient and everything starts sounding soft. With it, the attack stays intact and the room appears after.
Send two (long): 2.5 to 5 seconds, usually a hall or a vintage plate. This one is for atmosphere and gets sent to sparingly. One or two elements max. It builds the sense of a larger space without filling every gap.
Both returns are high-passed. Always. A reverb return that carries energy below 200 Hz is the single fastest way to muddy a kick and a bassline, because reverb tails are dense, broadband, and they don’t duck out of the way. The low end of a reverb is almost never the reverb you wanted. It’s just rumble with a decay curve.
This is exactly the kind of problem FabFilter Pro-R addresses directly. Its six-band EQ section lets you carve below 200 Hz, tame the harshness around 3 kHz, and add air above 10 kHz without leaving the plugin, which means you can shape the reverb’s frequency response in the same window you’re setting decay. That matters on vocals and on drum returns, where the tail wants to bloom in a range that fights everything else.
Where the mud actually comes from
It’s rarely the reverb itself. It’s the low end of the reverb, and the fact that too many sources are sending to it at once.
One bass and one kick in the low-frequency range. That’s the rule, and reverb returns are not exceptions to it. If your reverb bus is receiving a full kick and a full bass, you’ve just built a third low-frequency source with a longer decay than either of them. It will fight both.
The fix is boring and it works: high-pass the send, not just the return. And be selective about what actually reaches the reverb. Hats, percussion, chord stabs, vocal one-shots. Not the kick. Not the sub. If the kick needs space, that’s a depth problem to solve with pre-delay and a short room, not by feeding it into a plate.
The second source of mud is layering. When producers discover sends, the temptation is to add three reverbs and route everything through all of them. That produces the same pile-of-sounds problem you get from stacking four detuned saws to make one bass. Fewer layers, better sounds. Three reverbs on one snare don’t make a bigger snare. They make three reverbs.
When inserts still win
Some cases where I’ll break the send rule deliberately:
- A single kick or 808 designed around a specific reverb tail. Routing it through a send means you can’t print the tail with the sound, and the interaction between the two is the point.
- A drum bus where the reverb is part of the bus processing chain, before the compressor, so the tail is being squashed along with the kit.
- A sound-design layer where the reverb is a synthesis element, like a metallic hit where the tail is the note.
Everything else goes to a send. If you’re unsure, send.
Character reverb versus clean reverb
One thing worth separating: the surgical, mix-integration reverb and the characterful, “this specific hardware unit” reverb aren’t the same tool and don’t belong on the same bus.
Pro-R is clean and precise. It’s the right choice when the reverb needs to sit in a busy mix without fighting anything. It is not the right choice if you want the sound of a specific vintage plate. For that, Valhalla VintageVerb, UAD AMS RMX16, or Relab LX480 do the job. And if you’re after real-space emulation, convolution is a different category again: LiquidSonics Reverberate 3 loads impulse responses captured from real rooms and hardware, and its Fusion IR approach lets you modulate and EQ the convolution response almost like an algorithmic reverb.
Practically: use the clean, EQ-able one as your shared send, and put the character one on an insert when a specific element needs that specific colour. Don’t run your whole drum kit through a vintage plate just because the plate sounds nice on its own.
The part nobody says out loud
There’s a lot of contradictory advice out there about reverb chains, and it’s genuinely hard to know who to trust. One person says never put delay after reverb, another says that’s the standard move. Both can be right in different contexts, which is why the rule-based approach breaks down.
The useful mental model isn’t “send or insert.” It’s “shared space or part of the sound.” If the reverb is the space the element lives in, and other elements live in that same space, it goes on a send. If the reverb is a component of the sound itself, it goes on an insert. Everything else is detail.
Two sends, both high-passed, both pre-delayed, one short and one long. That setup handles 90% of what electronic tracks actually need, and it stops you from drowning the low end in tails you didn’t ask for.