Where do cables upgrades have the most impact?

Personally, I've found that the ICs have been the most profound improvement in my system.  I think the speaker cables would be right there also, but, most systems have SCs that are similarly priced as the PCs and ICs.  To get an equivalent jump in performance, the generally much longer SCs will need about a double the price difference to achieve the same gains.  The Cerious Technology Matrix SCs would be that type of jump for me.  The Teo ICs used as doubles to each component from source to preamp is an extremely big improvement.  I've kept the same sets of PCs for many years while I've changed the other two a few times.  

Big jump in performance of the ICs when I went to all silver cables from either Darwin or Amadi after having several Morrows, DNM, Audioquest, The change to Teo GCs brought a more organic sound that was also a real game changer in my system.  

Using a pair of Core Power Equi=Core 150's as my PCs in both the TRL DUDE preamp and as the power coming from all my sources was a very good improvement in dynamics, clarity and lower noise floor.  Other PCs have been good such as the Less Loss Original and Wywires Juice II's.
I probably will have to jump up in price to significantly beat the PCs I currently have in the system.

My speakers are VMPS RM40 BCSE MLS with most of the upgrades or the Vapor Audio Joule Black.  The amps are Nuforce Ref 9 V3 Se monos with the level 3 TDSS upgrades.  The VMPS are very detailed and transparent where the Vapor's are just a little more sophisticated and clean sounding.  The amps are just freakin' awesome with all the speakers I've tried them with.

Really depends on the whole system and you begin with the weakest links

If you split them up in digital, analog, power and speaker cables them I would start with the power section. This can trigger a debate with no right or wrong and only standpoints. 

Try it yourself. Feed at least your amp and sources with some decent powercords. I did not need to spend more than 350$ on a single power cable or block.  In general a factor 10 on the source is for me a non scientific rule. Common sense should do the math. Probably cheaper cables where out there that gave me the same results but I can’t try them all. 

In my opinion and based on my own test cases with cables diminishing returns will rise early. 

A test protocol for power-cords  that I use. 

Stick with two or three peaces of music. Listen for bloating , heavy bass  , cool and warm tones (timbre). Do some background reading on putting the devices in the correct fase.  I can’t detect in correct fase as some others do but it is easy to do. Peace of mind will also help. 

Try shielded and non shielded cables. Non shielded gave me less bloating or heavy bass and more alive music. There is no conclusion. 

Choose every time the sound you like the most.  Work your way up the chain and repeat the proces. You have to match and balance out the system. Work with what you have and try to get the most out of that fist. 

Try pro cables to. In most cases cheaper than consumer market cables and the the results can surprise you. 

 I have a mix of factory made (pro and consumer) and DIY cables. The main reasons for that is that I can test and swap cables without a big budget. You don’t need over expensive cables to get good sound. When I get bored listening to the same sound and want something else I have some cable options in my sleeve. 

Most of the time a come back to my reference point. I again know the contrast and enjoy what I have. 

I do reference with friends to check how my DIY cables sound. There is nothing telling me that DIY in the analogue section is worse than factory cables. Digital I leave to the pros. Meicord UTP is I non sponsored tip. I run it in all of my digital chain. For my digitaal section they are the end of the line UTP Cables. 

No Guaranty that my experience will be yours. 

Speaker cables wuld be my recommendation. Make sure that size and length match the amplifiers electrical performance. When this is done it ie easier to listen between the different interconnect cables. Best Mik

The shorter the better? When talking of cables, this statement is often presented as an axiom, although after several experiments it is not even a theorem for me. Using the example of intercon cables (the same manufacturer and the same model), I can say that in most cases a cable 1 or 0.8 meters long sounded better than half a meter long. I suspect that the matter is in some "wave" nuances, but I would be happy to know the expert's opinion.