Hm-m-m, what looked like a pretty straightforward question actually turned out to be a quite interesting research. There are actually many different ways to reduce the file size and it's very dependent on which file format you use and for what purpose. My guess is that you are trying to reduce the file size for .PNG or .JPG file for posting your comic page so it should be a pretty straightforward process.
In general, when you save .PNG files they have a compression option that lets you reduce the file size without losing quality (as PNG compression is lossless by default) and there are plenty of programs that can reduce .PNG even further by removing some extra data from it – for example, it looks like that PNGOUT plug-in is quite widespread and can be used within several different programs like XNView or IrfanView. By the way, both of these programs can batch convert images so even if there are several files to convert, it shouldn’t be a problem.
If you are going to use .JPG format as an output, there are also several options you can fiddle with and besides the usual “Optimize” and "Quality" options it looks like you can also choose different sub-sampling methods where some can yield smaller file but with more compression artifacts while other methods can yield bigger file with less artifacts. I think it’s up to you to find out which options are available in your software, whether you want to use additional programs or not and which options will give you best results.
If your question was about using other formats for other purposes like storing your working files, it all depends on the format itself and the general usage of that file – you can flatten unnecessary layers, use formats that can compress data and so on, but again, it’s pretty much up to you to find out the best possible solution for your working pipeline.
And by the way, resizing the image, while being the most obvious solution, also has a couple of things to keep track of: since there are various algorithms for upscaling and downscaling, in some cases it makes sense to fiddle with options and see which one has the best output. As far as I can see, most people seem to agree that Lanczos yields best results in most standard cases, but different programs can have different resampling algorithms so it’s hard to tell which one is the best one.
And the last but not least – I’ve heard that some people prefer to add text bubbles after the page was resized to its output size so text won’t be affected by compression and downscaling algorithms.