Note that the Högbom, Clark, and SDI versions of CLEAN can only properly deconvolve a region a quarter the area of your beam. If your the image and beam are the same size, which is the default behaviour of invert , you can deconvolve only a quarter of the image. For example, consider images that are 256 pixels in size (one dimension will do), with a point source located at pixel 192. A beam image centered on that point source will extend only to pixel 64, so that the full image cannot be CLEANed.
However there is a switch in invert , options=double, to make the beam four times the area (twice the linear extent) of your image. In this way, you can CLEAN almost to the edge of your image (you should make the image at least 5 pixels larger in each direction than the real emission).
In practise, you can sometimes violate the above rule that the beam must be four times the area of the CLEAN region, especially if the field is very sparse (e.g. a central source, and some simple confusing sources at a modest distance away). Generally the Högbom CLEAN will be more forgiving if you attempt to use a CLEAN region larger than a quarter the area of the beam. However there is a switch ( options=pad) in clean to improve the stability for Clark and SDI CLEANs as well.
Apart from restricting the area CLEANed to a quarter of the area of the beam, CLEAN works much better if the region being CLEANed snugly encompasses the real emission in the image. At first, you may not know where this is. However, after some quick and dirty (sic) CLEANing, you could redo it properly with a better idea of the CLEAN region. Essentially, you are giving CLEAN a priori information about the source, which better constrains the deconvolution problem.