OK, so there is a language barrier here. I understand from your other thread that you are using SQL Server in Spanish, which means that you have to make a guess on what the word is in English. The word is Shrink, not Reduce. (And to make it clear, this is the option which just above Back Up, right)
Comparing shrink to compressing with Rar is like comparing apples to oranges. Shrink is not a compression operation at all. Shrink moves allocated extents of 64 KB to the left side of the file, and unallocated extents to the right side, so to speak. Then it releases the part of the file that only has unallocated extents back to the file system.
If you delete data in table that may lead to extents being deallocated, but it does not always happen. Particularly in heaps (that is, tables without a clustered index) this is common. For maximum effect, it may be a good idea to run ALTER INDEX ALL ON tbl after the purge to minimise the amount of pages with unused space. For a heap, you also need to run ALTER TABLE tbl REBUILD. Once you have done this, you can try the shrink again.
But the size will still be way higher than you achieve with RAR, because, as I said, Shrink is not a compression operation. SQL Server does offer a couple of compression options, ROW, PAGE and two flavours of COLUMNSTORE. However, there is always a tradeoff. The more compressed data is, the more CPU power is needed to access it.
Yet an alternative is to use the compress() function on individual columns, so if you have long columns with a lot of text, you can save space that way. But this also mean that you will need to use the uncompress() function when you access the data.