Our concern is only at species level, but we are interested in the amount of conservation that happens at protein level across species when compared with gene level
there are many commonly used genes for bacteria taxonomy: gyrB, recA, rpoD, etc. Some of them were used for many genera (like gyrB), others - just for one genus. It is recommended to deal with several house-keeping genes (MLST/MLSA), located along the whole chromosome, because the gene flow between species was shown for some bacteria, and use of one gene may be ambiguous.
In general it is good to use single copy genes, or genes that have been shown to be single copy across multiple species. See here for an example, http://www.ncbi.nlm.nih.gov/pubmed/24950957. I agree with Alex, housekeeping genes are something to look into because there would be lots of data out there for you to mine. It is also good to use nuclear and mitochondrial or chloroplast genes. Different parts of the genome are inherited differently and have different mutation rates, so they provide different time scales of evolution.
In addition to the suggestions above...you need to see which genes will have divergence levels similar to the taxa of your interest. Genes useful in other taxa may not be appropriate for your taxa of interest...candidate genes with rate of divergence too low or too high than your taxa of interest may not serve the purpose....It would be advisable to do some bioinformatic analysis of available candidate genes in for divergence rates among the organism of interest.