Sequence profiles contain more information than for example consensus sequences. Imagine the followng sequence allignment:

ATTG[?] TAGT[?] ATGG[?]

Now the consensus sequence would be ATGG. However, by creating a profile you would say that there is 2/3 chance of an A and 1/3 chance of a T at the first place, 2/3 chance of a T and 1/3 chance of an A at the second place etc. This way you have a lot more information.

Even better and more used nowadays are hidden markov models though.

Well explained by Niek.. Profile is a better representation of information than consensus and MSA.

We can store the Profile in the form of either PSFM (Position specific frequency matrix) or PSSM (Position specific scoring matrix).

The use can be:

1) we can use this Profile (PSSM) to search for any homologous sequence. PSI- BLAST search is based on this. Here we take PSSM matrix and search against sequence database.

2) Another use can be, RPS-BLAST (Use by CDD database of NCBI): which is "Reverse Position-Specific BLAST. where you take sequence, and search against pre-calculated databases of PSSM.

One more important addition to profile is the SeqLogo diagram, which is a visual representation of any PROFILE.

Here is a very good background paper on Nature Biotechnology - "What are DNA sequence motifs?", which could resolve your query.