ETL is not a magic wand to create a data warehouse ready to be analyzed.
You have to determine the business processes you want to model and model them. You have to gather requirements, and identify source systems as well as other one-off systems or small systems which will need to be developed to store data that "exists only in someone's head."
So you cannot farm out the ETL and say "hey contractors, let me know when you have something for us to analyze."
These are the steps I would take:
1. Interview business users to get a feel for requirements. Are there certain inefficiencies with reports in certain areas?
2. Become an expert at those business processes. Understand them.
3. Come up with a data model to support the users' requirements for analysis.
4. Identify possible source systems to support the data model, as well as data that does not exist in electronic form.
5. Interview the administrators or developers of those source systems to understand them.
6. Profile the data to determine if the quality will support your data model. You may need to work with the source system owners to modify as needed.
7. Write ETL packages to load the data model from step #3, using information you gathered from steps 4 through 6.
8. Write reports.
You can farm out some of the technical work above, provided you provide very precise requirements. Or, you can hire long-term contractors to become experts at your business processes and help with the entire design. Or you can do it yourself. The actual development of the ETL will be a small piece of the pie overall.