The three key steps that a data architect should take when merging data from multiple systems into Salesforce are:
Analyze each system's data model and perform gap analysis. This step involves understanding the structure and meaning of the data in each system, identifying the common and unique data elements, and mapping the data fields between the systems. This step also involves assessing the quality and consistency of the data, and identifying any data cleansing or transformation needs.
Utilize an ETL tool to merge, transform, and de-duplicate data. This step involves using an ETL tool to connect to the source systems, extract the data, apply any data transformations or validations, and load the data into Salesforce. This step also involves applying de-duplication rules or algorithms to avoid creating duplicate records in Salesforce.
Work with stakeholders to define record and field survivorship rules. This step involves collaborating with the business users and owners of the data to determine which records and fields should be retained or overwritten in case of conflicts or discrepancies. This step also involves defining the criteria and logic for record and field survivorship, and implementing them in the ETL tool or in Salesforce.
Creating new fields to store additional values from all the systems is not a key step, but rather a possible outcome of the gap analysis. It may not be necessary or desirable to create new fields for every value from every system, as it may result in redundant or irrelevant data. Installing a 3rd party AppExchange tool to handle the merger is not a key step, but rather a possible option for choosing an ETL tool. It may not be the best option depending on the requirements, budget, and preferences of the organization.