Data Loader uses external Id fields to match records in the CSV file with records in Salesforce during an upsert operation. If the CSV file contains duplicate external Id values within the same file, Data Loader will throw an error saying "Duplicate Id Specified" and will not process those records. Therefore, it is important to ensure that the CSV file does not have any duplicate external Id values before using Data Loader to upsert records.

Using leading % wildcards, performing large loads and deletions, and using NOT and != operators are three options that prevent a query from being selective. These options can cause the query to scan a large number of records or indexes, which can negatively impact the performance and scalability of the query. Using a custom index on a deterministic formula field, on the other hand, can improve the query selectivity by allowing the query to use the index instead of scanning all the records6.

Analytics/BI Owners and Data Domain Stewards are appropriate roles as members of a team that makes decisions on enterprise-wide data governance. Analytics/BI Owners are responsible for defining the business requirements and metrics for data analysis and reporting, and Data Domain Stewards are responsible for defining and enforcing the data quality standards and rules for specific data domains. Salesforce Administrators and Operational Data Users are not suitable roles for this team, as they are more focused on the operational aspects of data management, such as configuration, maintenance, and usage.

PK Chunking is a feature of the Bulk API that allows splitting a large query into smaller batches based on the primary key of the object. This can improve the performance and avoid query timeouts when querying large data sets

According to the Get Started with Big Objects unit on Trailhead, one of the use cases for custom big objects is to store and manage loyalty program data for customers. The unit states that "From loyalty programs to transactions, order, and billing information, use a custom big object to keep track of every detail." Therefore, a custom big object can be used to capture the reward program data and display it on the contact record. Additionally, according to the Big Objects Implementation Guide, big objects can handle massive amounts of data (up to billions of records) and can be updated nightly from external systems using Bulk API or batch Apex. Therefore, a custom big object can meet the requirements of NTO's loyalty program scenario.