Data Preparation and Blending
Data preparation in Alteryx centers on the Input Data tool, the Data Cleansing tool, Select, and the Join family. Know the difference between a Join (inner by default, with Left/Right/Full options), a Join Multiple (two or more inputs, requires at least one common field per pair), and an Append Fields (cartesian, no key). The Union tool stacks records and configures field matching by position, name, or manual mapping; questions love the trap that "Auto by Name" silently drops fields that do not share a name across inputs.
Blending traps cluster around data types and nulls. A Join on a numeric field against a string field fails or yields mismatches; the Select tool is the fix. The Data Cleansing tool replaces nulls, removes unwanted characters, and can modify case, but it does not change field type. For fuzzy matching, the Fuzzy Match tool is distinct from Join: it generates match scores (Edit distance, Jaro, Levenshtein) and requires a threshold, and it can produce multiple candidate matches per record, so a Make Group step is often needed afterward.
Rule of thumb: before joining, always run a Select to confirm types and rename keys, and check join output counts - an unexpected record multiplication signals duplicate keys on one side. For unions, prefer "Manual" mapping when schemas differ so you can see exactly which fields merge.
Worked example: two customer files, one with CustID as Int32 and one as String, joined on CustID. The Join returns zero matches. Fix: Select on the string side to convert to Int32, then re-join; verify record counts match the smaller file's distinct key count.