Are you tired of recieving geodata with unacceptably high quality? No? Here is my go-to list for ensuring high quality for the vector geodata in my GIS projects.

Miscellaneous

• File format – What is the most suitable file format for the users of the data set? Perhaps SHP, GeoPackage, GeoJSON or DXF/DWG? For GIS, I usually pick GeoPackage, since it´s versatile and widely used.
• Coordinate system – Have you picked the right EPSG code?
• Data encoding – Ignore it and watch your data stab you in the back.
• Is the data up to date?
• Is the data documentation available?
• Describe the data license. Is it CC0 or for project use only?

Attribute fields

• Does the data use the correct set of attribute namnes? Remove unused/unwanted columns.
• Correct data types for the attributes? You probably don´t want to store integers as text.
• Which language should be used? English, Urdu or Norwegian?
• Avoid special characters in field names. No Swedish å/ä/ö for me.
• Should capital or lower case attribute names be used? This might be important for databases, such as PostGIS, which prefers lower case.
• If possible, include a commonly used field, making it easier to link/join the data with other data sets

Attribute values

• Which language is allowed for the data?
• Do all of the feaures contain data for the attributes used?
• Is the attribute information correct? A common mistake is to use outdated or semi-correct address information.
• How do you handle NULL data? Make sure it is consistent.

Geometries

• Geometry type – Have you picked the right geometry type for your data?
• Spatial coverage – Does the data set cover the whole area of interest?
• Resolution/generalization – What scale is your data intended for? Low resoultion presentations or a highly detailed map? 1: 10 000 or 1 : 1 000 000?
• Are the geometries valid? Did they pass an automated test?
• Do lines snap where they should?
• 2D/3D – Are you unknowingly distributing your 2D data as 3D?

This list will help you correct the most common errors. Many of these tests can be performed in an automated way, such as a FME workbench, QGIS model or an Python/R script.