Spreadsheets full of coordinates are the starting point of countless GIS projects: survey results, store locations, sensor readings, delivery addresses turned into lat/lon. QGIS turns that flat table into a true spatial layer in a few steps, and the whole workflow is worth learning by heart because you will repeat it constantly.

Preparation decides whether the import succeeds. Your CSV needs clean column headers with no spaces or special characters, one column for latitude and one for longitude in decimal degrees, and no thousand separators or degree symbols in the cells. Watch for text-formatted numbers — spreadsheet software loves to store coordinates as text, and QGIS may refuse to parse them. Save as UTF-8 CSV to avoid encoding surprises with non-English place names.

In QGIS, go to Layer, then Add Layer, then Add Delimited Text Layer. Browse to your CSV, confirm the delimiter, and in the geometry section set the X field to your longitude column and the Y field to your latitude column. This ordering trips up beginners: X is longitude (east-west), Y is latitude (north-south). Set the Geometry CRS to EPSG:4326 (WGS84), which is what decimal-degree GPS-style coordinates almost always use.

Click Add and your points appear on the canvas. At this stage the layer is temporary — it still points at the CSV file. To make it permanent, right-click the layer, choose Export, then Save Features As, and select ESRI Shapefile (or GeoPackage) as the format. Give it a file name and location, double-check the CRS, and click OK. You now have a real GIS layer with a geometry column and an attribute table.

Verify the result before doing anything else. Zoom to the layer extent and compare the point pattern against a basemap: a cluster of points in the ocean usually means X and Y were swapped or the CRS was wrong. This is where adding an OpenStreetMap basemap pays for itself as a five-second sanity check.

Two warnings travel with this workflow. First, remember the shapefile's attribute limits — long column names will be truncated and long text cut off, so consider exporting to GeoPackage instead. Second, a CSV point layer is only as good as its coordinates: scan for obvious outliers like a latitude of 900 or a missing minus sign, because one bad row can silently shift your entire analysis.