Switching from ArcGIS to QGIS (and a bit of R too)

I have been using ArcGIS for longer than I care to admit. I started with it during my postgraduate years, and for a long time it was just the GIS software, the one everyone used, the one you learned if you wanted to do spatial analysis seriously. Our university has an enterprise license, so it has always been available, and old habits die hard.

But lately, I find myself opening it less and less. And when I do, there is usually a nagging feeling that I could be doing this in QGIS instead.

Where this actually started

The turning point came while I was deep in the development of Regional Explorer, a spatial data tool I built to help users explore regional indicators across Wales. I needed to do a lot of map work: clipping boundaries, joining attribute data, styling layers, and exporting outputs. At some point, I switched to QGIS for a chunk of the workflow and just… never went back to ArcGIS for that project.

[Embed: Regional Explorer interactive tool]

What I noticed wasn’t one dramatic feature. It was more that QGIS got out of my way. The interface felt less cluttered, things were where I expected them to be, and the plugin ecosystem meant I wasn’t hunting around Esri’s licensing portal to access functionality I needed.

What QGIS does differently (and often better)

The plugin library is what I noticed first. The QGIS Plugin Manager gives you access to hundreds of community-built tools. During Regional Explorer, I used plugins for cartographic grid generation, automated map atlas production, and data validation — all free, all installable in under a minute. In ArcGIS, some of that functionality sits behind additional extension licenses.

Layer styling is also more intuitive. The QGIS Layer Styling panel updates the map in real time as you make changes. You can see your colour ramp adjustments live without clicking Apply repeatedly. It sounds minor, but when you’re iterating on a choropleth map across 20-odd local authority boundaries, that responsiveness saves real time.

The print layout does the same job as ArcGIS Pro’s, with less friction. ArcGIS Pro’s layout interface has always felt clunky to me — lots of right-clicking, awkward alignment. QGIS lets me produce atlas-style outputs (one map per geography, automatically) using built-in functionality with no additional extensions required.

It runs on Windows, Mac, and Linux, and it’s free. For anyone working with collaborators outside an institution, or students who lose software access after graduation, that matters. The enterprise ArcGIS license we have is great for us, but it doesn’t follow anyone home.

The Python console is also worth mentioning. Both platforms support Python scripting, but QGIS’s PyQGIS environment feels more accessible. The documentation has improved substantially, and being open source means you can actually dig into how things work rather than guessing at what an Esri function is doing internally.

R has quietly taken some of that workload too

A meaningful chunk of what I used to do in ArcGIS I now do in R.

For Regional Explorer specifically, the spatial autocorrelation analysis was done entirely in R using the spdep package. Calculating Moran’s I, running local indicators of spatial association (LISA), producing significance maps — all in R, not a GIS desktop application. The sf package handles spatial data manipulation cleanly, and tmap produces publication-quality maps that I can integrate directly into R Markdown reports.

The advantage here is reproducibility. A script is a record of every decision I made. Anyone can run it and get the same output. That’s harder to guarantee when the workflow involves clicking through menus in a GUI application.

leaflet in R also handles the interactive mapping side of Regional Explorer, which brings me back to the point: by the time I had finished that project, ArcGIS hadn’t appeared in the workflow at all.

Is ArcGIS going away entirely?

Probably not completely, no. There are still things it does well: the network analysis tools, the geodatabase format for certain enterprise workflows, and the 3D scene viewer. For very large raster analysis tasks, ArcGIS Pro with background geoprocessing can be fast.

But “we have a license so we use it” is a weaker argument than it used to be. QGIS 3.x is not the QGIS of ten years ago. It’s capable, actively developed, and the gap with ArcGIS has closed considerably in the areas that matter most for the kind of regional and socioeconomic spatial work I do.

If you’re still defaulting to ArcGIS out of habit, it’s worth spending a week or two trying QGIS on a real project. Not a tutorial — an actual piece of work you need to finish. That’s the only way to find out whether it fits your workflow.

It fit mine. And these days, that’s where I start.