Two ways:
First, for my 2nd year college-level class in Evolution, I had students use iNaturalist to explore variation. I chose Iguanidae as the subject family starting with Iguana iguana as the target species…something they should have an idea of what it looks like already. Then I had them score at least 20 different observations from a city in Brazil I assigned to them for things like colors present, striping through the eye, transverse or longitudinal body stripes, height of a dorsal crest, presence of an enlarged gular scale. Because I had just finished identifying all of those I had a good idea where the variation was, but this could be done with any taxon where the instructor knows there is some variation. I then assigned them a second city in northern South America and they did the same looking for differences and similarities in these organisms. Then again at a city in Central America. Each time they consulted with other groups who had done the same for different nearby cities. This gave them a sense of what traits are in common for all Iguana iguana and what traits vary. I then had them do the same for another iguanid (Ctenosaura similis) wheret they now tried to identify which traits were specific to that family and which were general to the family Iguanidae. It worked really well and helped them understand that variation as a thing is hughly important even within species we think we know what it looks like. Given that selection as a mechanism of evolution requires variation be present, it was a useful lab exercise (though it could be done in lecture) for getting students to understand what actually varies. That is, most things vary, except for the traits that we tend to use for identifications. I ended by making the point that those traits that vary are the fodder for evolution by selection and those traits that do not are fodder for taxonomic identification.
Second, for my upper-division course in Herpetology I put together a slide show of images that show the relevant traits for identifying each taxon that week. Each slide might have 4-6 images containing 2-3 taxa that rely on specific traits to identify. I also give them a list of those traits as well. I then tell them they have to take my slide show and replace each image with an image they find on iNaturalist that illustrates the same relevant trait that I show them. This requires that they understand, oh it’s THAT stripe, or THAT scale pattern that is identifying. The upside of this is that many many images on iNat are not good at showing a good trait for identification so they have to look at many images to find the one that replaces mine image. Because they’re working in groups to do this, each group produces a slightly different slideshow by the end of it. Oh and surprise, you now have study aids produced by each group that they share and can look at to see other versions of the thing they just did. For me the whole goal is to get them to actually look at specimens closely enough and in enough abundance to start to internalize the relevant traits.