The present-day population of supermassive black holes in low-mass galaxies offers a window into massive black hole formation in the early universe. While we cannot yet observe the formation of "black hole seeds" at high redshift, the fraction of small galaxies that host a supermassive black hole -- and the properties of those black holes -- are thought to depend on the mechanism by which these they form. However, black holes in the smallest galaxies can be difficult to find, requiring creative new approaches. I will discuss recent work showing that long-term optical photometric variability in low-mass galaxies can identify active galactic nuclei that are missed by other selection techniques. I will present an analysis of the nuclear variability of more than 70,000 nearby galaxies using data from Stripe 82 and Palomar Transient Factory and discuss our sample of low-mass, variability-selected AGNs. Using this sample, we are also able to place meaningful constraints on the present-day black hole occupation fraction at low galaxy stellar masses.