We use a large grid of photoionization models that are representative of observed planetary nebulae (PNe) to derive ionization correction factors (ICFs) for sodium, potassium, and calcium. In addition to the analytical expressions of the ICFs, we provide the range of validity where the ICFs can be safely used and an estimate of the typical uncertainties associated with the ICFs. We improved the previous ICFs for calcium and potassium in the literature and suggest for the first time an ICF for sodium. We tested our ICFs with a sample of 39 PNe with emission lines of some ion of these elements. No obvious trend is found between the derived abundances and the degree of ionization, suggesting that our ICFs do not seem to be introducing an artificial bias in the results. The abundances found in the studied PNe range from -2.88$^{+0.21}_{-0.22}$ to -2.09$\pm$0.21 in log(Na/O), from -4.20$^{+0.31}_{-0.45}$ to -3.05$^{+0.26}_{-0.47}$ in log(K/O), and from -3.71$^{+0.41}_{-0.34}$ to -1.57$^{+0.33}_{-0.47}$ in log(Ca/O). These numbers imply that some of the studied PNe have up to 65, 75, or 95 per cent of their Na, K, and/or Ca atoms condensed into dust grains, respectively. As expected, the highest depletions are found for calcium which is the element with the highest condensation temperature.