There are at least two major ways that COVID-19 vaccines could falter. The first might best be described as a memory lapse, and it’s a bit of a flub on the human side: Left to its own devices, the immune system slowly loses its intellectual grasp on the pathogen, and is much less prepared the next time it sees it. The second is a mismatch between what immune cells studied and what ended up on the final exam: a mutation in the coronavirus that alters its appearance so significantly that it becomes unrecognizable, even if immune memory of the vaccine remains intact. Designing and deploying boosters requires keeping tabs on these two fast-changing variables at once. Memory lapses can, in theory, be easier to detect and repair: Researchers take blood samples from vaccinated people and track the levels of different immune actors, such as antibodies and T cells. If those levels start to dip below a crucial protective threshold, it’s time to offer a booster. This approach works well in certain boosting regimens, such as the Hepatitis B vaccine for health-care workers, But sussing out this so-called correlate of protection typically takes gobs and gobs of data. For many vaccines, even ones that have been in use for decades, such as the mumps vaccine, those numbers still aren’t clear-cut. SARS-CoV-2’s correlate remains elusive.