Machinery relies on lubrication to regulate friction and wear at contacting interfaces. Aslubricants become less viscous to save energy and cost, and as new technologies like electric vehicles operate in harsher conditions, the risk of surface-initiated failure grows. We show that metal oxide nanocrystals (NCs) dispersed in lubricants form protective coatings, or tribofilms, in situ at contacting interfaces via tribosintering. Compared to state-of-the- art antiwear additives and surface coatings, metal oxide tribofilms have several advantages, but the lack of mechanistic knowledge about the growth and wear of these tribofilms hinders adoption of NCs in lubricants. We reveal mechanisms behind metal oxide tribofilm formation through macro-scale, application-relevant experiments. We use a benchtop tribometer capable of commercially relevant conditions – the mini-tractionmachine (MTM) with spacer layer imaging (SLIM) – to refine experimental study of metal oxide tribofilm formation. First, we consider the nanocrystals as anti-wear additives, demon- strating that the SLIM technique quantifies local wear events during tribofilm growth. The competition between wear and growth can be tuned by, e.g., the addition of S- and P-based co-additives, which increase the initial rate of tribofilm growth while contributing to a more polished steady-state tribofilm. We then use the nanocrystals to form in situ metal oxide surface coatings at low homologous temperatures, showing that ZrO2, TiO2, and BaTiO3 all form durable, anti-scuffing coatings and offer robust practical benefits in realistic conditions. We then generalize an under-appreciated feature of SLIM, disaggregating tribofilm thickness measurements to correlate pixels of data with local contact stress and time in contact. Local tribofilm thickness variations at a given time are best explained by differences in exposure to interfacial sliding, not differing contact stresses – an unacknowledged driver of tribosinter- ing. We then show that a property-dependent critical length scale governing the transition from mild atomic-scale wear to debris generation can explain the differing wear behavior of ZrO2 and TiO2 tribofilms, providing a physically-motivated design criterion. Finally, we dis- cuss data-driven modeling of tribofilm formation, demonstrating unique benefits including correction of corrupted data, and a process to test analytical models.