Monitor the rank correlations from t to t+1. If the correlation is perfect (or really, really high), there's likely a data problem (e.g. the data aren't updating). If the correlation is relatively low, then there could be a problem too.
Look at the distribution of correlations to see what's unusual and throw a flag for it.