I am currently working on the predictive modeling task. I have 5 years revenue and turnover data by account level and each account is categorized by industry (sector specific) and I am interested in predicting which customer account revenue go nil ( or certain lower threshold which is categorized as prone to disaster internally). Each sector has different level of revenue and turnover. I am interested in predicting revenue (continuous output) over next 12 month.
I would attempt modeling this type of revenue data as a time-series problem.
Since you assume some sectors will go nil, you effectively assume the data is not stationary(average changes across time).
Graph some of the sectors and see if seasonality patterns appear(higher revenue at the end of quarters/weekends), to my experience economic data usually have these.
Once you decide about you seasonality and stationary effects, look at the ARIMA model. Its quite simple, and of a quick brief the first google result has a nice tutorial for it(Python).
I would model each account separately, and use the average predicted change as a benchmark for evaluating risky accounts.