: Encountered abnormal data, such as sales fell 10%
First, do not rush to analyze how to how to properly think about it -
(1) This is not a problem
Everyone fell 10 percent default is a problem. But in fact I did not say what day ah? In case of normal fluctuations weekend effect is due to it.
(2) This is a big problem
If the normal volatility of 9% over the same period, 10% yesterday, considered how much a thing? This looks, but is sent up the chain closest to the truth, because there may be time to do up the chain will find: Oh, this volatility is not really so great. But this requires them to look long enough. The other two categories will not say, direct spike.
(3) they do not rule out abnormalities
If this fluctuation is large enough, it may be whether or not the data itself out of the problem? To confirm this first time, eliminate errors, holding the correct data to do analysis. Data error analysis wool?
Drill (old and new user area conversion)
(4) They did not rule out the policy implications of (competing products activities, which failed initiative to adjust, external factors (political and economic)
If this fluctuation is large enough, it may be whether the initiative to adjust? Are there relevant external policies, internal activity released? These must first do the analysis themselves understood. General volatility in the short term and special circumstances are more likely to first ask this, then consider how the analysis of specific indicators.
After ruling out these questions you look at the business on what a problem: there are problems such as lecturers or promotion is not to force the like.