Back in Blog Post #1, together we began the journey through this blog to assess the Key Performance Indicators (KPIs) that we labeled the KPI Assessment (KPIA) process. As the end of the annual performance period nears, an analysis team is like a fire battalion getting ready for the next emergency call. The analysis team will do whatever it can to prepare to complete an analysis in as timely a manner as possible while preserving the highest levels of accuracy. The last blog article outlined some considerations for calculating the quarterly values for each annual KPI. The question remains, how can you assess fourth quarter results and generate the final KPI value before the fourth quarter ends? The answer is quite simple: you can’t. One of the wisest people I know says “focus on what you can, not what you can’t.” In that spirit, although it is not possible to calculate the fourth quarter and final KPI values at this time, you can project what they will be with a high sense of reliability. Just before I began to write this blog article, I watched the election returns occur for the special senate election that in Alabama. I wasn’t watching commentary on a news station, or listening to the radio, instead I was watching realtime updates on the New York Times’s Election 2017 returns website. The data used on the site to allow for comprehensive monitoring was awesome. The data displayed was not only representative of what was happening at the moment, but there were various tools that the observer could use to interpret the validity of the votes that had been processed up to that moment. What you will notice when visiting the site is that there are more widgets to monitor the validity of the data than there are for analysis of the data itself. These verification tools are what make the site especially awesome. I viewed the site for more than an hour (beginning at 9:30 PM to about 10:45 PM (EST)). It was really interesting to watch the variation in the confidence of the data at any one moment. The site registered an update every minute whether new results had come in or not. Once 90% of the votes were counted the data verification widgets had little fluctuation, and shortly thereafter the race was called. I bring up the New York Times election result site as a point of reference to think about when calculating fourth quarter values before the end of the fourth quarter. Actually, given that the fourth quarter is not yet completed, you won’t be calculating fourth quarter results, rather you will be projecting the anticipated value. Projections are nothing to take lightly. They can be very dangerous if interpreted incorrectly and are subject to change. The main purpose of a projection is to give you an idea about the direction that something is headed toward. Is the value going to me more or less likely X? When making a projection as to where the fourth quarter may end up, the quality of the data set should be in better shape than real time election results pouring in. Real time data analysis has to factor in a level of uncertainty that is acceptable. Given how close the race was in the December 2017 Alabama special election, 90% of the votes needed to be in for teams of analysts to be comfortable with making a call about a winner before the remaining votes were tallied. Unlike processing a data set that is changing in real time, examining a data set that is still active, yet with the majority of the data collected, the levels of uncertainty should be less. With only 15 days or so until the end of the fourth quarter, out of a total of 92 possible days (84% of the quarter has been completed),[i] there is sufficient data to make an informed projection for the fourth quarter. Fourth Quarter Projections Having sufficient data does not mean that a projection is always a good thing to use. A projection is not a guess. As defined by Merriam Webster[i], a projection (as related to statistical analysis) is “an estimate of future possibilities based on a current trend.” Whereas a guess[ii] is defined “to arrive at a correct conclusion about by conjecture, chance, or intuition.” One is based on a series of facts from the past that estimate the future, whereas the other is based on conjecture and intuition, a true chance. Therefore, projections require methodology and past evidence. When projecting the values of the fourth quarter to best prepare timely evaluative analysis there are four things to check for:
Please note that this can be a tricky situation. By considering that enough time has passed also assumes that data has been readily collected. If data has not been collected, then the validity of the projection will be lower because it will solely rely on past performance data.
Once the fourth quarter KPI values are calculated and subsequently the yearend KPI, the team is ready to identify the limitations, deductions, completeness, and special circumstances. Both a limitation and special circumstance for these values is that they are projections. Like quarters one through three, each team member should signoff on the projected value for the fourth quarter and for the year as a whole signifying that the prescribed business rules were followed, the values were cross referenced, and notes relevant to the KPI value(s) are recorded correctly.
With these projections in place, the team can meet to do one final data quality check. The steps involved, and the next phase will be discussed in the next blog article #14. Blog #13 Question: Is it ok to project? Blog #14 Sneak Peek: The KPI Assessment wrapup. Resources as cited: [i]https://www.merriamwebster.com/dictionary/projection retrieved on December 13, 2017 [ii] https://www.merriamwebster.com/dictionary/projection retrieved on December 13, 2017 [i] Even when considering business days only, out of a total of 63 business days in the fourth quarter and 9 remaining 85% of the fourth quarter has been completed. The one percentage point difference when compared to all possible days is not significant.
0 Comments
Leave a Reply. 
AuthorThis blog is written by our founder/principal consultant Dr. Brandan Keaveny. Learn more about Dr. Keaveny here. ArchivesCategories
All

© 2018 Data Ethics LLC

