Over the years as our personal computers have become faster and more powerful, our databases have change from being a flat file to relational to business intelligence technology we are using today. Talk about data overload! We can slice and dice the data any which way we want so we can end up with a sculpture by Michael Angelo or end up with something a child has made from Play-Doh.
I recently went to an evening event presented by Dr Daniel L Moody, The Art (and Science) of Diagramming, understanding cognitive effectiveness of diagrams. Reporting is also like a diagram, using numbers to tell a story. Business analytics reporting has be able to tell the users the answer from just looking at the report without having to decipher the data. As the saying goes, a picture tells a thousand words 
.
As we look at our data today we are overwhelmed with the information that is given to us on a daily basis. We need to pick out particular numbers from our databases to get the correct analysis of the data. As business analysts we need to create an experience for the user that will make it enjoyable for the reader to use, to create a visual effect with the data by not using graphs 
. The user of the report has a choice, “Do I want to look at the ugly report today telling me that my numbers are bad?” or “My report is out, let’s see how that report is going to tell me how I can fix my numbers”. Of course, my preference is the later, you don’t want to be doing all this work so nobody will look at your report.
The first thing most people do when they build a new report is somebody has said to them “so and so has a great report, let’s replicate it” or migrating to a new database without enhancing any of the reports. Take the original report and modify it to accommodate new data and/or analysis even though the original report was for another purpose. The report is continually modified, adjusted for the new business rules and ends up being a dog’s breakfast that nobody wants to use.
Using a sales forecasting model as an example to tell a story for the day to the year using the same story board so the users know the story being told, it is just the journey taken will be different.
Time Series Data
Imagine time, how do we envisage time in our heads, do we look down at time, look up at time or do we look at time as an horizon going across from left to right. Is time ever broken? Does it stop when we know we have an end point to go to? There are rest breaks but we still need to get to the end point. In business, our end point is always that target number.
Before the advent of computers people had their pens, paper and then came calculator. So when creating reports, there would be a point in time where analysis needed to be done. You would have your columns to a certain point in time then add the analysis column, figure 1, therefore the time series is broken and stops the train of thought.
For our brains to take in data logically, the data needs to be shown as a straight line to the end. In figure 2, shows the time series unbroken with the analysis at the end of the total time period. Your eyes will automatically scan across at the weekly data and monthly data without stopping and starting at each new month. This allows for the data to be analysed initially as the week and then monthly then to the quarter, from small to large.
Lists
Using the analogy of a shopping list. We go around the house, opening cupboards, listing vertically on shopping list paper which is thin and long. The shopping list paper has been designed vertically as it is easier to read while shopping . As we are writing the list, there are duplicates but are not the same. We don’t write it at the end of the list, and include the new item as part of the item previously listed. We want to get the item at the same time as the other product in the same aisle.
This is the same when looking at a report, looking down the list to see what variables we have to look at to analyse the data. As with the shopping list, it is natural to not want to read duplicates. Once a set of variables have been looked at, the reader does not want to see the same words again in a list. The first thing they think is, “What is the difference?”. There is no difference in the words just the analysis in time.
An aside, you may think that dates may also be listed vertically. Lists are infinite but in business analytics time periods are fixed to a period. Therefore reports may have indefinite analysis but only fixed time periods, figure 3. There are situations when time is listed vertically. I can only think when the time variables are small intervals and long time periods and the variables list is short. eg daily tracking files and bank statements, (and we all know how much we like looking at those, not 
)
Creating the Story
Figure 4. Quarterly Sales Report
The report has been broken into 3 sections.
1. The actual data as inputted by the business.
2. Weekly analysis – Sales operations analysis for supply and demand planning, to ensure we are meeting weekly forecasts and are moving forward towards targets.
3. QTD analysis – Performance based analysis for finance to ensure sales are meeting initially forecasts and the quarter target.
Figure 5. Sales Operations Waterfall
Section 1 of the report shows a complete picture of the quarter, showing the different sources of data and specific time periods. These different sources and time periods are colour coded and follow throughout the report so the user can easily identify the sources of data or the analysis. This also allows the user to quickly glance at the data to understand the analysis.
1. When labelling, think about what you want the reader to concentrate on, the labels or the numbers. Ensure labels are easy to understand. There is already enough to read in these reports without having to read long labels, labels can be identified at the start of the columns without intruding on the numbers. Dates, create the date as a full month, 4/5/10, is this the 4th May or 5th April? It is easy enough to change the format to 4 May or Apr 5.
2. Inputs from the business – from finance for targets and initial forecasts by sales.
3. The weekly forecasts from sales by month by week and actuals. Note, if the waterfall had followed the same sequence as figure 1, we would not have a complete picture of the quarter and the waterfall. The waterfall will be a squashed picture of the quarter with missing analysis and an additional field, Figure 5a.
4. A continuous view of the quarter forecast over the 13 weeks.
Figure 6. What are we analysing?
Section 2 & 3 have the same analysis except section 2 is concentrating on the short term periods within the quarter and sales operations analysis for supply and demand planning. While section 3 is analysis for the quarter to ensure targets and initial forecasts are achieved, performance based analysis, eg commissions and business reporting periods.
1. Target vs Actuals, Target vs forecasts.
2. Forecast vs Actuals, Forecast vs previous week’s forecast.
3. To Go to Target, Initial forecast, Current forecast vs Last quarter and same quarter for last year.
Note, this list of variables could continue to include further analysis and business rules.
Figure 7. Time analysis.
As we slice the data the other way, we are now looking at the analysis in figure 6 but in relation to the past and the future.
1. What has happened
2. Where we are today
3. Where are we going and how we are going to get there
4. What has happened by month
5. Where are we going by month
6. Where are we going for the quarter
Figure 8. Sales analysis matrix for time.
Combining figure 6 and figure 7, we now have sales anlysis matrix for the quarter, where certain cells now can be ignored as time has passed and the eyes will concentrate on the data required to achieve the quarter targets.
1. Performance indicators vs Target
7. Performance indicators vs Forecast
d. Time has passed, so data can be ignored
2. Current situation vs Target
8. Current situation vs Forecast
e. Current situation vs previous quarter and same quarter last year.
3. Forecast vs target
9. WTW forecast
f. How we are going to achieve our goals, Weekly Forecast vs historical
4. Performance indicators vs Target for month
a. Performance indicators vs Forecast for month
g. Time has passed, so data can be ignored
5. Forecast vs target by month
b. WTW forecast by month
h. Month Forecast vs historical
6. Forecast vs target for quarter
c. WTW forecast for quarter
j. Quarter Forecast vs historical
Figure 9. For the Day to Year
By replicating the weekly model, we can have Monthly for Year and Daily for Week. We now have all the chapters of the story from Day to Week, Week to Quarter, Month to Year. The story is the same for each time period but the journey taken is different and for different purposes.
Figure 10. Other Dimensions.
Keeping the story the same, we can now create different journeys of our story depending on users purposes. We can now slice and dice the data any which way we like but keeping the story the same. The user now can move from journey to journey without the story changing.
Conclusion
According to Dr Daniel L Moody to achieve Cognitive Effectiveness in our diagrams we must have the following :
1. Discriminality
2. Modularity
3. Emphasis
4. Cognitive Integration
5. Perceptual Immediacy
6. Structure
7. Identification
8. Visual Effectiveness
9. Graphic Simplicity
Do I achieve this using the analogy of a report as a diagram? Dr Moody was only able to discuss point no. 1, relating to IT diagramming which doesn’t really excite me. So you tell me. Was I successful in translating my numbers into a picture that tells a story?
Recent Comments