Hichert Partner Logo

Follow @ibcs_a

Comments since 2015-07-01

Andrej Lapajne 2017-04-21 04:47 am 0
UN 4.1 Unify scenario analyses Introduction
Just to add a small detail on markers. If you use standard marker "size" (in Excel or other tools) for markers of different shape, then you will probably get different marker areas:

All the above markers have the size of 5pt. In my opinion that is not OK, that is why we are using a correction ("equalization") of area sizes in Zebra BI like this:

So the rule "Marker size is 0,5 font size" in IBCS (at least it used to be defined like that) is not completely pixel-accurate. It would be better to define the area sizes in FS2 or pt2.

Andrej Lapajne 2017-04-21 04:32 am 1
UN 4.1 Unify scenario analyses Introduction
Here is the detailed picture:

Andrej Lapajne 2017-04-21 04:30 am 0
UN 4.1 Unify scenario analyses Introduction
Hi, I agree with Rolf, I think the "lollipops" are correct. I have noticed that some people are still using the old notation from 2013 with the markers colored in red/green (it used to be called "Needles") or even an older one, where the drop line is black and the markers are red/green (do not do that please...)

Since I was the one who suggested that this is not OK and am thus responsible for this change back in IBCS 1.0, I am uploading the original PowerPoint file with a more detailed explanation:

Andrej Lapajne 2017-04-21 04:12 am 0
Missing Topics MT 3
PS. I have struggled to post my comment. I guess it was too long or had too many links and pictures. I try again to post the original PowerPoint file (some 5 years old...) with the ideas described below:
Andrej Lapajne 2017-04-21 04:03 am 0
Missing Topics MT 3

Hi everybody!


This is an impressive proposal. However I must say I am not in favor of it for the following reasons:

1. It's too complicated (SIMPLIFY!)

2. It interferes with semantic design that has already been well accepted (red/green for variances, open/closed for AC/BU, etc.)

3. It's not intuitive. Why would "open" shapes suggest "flow" and closed shapes suggest "stock"? Why is volume round and value rectangular? After all, the coins (value) are round and the number of pcs (boxes) are rectangular...

4. It's too symbolic (I would prefer a more perceptual solution). It could work as a pictogram system where icons are placed in front of the measure name, but not inside charts.


I do like the idea of beginning, average and ending periods. Only, I think the shapes should in fact be reversed: "down" triangle for ending of the period and "up" triangle for the beginning of the period.


Let me propose for discussion a few alternative ideas (in order of importance):


1. Volume vs. Value

In most cases, the volume measures (quantities, units sold, pcs, etc.) are integer numbers, so I think we need a solution that suggests counting. Visually, this would mean repeating the same shape:

As shown in the above picture, the system works for scenarios as well. It has it's cons and it's not very practical to implement. On the other hand, it has been used quite often in datavis (most notably in the brilliant work of O. von Neurath).


More importantly, this method of repetition reduces the need for visual variables. As a consequence, we can now use the shape for even more detailed semantics, for example: "circle" = "headcount" (after all, our heads are more-or-less round) and "rectangle" = "boxes" (pcs, units sold, production, etc.):

2. Good vs. Bad (Refered to as positive, negative and neutral in the above proposal)

Good vs. bad is one of the world's oldest dichotomies. While it could arguably be presented as red vs. green, the light vs. dark or white vs. black is a much stronger archetype:

Not using light vs. black colors is in my personal opinion one of the most important opportunities currently missing in IBCS.


Which brings me to the "forbidden" topic of colors ;)


Many times before I have proposed to use lighter colors for "good" KPIs like Revenue or EBITDA and dark or completely black colors for "bad" KPIs like Expenses. In the current IBCS version, the "calculation" waterfall charts in EX 1.1 are still rendered in this way:



Unfortunately, the (same!) Profit in the very next chart (on the same page!) is rendered in darker color:



We had many discussions about this, I say it's time we fix it. The color of the same KPI simply cannot change from one chart to another.


The key argument against the use of light vs. dark colors for good/bad is the PY notation. And while I absolutely agree that using a lighter color for PY - I prefer to call this "faded" color - is a very smart idea, it does present problems. The main one is that we are using two different visual variables to code business scenarios: (1) color hue in case of PY and (2) pattern in case of AC, FC, BU. That's the original sin. If we would use only pattern for scenarios, then this problem would fade away.


An absolutely trivial solution of this problem is to use the color hue dimension for good/bad. For example using bluish or beige color for good and spot black for bad. Then you can still apply e.g, 50% lighter color for PY and still get visually distinct picture. Has this gone too far or perhaps not?


3. Ratios!

The proposal of "SEMANTIC NOTATION OF BASIC MEASURES" completely omits the ratios. But it should not, as ratios are extremely important KPI types in business. They are defined in UN 3.1 separately for column and bar charts as "1/3 of category width" and "thin lines" in line charts. I do not agree with this definition for the following reasons:

1. If possible, we should not separate the design definitions based on chart types. Rather, we should strive to find universal solution that work across different chart types.

2. It recommends the same visual design for Value/Volume ratios (e.g. "Sales per capita" or for example "Price") as well as pure Value/Value ratios (e.g. "Profit as % od sales"). This is not OK, because in the first case the unit is for example USD/person or USD/unit, while in the second case there is no unit, just a % value (e.g. 18% EBITDA margin).

3. In reality there are many quite different types of ratios in business reporting, most notably:

    (1) Structure % ("Portfolio Share"): division of an element (Part) by the Whole within a category dimension of exactly the same measure (either a volume or a value). Example: % of Net revenue of Laptops within the total Net revenue.

    (2) Pure ratio in %: Division of two different measures of the same unit. Examples: EBIT% = EBIT in USD / Net Sales in USD, Value market share in % = Sales in USD / Market size in USD), Volume market share = Sales in units / Market size in units

    (3) Volume/Value: "per unit" measures, such as prices (e.g. USD/unit), Sales per capita, EBITDA/Employee, etc. These KPIs are very different from previos two, as they have the unit, while above two do not have a unit. This, they should not look the same!

    (4) Percentage points: The difference between two percentages. These should be covered somewhere in UN 4.1., but I can't find it online at the moment. As far as I know, they are supposed to have 50% column/bar width, but in my opinion this is also not OK. To be discussed.


Here's an old idea how to differentiate between the KPIs:

This proposal still seems interesting to me:

On the left chart we have revenue per unit (box or package, that's why it's rectangular) and on the right we have revenue per "head" ;).


This system clearly distinguishes between prices (value/volume) and pure ratios (% margins):

The left chart is Gross Profit (in USD) per "head" (unit is USD/Employee) while the right one shows the Gross margin in % (Gross Profit in USD / Net sales in USD) - no unit, pure % value.



4. Stock vs. Flow

In my opinion this one is most controversial and perhaps even the least important. Why? 

Reason #1: What exactly is flow? 

Reason #2: Is "Stock" the right term?


 Arguably, we do not really have any "flow" data at all. Continuous data simply does not exist. We only have measurements in a certain point in time. The "flow" of an indicator is just an approximation of a series of discrete measurements, which is now explained in EX 1.1 under Line charts.


For practical reasons: What is the real difference between monthly values of, say, Net sales and Inventory? There is only 1 measurement per month in both cases, however the true difference in the nature of these KPIs, as defined in terms of business intelligence, is:

 - Net sales is a (fully) additive measure

 - Inventory is a semi-additive measure (aggregatable accross all dimensions except time)


The idea of using lines and areas for "flow" and discrete visuals like columns or dot-plots for stock appears compelling at first sight, but how would you visualize daily inventory vs. quarterly sales? Probably lines for daily inventory and columns for quarterly sales, right?



This will make a very interesting discussion in Barcelona. My point is that "flow vs. stock" is not a property of "continuum" vs. "point in time", but rather just a different aggregation type in the time dimension ("beginning value" vs. "end value" vs. "sum" or perhaps even "average" or "YTD"). This is why I have mentioned in the beginning that the idea of marking the beginning, avergae and ending points in time looks very promising. Can we simplify it to show small dots at the left or right side (under the chart axis) for opening/closing values, respectively?


[Of course all ratios are totally non-additive, but this is probably not relevant for "stock vs. flow" discussion.]

May the force be with you and I'm looking forward to discuss all this stuff in Barca (even though they're sadly out of the UEFA championship)!

Jürgen Faisst 2017-03-25 07:44 pm 0
UN 4.2 Unify time series analyses
Hi Beat,
I think you have to show all the years if you use "..". 
If you want to show the average over the four years from 2013 to 2016 I would suggest to use 2013_2016Ø. 
Jürgen Faisst 2017-03-25 11:46 am 0
UN 4.1 Unify scenario analyses
I would prefer a sequence representing the actuality of the scenarios from left to right:

2016 PY is in fact from 2015 and therefore the oldest one
2016 PL has been created end of 2015, which is before 2016 really started to happen
2016 FC is created while 2016 happens
2016 AC is the documentation of what has happened in 2016 after it happened

However, different views on this are exactly the reason why "No rule governs the sequence of scenarios referring to the identical time period".
Yvan Clot-Goudard 2017-03-22 01:21 pm 0
Thanks for clarifying, Jörg, point noted.
Joerg Decker 2017-03-21 08:52 pm 1
I would like to post your question to the LinkedIn Forum where you already posted another question. This place here is to discuss the standard and its changes. The LinkedIn Form is the place where all consultants meet and (hopefully) help quickly.
Jürgen Faisst 2017-03-21 12:31 pm 0
UN 4.2 Unify time series analyses

Rolf and I suggest to complete this section by adding a symbol for year-to-go in analogy to year-to-date: 


Year-to-go analyses (YTG) refer to the period from the present to the end of the (fiscal) year (YTG time span). Where helpful, visualize analyses showing YTG values by appending an underscore to the time period name, e.g. “Jun-2015_”.


© 2017 IBCS Association. Licensed under Creative Commons BY-SA 4.0 International.