Do You Really Own Your Consumer?
If agencies in Pakistan want to elevate their game, they need to start talking to real people, argues Arshad Awan.
☰
If Only Florence Nightingale Had a Dashboard
You may be surprised, or at least I was, to find out that one of the catalysts in making data-driven decisions is a person commonly known as the founder of modern nursing, Florence Nightingale. It was Nightingale and a team of experts who leveraged statistics to promote sanitary reform in healthcare facilities in the British Army. This cascaded into a Public Health Act that established the requirements for well-built sewers, clean running water and regulated building codes.
What was the secret behind this success? It was the effective use of the humble pie chart and other graphs. Realising that few people bother reading statistics, Nightingale and her team resorted to graphics to attract attention, witty prose to engage readers and specific narratives to avoid overwhelming people. Nightingale wanted to make science more accessible to the public, including ambassadors, ministers and even Queen Victoria.
Since then, advances in computing technology have allowed us to move on from hand-drawn diagrams to printing quantitative graphs and results. In the eighties, with the advent of graphic user interfaces, a concept called ‘Executive Information System’ helped create summarised views of data for the top layers of management. With the emergence of cloud computing, organisations have access to real-time data. As someone who works with numbers, this rise in the use of data and dashboards should be exciting. Right?
Wrong. I hate dashboards. Allow me to clarify.
I mostly work in analytics related to consumer behaviour, so my work is usually linked with marketing or product teams. The issues I have with data dashboards can generally be applied across the board but really stand out in marketing-related work. For starters, measuring behaviours is a subjective and messy proposition that is open to interpretation, depending on people and their biases. This is unlike natural sciences; for instance, measuring speed only means one thing as opposed to brand equity or customer satisfaction scores.
Do not get me wrong, dashboards have their place in an organisation. They are great for getting a quick snapshot and, if built well, they allow one to process a lot of information at a glance. They are effective in driving operational decisions, as they only require a handful of performance measures to be communicated – for example, are you following the speed limit? Basically, it is much easier to ensure that one is following the speed limit from their car’s dashboard (operational decision), as opposed to understanding why high awareness may not be translating into high sales from a marketing dashboard.
This leads me to the next problem I have with dashboards. The lack of storytelling. Across the different industries I have worked in, a common sentiment is ‘this dashboard is not insightful.’ I think dashboards are shrewdly pitched as a silver bullet solution to all data needs by vendors, and most people incorrectly assume that they will somehow do the heavy lifting and uncover ‘insights’. Data visualisation is not analysis; remember, even Nightingale added witty prose to her pretty charts to build a coherent narrative. This is because dashboards are ineffective for deep analysis as they struggle to provide context.
Firstly, they are static and do not have the flexibility of filtering data through different lenses as they are built to serve a wide audience. Secondly, while charts and graphs look pretty, tables are the most effective way to understand data across cuts. Personally, I do the same as well – analyse tables and then shortlist data points that can be fed into charts. Dashboards usually refrain from tabular visualisations; otherwise, what would be the difference between those and an Excel sheet?
Furthermore, dashboards do not scale well when the number of measures increases. I have lost count of the number of times a single-page dashboard evolves into a multi-page, multi-section monster that ceases to effectively answer the first question it was designed for. Either that or the business intelligence team ends up building so many dashboards that they end up losing track of how many were built and for what purpose.
It is important that stakeholders remember that a dashboard is a means to an end, not the end itself. Generating insights is not visualising or analysing data; it usually means explaining quantitative measures with qualitative factors.
And, although it is easier said than done, the best way to do this is for analysts and end users to collaborate. The end user’s business acumen often explains the reason behind the movement of a certain indicator and the analyst can then reinforce that by looking at other numbers that validate the hypothesis.
AI is also playing a role in making dashboards more user-friendly. Many data platforms now have an ‘AI assistant’ to help identify measures that stand out. I have also seen instances where these tools summarise social media data into a few bullet points. However, end users and business stakeholders should take some time to understand the indicators and measures they are working with; for instance, how they are calculated and what are the limitations of each number.
There will never be a perfect dashboard or visualisation, in my opinion, mainly because business questions are ad hoc and require deep analysis.
My suggestion to end users is to try to pick up skills to query data from databases or become familiar with creating reports/views that serve their purpose. A trick I use as an analyst is to arrange several tables in an order that builds upon the key message I want to communicate. I then leverage cross-filtering functionality to drill down on the metric that I want to explore.
The term ‘dashboard’ has an interesting origin from the times of horse-drawn carriages. It began as a ‘plank-like’ device placed in front of the driver to protect passengers from the dirt ‘dashed up’ by the horse’s hooves. The dashboard then protected passengers from dirt kicked up by the wheels of the early motor vehicles and progressed into its current function of protecting passengers from engine heat. It finally evolved to having dials like speedometers and now touchscreens have Bluetooth connectivity. This spirit of evolution must be reflected in data dashboards, both by their users and creators.
Ans Khurram is an analytics and insights professional. anskhurram@gmail.com
Read Comments
Related Stories
I mostly work in analytics related to consumer behaviour, so my work is usually linked with marketing or product teams. The issues I have with data dashboards can generally be applied across the board but really stand out in marketing-related work. For starters, measuring behaviours is a subjective and messy proposition that is open to interpretation, depending on people and their biases. This is unlike natural sciences; for instance, measuring speed only means one thing as opposed to brand equity or customer satisfaction scores.
Do not get me wrong, dashboards have their place in an organisation. They are great for getting a quick snapshot and, if built well, they allow one to process a lot of information at a glance. They are effective in driving operational decisions, as they only require a handful of performance measures to be communicated – for example, are you following the speed limit? Basically, it is much easier to ensure that one is following the speed limit from their car’s dashboard (operational decision), as opposed to understanding why high awareness may not be translating into high sales from a marketing dashboard.
This leads me to the next problem I have with dashboards. The lack of storytelling. Across the different industries I have worked in, a common sentiment is ‘this dashboard is not insightful.’ I think dashboards are shrewdly pitched as a silver bullet solution to all data needs by vendors, and most people incorrectly assume that they will somehow do the heavy lifting and uncover ‘insights’. Data visualisation is not analysis; remember, even Nightingale added witty prose to her pretty charts to build a coherent narrative. This is because dashboards are ineffective for deep analysis as they struggle to provide context.
Firstly, they are static and do not have the flexibility of filtering data through different lenses as they are built to serve a wide audience. Secondly, while charts and graphs look pretty, tables are the most effective way to understand data across cuts. Personally, I do the same as well – analyse tables and then shortlist data points that can be fed into charts. Dashboards usually refrain from tabular visualisations; otherwise, what would be the difference between those and an Excel sheet?
Furthermore, dashboards do not scale well when the number of measures increases. I have lost count of the number of times a single-page dashboard evolves into a multi-page, multi-section monster that ceases to effectively answer the first question it was designed for. Either that or the business intelligence team ends up building so many dashboards that they end up losing track of how many were built and for what purpose.
It is important that stakeholders remember that a dashboard is a means to an end, not the end itself. Generating insights is not visualising or analysing data; it usually means explaining quantitative measures with qualitative factors.
And, although it is easier said than done, the best way to do this is for analysts and end users to collaborate. The end user’s business acumen often explains the reason behind the movement of a certain indicator and the analyst can then reinforce that by looking at other numbers that validate the hypothesis.
AI is also playing a role in making dashboards more user-friendly. Many data platforms now have an ‘AI assistant’ to help identify measures that stand out. I have also seen instances where these tools summarise social media data into a few bullet points. However, end users and business stakeholders should take some time to understand the indicators and measures they are working with; for instance, how they are calculated and what are the limitations of each number.
There will never be a perfect dashboard or visualisation, in my opinion, mainly because business questions are ad hoc and require deep analysis.
My suggestion to end users is to try to pick up skills to query data from databases or become familiar with creating reports/views that serve their purpose. A trick I use as an analyst is to arrange several tables in an order that builds upon the key message I want to communicate. I then leverage cross-filtering functionality to drill down on the metric that I want to explore.
The term ‘dashboard’ has an interesting origin from the times of horse-drawn carriages. It began as a ‘plank-like’ device placed in front of the driver to protect passengers from the dirt ‘dashed up’ by the horse’s hooves. The dashboard then protected passengers from dirt kicked up by the wheels of the early motor vehicles and progressed into its current function of protecting passengers from engine heat. It finally evolved to having dials like speedometers and now touchscreens have Bluetooth connectivity. This spirit of evolution must be reflected in data dashboards, both by their users and creators.
Ans Khurram is an analytics and insights professional. anskhurram@gmail.com
Read Comments
Related Stories
And, although it is easier said than done, the best way to do this is for analysts and end users to collaborate. The end user’s business acumen often explains the reason behind the movement of a certain indicator and the analyst can then reinforce that by looking at other numbers that validate the hypothesis.
AI is also playing a role in making dashboards more user-friendly. Many data platforms now have an ‘AI assistant’ to help identify measures that stand out. I have also seen instances where these tools summarise social media data into a few bullet points. However, end users and business stakeholders should take some time to understand the indicators and measures they are working with; for instance, how they are calculated and what are the limitations of each number.
There will never be a perfect dashboard or visualisation, in my opinion, mainly because business questions are ad hoc and require deep analysis.
My suggestion to end users is to try to pick up skills to query data from databases or become familiar with creating reports/views that serve their purpose. A trick I use as an analyst is to arrange several tables in an order that builds upon the key message I want to communicate. I then leverage cross-filtering functionality to drill down on the metric that I want to explore.
The term ‘dashboard’ has an interesting origin from the times of horse-drawn carriages. It began as a ‘plank-like’ device placed in front of the driver to protect passengers from the dirt ‘dashed up’ by the horse’s hooves. The dashboard then protected passengers from dirt kicked up by the wheels of the early motor vehicles and progressed into its current function of protecting passengers from engine heat. It finally evolved to having dials like speedometers and now touchscreens have Bluetooth connectivity. This spirit of evolution must be reflected in data dashboards, both by their users and creators.
Ans Khurram is an analytics and insights professional. anskhurram@gmail.com
Read Comments
Related Stories
My suggestion to end users is to try to pick up skills to query data from databases or become familiar with creating reports/views that serve their purpose. A trick I use as an analyst is to arrange several tables in an order that builds upon the key message I want to communicate. I then leverage cross-filtering functionality to drill down on the metric that I want to explore.
The term ‘dashboard’ has an interesting origin from the times of horse-drawn carriages. It began as a ‘plank-like’ device placed in front of the driver to protect passengers from the dirt ‘dashed up’ by the horse’s hooves. The dashboard then protected passengers from dirt kicked up by the wheels of the early motor vehicles and progressed into its current function of protecting passengers from engine heat. It finally evolved to having dials like speedometers and now touchscreens have Bluetooth connectivity. This spirit of evolution must be reflected in data dashboards, both by their users and creators.
Ans Khurram is an analytics and insights professional. anskhurram@gmail.com