Visualizing Data: How Data Insights Could Be About To Change

Visualizing Data

Gartner recently noted a shift away from data dashboard. Is this a trend that will continue? As businesses collect vast quantities of information from their customers and commercial partners, how can this data be visualized for insights? Is the data dashboard now not fit for purpose.

According to Gartner’s top 10 data and analytical trends of 2020, the use of analytics dashboards will decline being replaced by automated data stories over the next three-to-five years. Is visualizing data about to radically change?

The dashboard has been a cornerstone of data analytics for decades. Is this about to change, as AI, machine learning and new ways to visualize data such as VR and AR mature?

Gartner states: “Dynamic data stories with more automated and consumerized experiences will replace visual, point-and-click authoring and exploration. As a result, the amount of time users spends using predefined dashboards will decline. The shift to in-context data stories means that the most relevant insights will stream to each user based on their context, role or use. These dynamic insights leverage technologies such as augmented analytics, NLP, streaming anomaly detection and collaboration.”

To gain an insight into whether visualization using the dashboard approach is coming to an end, Silicon UK spoke with several leading experts to gain their views and opinions.

Clive McDonald, Head of Sales Engineering EMEA, Sumo Logic [CM]

Clive McDonald, Head of Sales Engineering EMEA, Sumo Logic.
Clive McDonald, Head of Sales Engineering EMEA, Sumo Logic.

Richard Pilling, Head of Data and Analytics at Cloudreach [RP]

Richard Pilling, Head of Data and Analytics at Cloudreach.
Richard Pilling, Head of Data and Analytics at Cloudreach.

David Cooper, Marketing Manager for Fountain Partnership [DC]

David Cooper, Marketing Manager for Fountain Partnership.
David Cooper, Marketing Manager for Fountain Partnership.

Krishna Subramanian, COO at Komprise [KS]

Krishna Subramanian, COO at Komprise.
Krishna Subramanian, COO at Komprise.

Alex Smith, Global Product Management Lead for iManage [AS]

Alex Smith, Global Product Management Lead for iManage.
Alex Smith, Global Product Management Lead for iManage.

Philip Miller, Co-Founder, Solidatus [PM]

Philip Miller, Co-Founder, Solidatus.
Philip Miller, Co-Founder, Solidatus.

John Tobin, Data Architect, Solidatus [JT]

 John Tobin, Data Architect, Solidatus.

John Tobin, Data Architect, Solidatus.

Tim Archer, Analytics Director and Senior Manager, TrueCue [TA]

Tim Archer, Analytics Director and Senior Manager, TrueCue.
Tim Archer, Analytics Director and Senior Manager, TrueCue.

Is Gartner right when they predict the decline of analytics dashboard in favour of more automated systems?

[CM] “One area that Gartner has called out is continuous intelligence. This builds on previous approaches to analytics and visualization by taking place in real-time. An excellent example of this would be the software that we use today – it creates data on what it is doing all the time. We can take this data and use it to find out various things, from how well our software is performing through to the impact of the changes that we make.

“For the software team, getting information on potential issues means that they can be fixed before they affect customers. For the business operations team, this information can be analyzed to see whether a change in their applications has produced the right result in terms of better performance or more customer sales. For the security team, it can show potential compliance problems or risks, and more importantly what to do about them. These recommendations can be automated to help people faster.”

[TA] “In some contexts, it may be useful to see customized datasets and metrics sent directly to the most relevant person based on their specific role or function, however, this would not be robust enough to justify what Gartner is referring to as a ‘demise of discovery-based analytics dashboards’.”

“Data-driven, analytically mature organizations are well informed and understand their business through descriptive and visual analytics. Having point-and-click discovery dashboards that can visualize complex data will continue to play a vital role in building a baseline understanding of the business’s data quality and this can be used to provide new users with training on the basics of analytics.

“As Gartner notes, there will be a gradual transition towards augmented insights that are contextually relevant, but we are some way off seeing ‘the decline of the dashboard’. We actually expect to see continued global growth in the usage of advanced analytics dashboards by businesses until total analytics ubiquity is achieved.

“Data and analytics maturity are an ongoing journey and trust isn’t gained in a short period. Whilst we may see the trend moving towards personalized, dynamic insights, we also believe that many organizations have not yet started their analytics maturity journey yet, so they will not be ready to receive automated data stories in place of standard intelligence dashboards anytime soon.”

[RP] “Something needs to change. How often do you look at a dashboard and wonder in what decade it was designed, and if there were any guidelines for usability given how inflexible they are? This is especially true when compared to modern web authoring/app creation tools. However, this is not unique to BI dashboards; all mature technologies stagnate and are then compared to new advancements in other areas.

“My biggest data visualization bugbear is CRM and ERP platform visualizations. A rare few are good, but most are built upon complex legacy technology stacks, which are too expensive for the provider to re-tool. The technical debt they have is overwhelming. Just as the era of having to install a thick client, or even often insecure and finicky browser extensions is behind us, so should the use of those cranky old visualization systems, which are slow to update and hard to work with. These factors drive people away from using dashboards and, making fewer data-led decisions.”

[DC] “It’s true that automation is rising in its prominence within data-handling. On a release Fountain Partnership did, I predicted that in the Google Ads platform, in the future, only one particular page will really be needed, whereby you simply click a button and numerous changes are made to the campaigns through Google’s automated optimizations.

“New third-party reporting software is appearing all the time, with an entire industry now based on reporting methods and the focus especially being on customizing reports to fill in data automatically. Funnel is an example of this, as well as software that links to Google Sheets, such as Data Studio and Supermetrics.

“Seeing the analytics is fine, even if it is automatically filled, but the trick is to be able to implement changes based on this data immediately. The Optimization Page in Google Ads is one of the most advanced systems that is trying to accomplish this, with Microsoft Ads’ own version not far behind. From this page, Google’s automation assesses the campaigns’ results, displays them for you and gives you suggestions to improve. Now they have also made these suggestions implementable at the click of a button. Having worked closely with Gartner on implementing more automation in their digital marketing campaigns, I can happily agree with them, that this form of automation will become the norm in the future.”

[KS] “Analytics is very powerful for the insights they deliver. How can you manage something you know nothing about? However, analytics alone is not actionable – if you have great insights but it takes a lot of effort from your team to turn that into meaningful productivity gains, then the insights are not that useful. This is why as Gartner predicts, the future is analytics-driven automation. Take data management for example. We can analyze corporate data, which is usually billions of files, and show savings of 60% or more by managing it better. But without automation, realizing those savings by manually moving billions of files around would be virtually impossible. This is why analytics-driven data management and automation is taking off.”

How will AR and VR systems become commonplace to make sense of the vast datasets businesses are now managing?

[RP] “Imagine you were looking to make a visualization interface from scratch. Much of what limits every current system today is that it’s based on a two-dimensional display technology – we’re trying to view multidimensional data so we have to flatten it down to an anaemic two dimensions. Our brains are great at understanding three-dimensional information, so why don’t we present data visualizations this way?

“This hasn’t been feasible in the past due to the limitations on headsets, but this is changing rapidly and the next five years will see the potential open up vastly. I believe that taking the step into three (and four) dimensions for visualizations will spur the next ‘killer app’. We’re definitely in a ‘visualization overhang’, and those who get there first will define the visual interface language for decades to come.”

[PM] “The ability of AR and VR systems to provide context will become more important. In the 2002 film of Philip K. Dick’s Minority Report, we are shown how this can be used to sell goods in the context of a person passing a shop – tailoring the images to the ‘biggest’ spender nearby. This is actually a very practical commercial use – in terms of presentation, bespoke results of personal analytics.

“There will be more like this, in the field we will see people used to seeing the world through an AR lens while working – everything from farming to law enforcement will change. It will democratize the use of data by getting it closer to the end-use case and the end-user. We can see a small example of this with something like an instant translator overlaying an image with translated text. The practical uses are almost limitless.”

What does a data analytics strategy look like today?

[RP] “Data strategy will become the number one business priority in due time, as the business value derived from data increases daily. How important is data to business? The best way to answer that is by asking yourself a simple question which reveals what is most important about IT systems: ‘What’s the room called where a company’s computers live?’ The infrastructure centre? Nope. It’s called the data centre for a good reason: D&A is the core of what companies do. The thing they protect the most and hold most dear. Without data, they have no business. It is the proper use of data and analytics that enables our customers to fully realize the potential of the cloud.

“Innovation at board level is often constrained by best practices that minimize risk and emphasize short-term results; this can slow or block technology transitions like a move to the cloud, which could realize huge future value. We’re increasingly seeing that effectiveness outweighs efficiency, or, as my dad would put it: “you don’t fatten a pig by weighing it all of the time”. Data strategy helps enterprises that are struggling to formulate a vision for their data and analytics activities and, want to understand the “art of the possible” of using the cloud to gain insights into their data. This is a holistic activity, not simply focus on technology – culture is a huge influencer on how successful a company can be, so we must also consider that aspect too.”

[AS] “The common strategy, which is based on the historical model of data lakes, doesn’t work any longer. The strategy needs to be about empowering people to solve problems using data. Tactically it means providing the capability to make connections between the data points to answer the questions of the business.

“To this end, establishing a robust and comprehensive ‘information architecture’ is a core component of a data analytics and visualization strategy. Organizations must determine where the data is stored/available, what data points might answer the business’ questions, are all the data points available and clean, what data points are missing and where can it be acquired, taking into account both internal and external sources such as market data, from academic institutions and so on.

“It’s worth highlighting the importance of the availability of all the data points when looking to solve a business problem. To illustrate, using 18 of the 20 data points required to solve a business problem, will almost certainly skew the findings, in turn being counterproductive to the cause. I’ve seen this approach leading to even bigger issues in organizations than the original that they set out to resolve. In other times it’s led to the failure of expensive projects to deliver any value. With information architecture in place, the next phase of the strategy is to adopt artificial intelligence and machine learning technologies to flexibly link the data points together in a semantic way.”

How will AI – and Machine Learning in particular – impact on the development of data visualization?

[DC] “Put simply, it already has; in a big way. Over the last quarter of a century, artificial intelligence has made waves in its development, already being utilized in many areas of the world, data visualization being just one of them.

“In terms of reporting what platforms like Google Analytics have done to address this, is to react to how we, as data analysts use their interface, then offering us the data in a readable format. This is just one example of the many third-party platforms for reports, some of which now have infinitely customizable interfaces that adapt to the data you need. The more we use specific data points, the more that platforms will make this easier to access.

“Certain platforms have also been learning how to use the data they present themselves. Whilst this may seem like humans are becoming obsolete in the process of applying the learnings of data, the reality is that our role is elevated. The mind-numbing and time-consuming tasks of forming data and then actioning the granular tasks that emerge have been retired to machines that can perform billions of equations in a second, compared to humans unable to do even one. Therefore, the task of compiling, presenting and making conclusions from data have been immensely sped up and are also more accurate.

“Our role has become the strategist. We create the boundaries into which these machines learn and action, we can then increase or decrease these boundaries depending on how these machines perform and where we decide to expand into.”

[RP] “This depends on one’s definition of a data scientist. Many are deep into the data, developing and working out the best algorithms to solve a problem – experimenting and discovering. This experimental approach is one of the reasons that data science and Agile do not mix well.

“Another type of data scientist – one who can understand the technology and explain it in business terms – I would think of as a storyteller, as any good communicator should be. However, this is not limited to just data scientists; anyone who works in a field where explaining their deep speciality to those who rely on their expertise, is a storyteller – or should be if they are to be effective. But where it really becomes apparent is when one has to present to a public/industry audience. Then it becomes vital to see the topic from your audience’s viewpoint – Chris Anderson’s book on speaking at TED is a great guide of putting this into practice.”

[AS] “There are two aspects to using AI. The first is for data cleansing. AI can help clean data through extraction techniques. Second is that machine learning can be used to automate the connection between the data. For instance, it’s possible to feed graph data with numerous data points and use machine learning to determine the optimal connections between the data points while also learning the impact of changing the data points on the visualization outcomes.

“Using a real-world scenario, enterprises typically have a large contract estate – from employment and IP, through to sales and all manner of commercial agreements. AI can be used to interrogate the entire landscape for say Covid-19, recession, or Brexit. To ensure accurate results, AI can be used to classify and extract the relevant information to enable the enterprise to understand foremost the data that exists in the contracts within the specific contexts.

“So, say that there are 100,000 contracts and 20 data points that are pertinent to the exercise. AI, machine learning and statistical approaches can be used to build a model, identify patterns and test for trends and outcomes. Crucially, by using relevant and varied data points in the model, the organization’s leaders – e.g. CEO, Head of Risk, Head of Legal Sales Director, etc. – can have distinct and customized views on the business. This capability is lacking in the typical analytics dashboards.”

Are data scientists now becoming data storytellers?

[AS] “Data scientists are becoming the storytellers, and so are the professionals – i.e. the tax investigators, the lawyers, the regulatory specialists, and such. The data scientists are also the ones who are developing the low-code/no-code tooling and automation to help these professionals work with varied data points and use the fluid visualization for storytelling. For example, typically, a lawyer outlines the chances of winning a litigation case to the client via a memo. But today, the capability exists that allows the lawyer to convey the same argument via more impactful visual storytelling.

“In fact, today there is growing interest in visual contracts – mainly because most contracts are utterly incomprehensible for non‐lawyers – despite attempts by lawyers to write contracts in ‘plain English’. Similarly, for a CEO, a visual depiction (storytelling) of hidden risk to the business in the enterprise’s contract landscape played out over five years is more consumable, powerful, and actionable, compared to a large volume of text. This is contrast to the analytics dashboards that are rigid and very business metric orientated.”

[JT] “As AI and ML become more accessible and more powerful, it is clear that the mystique around them needs to be cleared away and there needs to be a clear path from the inputs via some reasoning to the outputs. Many highly publicized machine learning issues where results depend on training data (facial recognition, Tesla auto-pilot incidents) show the importance of understanding the inputs, how processing works, and the limitations of the process. Data scientists must take responsibility for clearly explaining what is going on. The ‘magic box’ nature of AI and ML means data scientists must tell a clear story (and be sure it’s not a fairy tale).”

There is little doubt that data visualization is rapidly evolving the tools it uses. It may be too early for the demise of the data dashboard as we know them today. Tom Davenport, professor and data analytics though leader says: “Whenever I speak with successful analytics people— and I do that all the time—it’s usually not long before they mention the phrase ‘telling a story with data.’”

What is likely to happen is the AI, VR and AR influence how dashboards change to meet the needs of businesses that need to understand the information they have and, extract tangible, actionable insights.

Photo by from Pexels.