Equipping a team of citizen data scientists

By Dr. Dorian Selz, CEO, Squirro

The idea that data is now an organisation’s most valuable asset to starting to take hold. In 2017, an Economist article stated that the ‘world’s most valuable resource is no longer oil, but data’ and businesses all over the world are seeking to use technology to extract insight from the data they hold.

Yet even the world’s biggest CRM firm Salesforce has estimated that only 1% of a company’s data is used by its CRM system. As so much enterprise data is unstructured, what can be done to open up more of an organisation’s data so they can get more insight from it? The answer lies in addressing who uses it. When responsibility for data analytics lies with the data scientists within an organisation, it automatically puts restrictions on how data is used.

Dr. Dorian Selz, CEO, Squ

The true value of data only comes when each employee can benefit from the insight it delivers – the democratisation of data. Putting data in the hands of business users, or citizen data scientists, in this way is a real trend within business, and anyone within an enterprise should be able to access data and extract actionable insight from it.

To so though, they need to be equipped with the right tools. What tools are required for citizen data scientists to make best use of the data within their organisation?

Smart dashboards

A good dashboard and user interface is, in theory, one of the most straightforward things to address. Any analytics vendor should be able to provide this, but it is not always the case. Because such tools were initially designed for data analysts to use they can be complex and more difficult to manage than they could be.

With users accustomed to using tools such as Microsoft PowerPoint and Adobe Photoshop, and dashboard must really offer similar levels of user-friendliness. For a business user within a financial services (FS) firm to use data effectively, an easily navigable dashboard is an essential item.

Powerful visualisation tools

Many of us think in visual terms. A list of a bank’s highest performing branches conveys the right information, but a map of those branches, that allowed the user to zoom in and out to get more in-depth data and information on a specific branch is both more easily digestible and a much more powerful proposition.

With FS an increasingly global industry, the ability to tag data with coordinates to create a dynamic and interactive geographic map is a highly valued capability, allowing a user to demonstrate results and trends in an effective and visual way.

Effective data loading

If users across a business are going to be crunching data, running their own analyses and reports, then it stands to reason that data must loaded and ready to use whenever that user requires. This requires an effective data pipeline workflow.

Data must be loaded and enriched much quicker, and data from different sources should be processed in different and highly customisable ways, according to the requirements of that particular FS user. This will deliver more powerful results and give the user full control over the data loading workflow.

The right recommendation engine

FS is a highly competitive industry – particularly corporate FS – with a number of competitors all too ready to approach clients with deals and opportunities. But account-handlers within corporate FS are usually time-pressured and are dealing with such vast volumes of data that going through that data and sourcing leads for their clients is a time-consuming and costly exercise.

So for users to become an effective citizen data scientist, a smart recommendation engine is essential. It should look at data from internal sources (CRM systems, call notes,) as well as external (premium data sources such as Bloomberg or Thomson Reuters, public RSS news feeds) and use that to identify opportunities for a client. It should also score and ranks every opportunity and provides actionable recommendations to users so they can contact clients confidently with the best investment option for them.

Intelligent machine learning capability

Machine learning is much discussed within FS, and the key concept behind it is the ability to progress and do things better than before. Using traditional data modelling, business users in a bank could use untrained data sources and it would generate good results for them. Over time though, that performance will drop off, or certainly level out.

But what if a data source could be trained? What if multiple models could be run at once to constantly evolve and offer improved performance? This machine learning has previously been the preserve of more technical staff, but making it accessible to business users will be an important factor in boosting insight and understanding on an on-going basis.

 Cognitive search capability

Finding the right information within the multiple platforms and siloes of a major FS organisation is a major challenge for business users. Traditional search is proving unfit for purpose in the knowledge economy, and cognitive computing is becoming more widely deployed.

Cognitive search goes far beyond traditional search, by recognising context, relevancy, intent and interest to deliver vastly superior search results. Instead of the user looking for the most relevant information, it comes to them automatically by cognitively understanding their interests and matching this with the data available. With the pressure tosource information and insight quickly, cognitive search will become de rigueur for business users in FS sooner rather than later, becoming another key tool in the armoury of the citizen data scientist.

By Dr. Dorian Selz, CEO, Squirro

The idea that data is now an organisation’s most valuable asset to starting to take hold. In 2017, an Economist article stated that the ‘world’s most valuable resource is no longer oil, but data’ and businesses all over the world are seeking to use technology to extract insight from the data they hold.

Yet even the world’s biggest CRM firm Salesforce has estimated that only 1% of a company’s data is used by its CRM system. As so much enterprise data is unstructured, what can be done to open up more of an organisation’s data so they can get more insight from it? The answer lies in addressing who uses it. When responsibility for data analytics lies with the data scientists within an organisation, it automatically puts restrictions on how data is used.

Dr. Dorian Selz, CEO, Squ

The true value of data only comes when each employee can benefit from the insight it delivers – the democratisation of data. Putting data in the hands of business users, or citizen data scientists, in this way is a real trend within business, and anyone within an enterprise should be able to access data and extract actionable insight from it.

To so though, they need to be equipped with the right tools. What tools are required for citizen data scientists to make best use of the data within their organisation?

Smart dashboards

A good dashboard and user interface is, in theory, one of the most straightforward things to address. Any analytics vendor should be able to provide this, but it is not always the case. Because such tools were initially designed for data analysts to use they can be complex and more difficult to manage than they could be.

With users accustomed to using tools such as Microsoft PowerPoint and Adobe Photoshop, and dashboard must really offer similar levels of user-friendliness. For a business user within a financial services (FS) firm to use data effectively, an easily navigable dashboard is an essential item.

Powerful visualisation tools

Many of us think in visual terms. A list of a bank’s highest performing branches conveys the right information, but a map of those branches, that allowed the user to zoom in and out to get more in-depth data and information on a specific branch is both more easily digestible and a much more powerful proposition.

With FS an increasingly global industry, the ability to tag data with coordinates to create a dynamic and interactive geographic map is a highly valued capability, allowing a user to demonstrate results and trends in an effective and visual way.

Effective data loading

If users across a business are going to be crunching data, running their own analyses and reports, then it stands to reason that data must loaded and ready to use whenever that user requires. This requires an effective data pipeline workflow.

Data must be loaded and enriched much quicker, and data from different sources should be processed in different and highly customisable ways, according to the requirements of that particular FS user. This will deliver more powerful results and give the user full control over the data loading workflow.

The right recommendation engine

FS is a highly competitive industry – particularly corporate FS – with a number of competitors all too ready to approach clients with deals and opportunities. But account-handlers within corporate FS are usually time-pressured and are dealing with such vast volumes of data that going through that data and sourcing leads for their clients is a time-consuming and costly exercise.

So for users to become an effective citizen data scientist, a smart recommendation engine is essential. It should look at data from internal sources (CRM systems, call notes,) as well as external (premium data sources such as Bloomberg or Thomson Reuters, public RSS news feeds) and use that to identify opportunities for a client. It should also score and ranks every opportunity and provides actionable recommendations to users so they can contact clients confidently with the best investment option for them.

Intelligent machine learning capability

Machine learning is much discussed within FS, and the key concept behind it is the ability to progress and do things better than before. Using traditional data modelling, business users in a bank could use untrained data sources and it would generate good results for them. Over time though, that performance will drop off, or certainly level out.

But what if a data source could be trained? What if multiple models could be run at once to constantly evolve and offer improved performance? This machine learning has previously been the preserve of more technical staff, but making it accessible to business users will be an important factor in boosting insight and understanding on an on-going basis.

 Cognitive search capability

Finding the right information within the multiple platforms and siloes of a major FS organisation is a major challenge for business users. Traditional search is proving unfit for purpose in the knowledge economy, and cognitive computing is becoming more widely deployed.

Cognitive search goes far beyond traditional search, by recognising context, relevancy, intent and interest to deliver vastly superior search results. Instead of the user looking for the most relevant information, it comes to them automatically by cognitively understanding their interests and matching this with the data available. With the pressure tosource information and insight quickly, cognitive search will become de rigueur for business users in FS sooner rather than later, becoming another key tool in the armoury of the citizen data scientist.