Treasury Today Country Profiles in association with Citi

Thinking big

Corridor of a working data center

Big data has much to offer corporate treasurers – but at this early stage, most of the possible benefits have yet to be realised. Treasury Today Asia reviews the current state of affairs, the possible applications of this technology, and the pitfalls that treasurers should be aware of.

How big is big data? The short answer: not as big as it’s going to be. A report published by big data consultancy WHISHWORKS found that while only 18% of UK companies have fully adopted big data, 47% have tentatively started initiatives and 29% are actively investigating the opportunities. Meanwhile, Wikibon predicts the worldwide big data market will reach US$103bn by 2027, up from about US$40bn currently.

As far as treasurers are concerned, big data has plenty to offer. Research by the Economist Intelligence Unit on behalf of Deutsche Bank found that over half (56%) of treasurers believe big data analytics systems will be beneficial to their organisation in the future, ahead of AI/ML systems, instant payments, RPA and blockchain solutions.

Where specific areas of interest are concerned, a 2017 survey by Euromoney and J.P. Morgan asked which areas of treasury big data would impact the most. In first place was customer behaviour insights, cited by 30% of respondents, while 27% opted for cash forecasting. Other areas of interest included real-time visibility/tracking of payments (18%) and reporting (14%).

The story so far

In practice, though, most of the potential benefits are yet to be realised. “‘Big data’ is certainly a real buzz word,” comments Andrew Marshall, Managing Partner at Covarius. “But treasurers aren’t really aware how it can be utilised and how it can benefit their organisation.”

There is certainly plenty of talk around this topic. “I really think we are just scratching the surface in terms of what could be done with data in the treasury space,” notes Peter Fox, head of Data and Insights for GTS at Bank of America Merrill Lynch. “We have been involved in hundreds of conversations with treasury team members – this topic tends to come up in strategic reviews with customers and is a very common topic at our advisory boards.”

Where practical applications are concerned, progress is happening, albeit more gradually. “Where progress has been made so far by treasurers is in automating certain processes,” says Magdalena Mielcarz, EMEA Head Channel and Enterprise Services, Treasury and Trade Solutions at Citi. “So I’ve been hearing a lot from the clients about deploying robotic process automation (RPA) into processes. But when we think about big data, what we are talking about is taking this to the next level.”

“It’s about using artificial intelligence and machine learning to deliver insights, predict the future or detect certain future risks.”

Magdalena Mielcarz, EMEA Head Channel and Enterprise Services, Treasury and Trade Solutions, Citi

Mielcarz explains this is not just a question of robotising repetitive steps that are already being carried out by a human – “it’s about using artificial intelligence and machine learning to deliver insights, predict the future or detect certain future risks”.

“We think the low hanging fruit lies in the area of operational efficiency,” adds Fox. “In our interactions with customers, we have been able to see first-hand the power of sitting down with a broad set of individuals within a client’s organisation, aggregating their data into one dashboard and benchmarking against similar companies. The conversations that can lead to, and the immediate actions that customers are able to take, are frankly astounding.”

The next frontier, Fox adds, will be to look at activities that are characterised by underlying patterns. “So when you think about cash forecasting, understanding the underlying pattern of when you get money and when you pay it out is an area that is very ripe for data analysis. Fraud detection and error recognition are other areas that data analysis is very well suited to solve.”

Practical applications

Where practical applications are concerned, Mielcarz cites the example of Citi Payment Outlier Detection (CPOD), which uses AI to detect transactions that are outliers compared to clients’ historical flows, so that clients can confirm whether or not a transaction is legitimate.

“We also deploy big data analytics into our current processes at Citi,” she says. “For example, we use certain biometric-related features that can indicate suspicious activity.” This could include analysing different users’ typing patterns and judging whether the person currently using a computer is the same person who has been enabled on the system.

In addition, Mielcarz says the bank draws upon big data to generate client relationship insights and thereby help clients optimise their treasury operations. “For example, we can analyse clients’ existing cash management structures with Citi alongside external or public sources,” she says. “This enables us to advise the customer that in the context of Brexit, for example, we would make certain recommendations for an efficient structure in the future, based on the company’s account structure, currencies and flows.”

She explains that these insights could be based on different topics such as accounts structure, liquidity structure, working capital analytics or transaction flow optimisation. Benchmarking can also be deployed to provide further recommendations. “While we can’t include every type of data in light of the current regulatory framework – especially GDPR – we can still provide some benchmarking and peer analytics, comparing certain structures, flows or corridors to those of the customer’s peer group,” she comments.

Too much data?

Embracing the power of big data can bring many benefits, but treasurers also need to understand the risks. Fox notes there is a real risk of spreading a particular team’s focus too thinly across too many objectives. He adds, “There is a lot you can do with data that can be a double-edged sword – so it makes sense to find the problems you really want to focus on and resolve these.”

Treasurers should also consider some other possible pitfalls:

  • Losing focus on business problems.

    For example, Fox says that while there is a lot of attention on data infrastructure development and the use of data lakes to collect data, this should not come at the expense of focusing on new processes and new data products.

  • Losing sight of the human element.

    In addition, Fox says it can be easy to lose sight of the human element of big data. “In many cases, your new dashboard, product or analysis will change how other people do their jobs,” he says. “So it’s important to recognise that these developments bring changes, and address this via good management processes and excellent storytelling.”

  • Poor quality data.

    Data quality is another consideration. The data that companies use is of little value unless it is complete and accurate – which can be a challenge in some cases. “You have to make every effort to improve the quality of the data you are using in order to get the right outputs in the future,” notes Mielcarz. “That’s a challenge that everybody faces, which can be tackled by identifying the lower quality elements and improving quality over time.”

Meanwhile, Enrico Camerinelli, Senior Analyst at Aite Group, highlights another possible barrier to adoption: “Ironically, the companies that could benefit the most from this are the ones that may not have the budget needed to afford the necessary investments.”

Plunging into data lakes

One challenge where big data is concerned is for treasurers to be able to access data in a way that is genuinely useful. Data lakes may provide an opportunity to achieve this.

Simply put, a data lake is a repository of raw data which can provide more flexibility than a more traditional data warehouse. Andrew Marshall, Managing Partner at Covarius, observes that the “big data dream” is typically driven by IT, and that it tends to involve creating large, accessible data lakes which treasurers can use to extract data – in contrast to ERP data sources, which are typically locked down.

“This may give the treasurer visibility across historic and future data (such as invoices) running to millions of transactions, which in theory can be used in all manner of ways to provide forecasts and predictive analytics,” comments Marshall. “However, the role of IT usually stops at the data lake/reporting level. This puts serious limitations on the value of the data to the treasurer.”

More recently, Marshall says that bigger players in the TMS space are beginning to invest in bringing data across from the data lake and into their reporting and analytics tools. “This will enable some seriously impressive and powerful output, from which the treasurer can make complex decisions,” he adds.

However, one obstacle around the data lake model is that static models may mean that changes to the underlying data in the ERP system will not flow quickly across the data lake, meaning the treasurer may be using out-of-date information. Such changes could include corrections for inputting errors, updates based on trading performance and changes to credit terms.

“To combat this, TMS vendors are looking to build real-time connectivity – for example, via APIs – between the ERP, data lake and TMS such that changes to underlying data are instantly updated in the TMS, with status codes updated via messaging services to reflect the changes,” explains Marshall. “This gives the treasurer that level of confidence that the data is good to work with and never out of date.”

Where to start?

Given the potential benefits, what should treasurers be doing now to take advantage of big data?

  1. Focus on a specific business problem

    Fox notes the importance of focusing on a specific business problem in the first instance. “Pick out a moderately sized pain point and start to use different data visualisation tools and different data science tools within your own organisation to chip away at that problem,” he suggests. “I think people will be surprised at how quickly you can get momentum if you bring the tools and the people together around the common problem to try and solve it.”

  2. Translate initiatives into actionable data

    Likewise, treasurers should ensure that any big data initiatives result in actionable data. “For example, a company may always have problems with specific suppliers when they send invoices, resulting in delays in accounts payable,” says Camerinelli. “You can carry out an intelligent analysis to find out which suppliers are the best and worst performance – but then you have to translate this into actionable items that someone else has to follow through on.”

  3. Appoint a data champion

    Mielcarz, meanwhile, emphasises the importance of having the right people available to test new opportunities. “At Citi, we have data scientists in our labs – and we also identify data champions across our current teams to ideate and feedback from discussions with clients about what types of insights clients are interested in seeing.”

    She recommends that corporate treasurers should likewise appoint data champions within their organisations in order to provide insights from a data perspective.

  4. Engage with partners and the treasury community

    In addition, treasurers can engage with technology and banking partners to discuss the problems they are trying to solve and learn more about the relevant opportunities. There is also much to be gained by engaging with the treasury community and finding out how different groups are currently using data to tackle similar challenges.

Future developments

Looking forward, it will be important to monitor the impact of the evolving regulatory climate and the expectations regarding privacy and data security. While GDPR is a European regulation, other jurisdictions are beginning to follow a similar approach – California, for example, has adopted a similar law covering data privacy. “We need to be mindful of the fact that we need to keep embedding an infrastructure that makes it easy for us to comply with all those regulations, existing and upcoming,” Mielcarz notes.

Where future opportunities are concerned, another notable area of opportunity is the way in which different data sets can be combined. “From having conversations with clients and studying the industry more broadly, some of the more interesting developments in the data space have come from unique combinations of data sets that are related but distinct,” says Fox. “So that might include GPS data in the logistics space, or weather data in the farming space. It’s one thing to mine your own data, but being able to combine your data with related third-party data could lead to tremendous insights.”

Fox says that marrying bank data with a company’s internal accounts receivable data, for example – and potentially layering in external data – could lead to greater efficiencies and automation within customers’ systems. This could also result in a situation where treasurers increasingly adopt different workflows and ‘manage by exception’.

“In the future, we think that leveraging data could help automate a significant number of repeated activities – freeing up the treasury team to focus on exceptions and contribute more to the broader strategic objectives of the organisation,” he concludes.