At first sight, Financial Advanced Analytics can seem overwhelmingly complex. Not only do we have to consider factors that are relevant to advanced analytics projects in general, but we must also consider industry-specific factors within finance processes. These include layers of legislation, an emphasis on accountability and traceability in relation to decision-making (even down to individual cases), and the fact that decisions made within financial processes can have far-reaching implications for consumers or companies (e.g. at which interest rate do I obtain a credit).
The good news is that the benefits are enormous. Advanced analytics can, and already is, revolutionizing the finance world and finance processes by uncovering opportunities, augmenting our understanding of our customers, and enabling successful and more efficient financial decision making across the customer lifecycle and order-2-cash process – also in the interest of customers. Effectively-implemented Financial Advanced Analytics (see our insights below) opens up completely new possibilities for finance departments, also when it comes to their internal positioning. They can now assist other departments such as Marketing, Sales, CRM and SCM even more than before by providing actionable insights. Cross-departmental data consolidation allows for a holistic view of the relative use case and customer. This contributes positively to the goals of all departments as well as the overarching company strategy and in the end creates a win-win situation for the entire company.
So, how can you take your financial decision making to the next level? There are many factors to consider, but I’d like to focus on the six key elements that are necessary for a holistic strategy and the basis for above-average ROIs in Financial Advanced Analytics Projects.
1. Data – the heart of business analytics in finance
Every business insight begins with the harvesting and processing of data. Finance processes are generating an immense amount of data – both structured and unstructured – thanks to the need for traceability, accountability and security. According to Veritas, approximately 50 percent of all the data stored and processed by enterprises globally today is “dark data”, the value of which has not been quantified yet. Only 15 percent of all stored data today is considered business critical in terms of decisions. That’s why a business-impact driven approach towards data harvesting is crucial. However, what is considered “critical” data may change over time, so a regular monitoring process (for example during model validations) should be established along with a sustainable data governance policy that considers data quality and quantity.
As an example, a bank might traditionally calculate the value of a customer (and use it for steering in different finance processes) by looking at their products and connected account transactions. But additional information can come from other sources. Banking app/portal data (locational, behavioral, device and login information acquired with the necessary consent and compliant with the latest regulations) can be harvested and can add to the understanding of a customer when combined with traditional data sources.
Beyond the amount and source of data, the speed at which it can be processed, as well as its availability at a certain point in time for decisioning, should also be considered. This is especially relevant for real-time automated decisioning in a financial context, where decisions (such as accept/reject/refer) need to be made immediately.
2. Tools to enable successful financial decision making
The implementation of advanced analytics in finance does not begin with the technology. Instead, it is better to think about the use cases and data you have available, and coupled with your business goals, formulate a coherent strategy accordingly.
When the time comes to select the tools, the first question is suitability for your particular business needs. Will they be able to meet your requirements as your transformation progresses? There is strong intercorrelation between all of the six key elements of the Financial Advanced Analytics strategy. In the case of platform selection, for example, one important question to ask is should you opt for an open source or commercial solution? Should the platform reduce complexity with a graphical user interface or do you really want to go down to the code level, or somewhere in between? How to bring the results into production?
In the end you shouldn’t search for the cutting-edge solution – Pareto for example may offer a workable approach to the technology question. However, some requirements aren’t negotiable – such as compliance, regulatory framework and requirements from authorities with regards to documentation and traceability, as well as role-based access capabilities.
3. Methodology – gain transparency and make it scalable
Any methodology applied to the implementation and operation of Financial Advanced Analytics needs to be based on a thorough understanding of your business and goals, of the concrete use case, the potential ROI and the corresponding data. There are existing methodologies that can serve as inspiration such as CRISP-DM, which divides the implementation of data mining projects into 6 key stages and which can also be applied to Financial Advanced Analytics projects: business understanding, data understanding, data preparation, modelling, evaluation and deployment. It needs to be remembered though that any proprietary model will need adaption to your individual business requirements.
After you have decided on a methodology, planning and communication are crucial. A very common pain point is failing to adequately plan for deployment, often referred to as “the last mile”. It’s easy for a project team to get isolated in the transition phase and present a solution that has not been communicated, so keep key stakeholders informed about the why and the how of the project, the expected ROI, and the project duration to ensure buy-in at the deployment phase. In the end, the results of Financial Advanced Analytics projects change the status quo, which needs to be prepared and managed accordingly.
Finally, while financial process knowledge is the foundation for success in implementing Financial Advanced Analytics, keep an open mind as to the algorithms and concepts you use. An experienced partner can often apply best practice from other industries and help you benchmark your competitors to identify the most beneficial areas to apply Financial Advanced Analytics – or even areas which might not be promising to start with.
4. The people – combining data with financial process understanding
All the talk of technology in relation to Financial Advanced Analytics can sometimes overshadow one of the most important elements of creating a data-driven business and culture – the people. In line with advances in technology, roles and skillsets have also evolved.
At one time, for example, it was common for data scientists to work alone on isolated proof of concepts, collecting and interpreting data in their labs. In a complex data-driven environment however, data scientists that work in isolation will struggle to provide value – one of the main reasons why most proof of concepts don’t make the last mile. Today, they need to be proactive, commercially-aware team players who understand business, as well as analytics/data and can proactively liaise with other stakeholders. This also applies to all other stakeholders who are part of the project team or get involved with Financial Advanced Analytics projects – it’s all about open-minded data-driven thinking.
It’s critical to conduct an assessment of what talent is needed, and then develop a strategy that ensures the availability and continuity of mission-critical talent, formulated alongside any project to implement Financial Advanced Analytics. Once the new or modified roles have been identified, it can then be assessed which roles are best kept in-house, and which are supplied by external partners. The ratio between the two can vary significantly depending on time, money and business strategy.
5. Change management for successful Financial Advanced Analytics projects
Change management in the implementation and deployment of Financial Advanced Analytics is probably one of the most overlooked aspects in the process. As data is often siloed in departments within large organizations, a culture of collaboration needs to be fostered in order to maximize its benefits on a corporate level. One way to achieve this is to create win-win situations whereby specific departments cooperate with analysts to achieve mutual goals.
As an example, what would be the benefit if you as a Key Account Manager for one of your company’s most important clients, adjusted the credit limit proactively in advance of large deals, instead of receiving escalations afterwards? Lots of potential can be uncovered, for example, by synchronizing the objectives of the sales and finance departments and merging datasets from both departments.
It should also be remembered that at the end of the day, not only machines but also people will interact with the insights and decisions made actionable by the data. In this respect it is vital to get all stakeholders, such as front office staff, onboard at the earliest possible stage to ensure that they are ready at time of deployment. There are many cases of Financial Advanced Analytics projects being delayed because staff were unprepared for the new processes and predictive capabilities. The lesson? Open dialogue and clear communication is essential – framed by a clear strategy map that explains the why and how of the initiative.
6. Operational models for Advanced Analytics in financial processes
According to CRISP-DM, the operational model is actually part of the implementation phase of the methodology. However, our experience shows that this factor is critically important and merits separate discussion. Whichever model you decide upon, there are a couple of factors to consider that can affect the effectiveness of the model. The first is failing to plan effectively for deployment. As with change management, communication plays a key role in ensuring that people across the organization are onboard with the implementation and deployment both in terms of what to expect and why.
A second factor that can limit the effectiveness of the operational model is the failure to monitor, maintain and update the model over time. A static model cannot remain truly relevant with changing market conditions, consumer/client behaviors and portfolio structure. Therefore, it’s mandatory to establish a sustainable monitoring framework which ensures that the decisions are heading in the right direction over time and that models are adapted when necessary. Interestingly, in this regard, non-regulated businesses sometimes forget this whereas regulated businesses are very mature.
Last, but not least…
Having highlighted the six key elements for successful Financial Advanced Analytics projects, perhaps it’s time to ask how you should start? I would say take some time to define your strategy covering at least the 6 key elements identified above and explain where you want to go and most importantly why (create a sense of purpose)! Afterwards, start focused and be open-minded about the data sources, and seek to combine data to leverage insights. And don’t forget that the human element, change management and communication are vital parts of the mix.
If you’re ready to take the next step, find out more about how Financial Advanced Analytics can transform your organization!
If you’d like to learn more about our Financial Advanced Analytics insights series, or if you have questions and wish to get in direct contact with our experts, send us an email here.
Vice President Analytics & Consulting Services | Arvato Financial Solutions