Enterprise digital transformation occurs when digital technology enables businesses to function better than before, improving efficiencies and business processes. At the core of this process is the transition of analog information to digital formats, with significant value available to organizations that adopt the right cost reduction strategies, spend management strategy, spend analysis process, and expense management software while doing so.
The ability to decipher detailed data about complex services can unlock immediate and long-term value, both in terms of savings and overall efficiency. Tackling the right pain points with the right technology results in an enhanced approach to transformation.
To achieve these outcomes, organizations require a spend analysis solution that can improve supplier management and performance. Charging inconsistencies can commonly be traced back to the supplier, due to factors such as inaccurate data and methodologies.
A subset of improved supplier performance is cost savings, and the right technology can enable you to optimize existing service plans without having to renegotiate deals. These improvements can be taken a step further when service data is aggregated across multiple suppliers, resulting in better deals.
Finally, driving operational improvement can further enhance enterprise digital transformation. This can be achieved by equipping project managers, third-party applications, and vendors with accurate service data, helping objectives to be achieved in each category.
Complex spend data visibility obstacles
Having identified the key opportunity areas, next we must review the challenges. The first primary obstacle is the gathering of invoices and building a clean fact base. This is made complex by e-invoicing and accounts payable systems that simply summarize and store the source documents. Further difficulty is added by variations in invoice formatting, as well as terminology used by different providers and suppliers.
The key to solving the invoice gathering challenge is having a system with flexible entry points, which can be achieved by connecting an API with suppliers. This can then be reinforced by setting up access methods directly from online billing portals.
In the case of Thinking Machine Systems, the solution leverages Named Entity Recognition (NER) to identify matching entity types across numerous formats. Supervised training models are essential in this process, ensuring that quality measurement is always in place to promote accuracy.
In addition to gathering invoices, identifying the relevant contract data can also present a significant challenge. Contract documents commonly have highly variable lengths and structures, which makes it a challenge to pinpoint relevant information across many documents.
In the case of our solution, we train systems to identify relevant entities, with examples including pricing inclusions, exclusions, and specific product types. Training the machine to identify specific entities makes it easier to accurately process larger quantities of data.
Associating the relevant contract information with the corresponding invoices also presents a costly challenge for organizations. In some cases, contracts are explicit, but the way information is presented can still differ drastically from the way it appears on invoices. For example, product descriptions can be named differently, rates may or may not include discounts, and abbreviations are often used without description. The Thinking Machine Systems solution overcomes this challenge using its powerful artificial intelligence and a machine learning driven Knowledge Model.
How we use digital technology to drive savings
Once we have established the clean fact base, we organize into different segments to provide a logical way of identifying and grouping key information. For instance, we separate site location data from service information, where we can identify every service type. We also separate product information, enabling our solution to identify all product types and categorize them effectively. Contracts also account for a huge segment of the data, which derive key information for understanding how products and services are priced.
We are then able to look at this information from multiple angles. By visualizing these data subsets in a way that they can be analysed all in one place, we are able to raise questions that reveal opportunities to make savings. This is also a valuable process for identifying any inaccuracies, as well as providing the customer with additional peace of mind. We can analyse the services used by a customer and clarify whether the employees using them are still at the company, potentially unearthing significant, immediate savings.
Our machine learning technology is trained to recognize these different information categories, automating the process of asking critical questions. This could include reviewing all services that are connected to each of their addresses or checking whether contracts match correctly with associated invoices.
The machine can also identify new questions that need to be asked in a dynamic way, as well as asking the original questions it was trained to use. The machine is then able to naturally predict many different failure points and savings opportunities, leveraging the information built into the fact base. We can then provide the level of context that is required to act on the savings recommendations.
The result of this rich data visualization enables key data sets to be explored in an easy way, with the critical questions having already been addressed by the solution. These elements are all brought together in a unified dashboard, where you can view categories like products, sites, and services all in one place. Within this single dashboard, the user can also explore the overarching fact base that drives the whole process.
Real company spend savings that you can trust
Once this process is complete, the customer needs to be able to confidently relay these savings opportunities and recommend how to reduce the costs to their CIO, CFO or other procurement teams. To support you at this stage of the process, the machine creates fact base subsets and presents it back to you in a concise, easy to read format. This equips you to efficiently explore the evidence behind savings recommendations and opportunities, enabling you to validate the results generated by the machine.
In one example, we were matching invoices against contracts based on three elements: products, sites, and charges. The contract stated that the associated products should be charged at a specific site at a specific amount, including a monthly charge and a one-off payment. When we analyzed this data, we found instances where the three elements failed to match properly, allowing us to investigate and identify the specific postcodes of sites where overcharging was taking place. The result of this was a verified savings recommendation of £100,000.
Driving enterprise digital transformation through spend analytics
As digital transformation enables enterprises to leverage emerging capabilities, organizations are set to face an increasingly complex technology service landscape. To fully optimize the benefits of digital transformation, it will be essential to gain an enhanced level of control and insight when it comes to service-related contracts and invoices. This will ensure that significant value is not being lost during the transformation process, while also promoting general data hygiene to better inform decision-making.
The Thinking Machine Systems solution is designed to overcome the challenges traditionally associated with data gathering, regardless of the origin or complexity of the documents involved. This is achieved through our innovative machine learning approach, made possible by our unique Knowledge Model.
Crucially, our solution not only constructs a clean fact base, but also automates the identification of immediate and long-term savings opportunities. The machine then equips delivery teams to effectively communicate these recommendations to senior stakeholders, promoting an effective data-driven process.