• Point of view

10 principles for building your data foundation

Augmented intelligence is the future of decision-making – in which the power of artificial intelligence (AI) blends with human judgment.

Machine intelligence is based on a data foundation. Without a solid data foundation, as part of a holistic data approach, organizations will struggle to train machines to deliver intelligent and trustworthy insights and recommendations.

But what is a data foundation? How can it be used? And which simple best practices can help you create one?

The baseline for business

Your data foundation drives actions and decisions that can make or break your business. On top of your data foundation, digital experts can layer in business and process knowledge, data science, AI, and machine learning capabilities to create a scalable, centralized repository of reliable data.

The power of a solid data foundation is evident when looking at how one financial services firm seeks to use artificial intelligence to accelerate credit decisions for commercial customers.

With customers across 35 countries, the firm needed to manage and extract data quickly while adhering to various global regulations, society expectations, and regional conditions in multiple languages.

Processing data across multiple locations and formats is possible by using natural language processing and computational linguistics to establish and manage a data foundation. Here, industry and process expertise are essential to creating a data foundation that handles information from public and private companies sensitively. In addition, prescriptive engines now run the necessary compliance checks.

With smart processes integrated with spreading and credit-scoring tools, the firm was able to automate 80% of financial statement processing, freeing up employees to focus on more strategic credit decisions. As a result, application-to-funding cycle times decreased from eight days to 48 hours, and customer satisfaction improved. And, by developing this AI-powered spreading solution for commercial credit-risk evaluation, the firm was able to improve its on-time credit decisions and compliance, too.

Take a copy for yourself

Download PDF

Avoiding common roadblocks

When businesses try to prepare a data foundation for AI and analytics programs, there are plenty of roadblocks that may appear. For instance, data often resides in different places, sometimes hidden from plain sight. This inability to access the right data at the right time creates higher costs, siloed decision-making, and slower achievement of business objectives. Plus, data is often found in a variety of formats – including structured data from enterprise resource planning, financial or other systems, or unstructured data, in websites, emails, and miscellaneous documents. This just adds to the headache.

To avoid this and other roadblocks, a method called extract, transform, load (ETL) can help businesses source and transfer data to a single data foundation. Unfortunately, ETL also has its limitations. It can't keep pace with large volumes of data and the flexible and adaptive nature required for successful augmented intelligence.

So, how do you create a solid data foundation for augmented intelligence?

Let's look at some key principles – adopted by many of the leading Fortune 500 businesses – that create a data foundation primed to unleash the power of augmented intelligence across the enterprise.

  1. Consider data integration: Businesses will always have to grapple with disconnected sources of data. Bringing enterprise data together requires a deep understanding based on industry-specific business challenges and objectives for each function to which the data applies. If you start with data integration, the next steps become easier to follow.
  2. Take a horizontal view: Rather than creating independent use cases – wherein data doesn't reconcile effectively – build out a unified function and add other use cases sequentially. In short, focus on horizontal use cases for your data foundation for less rework and an easier way to scale your efforts later down the line.
  3. Leverage bilingual talent: Employees that can speak to both industry-specific business processes and emerging digital technologies are known as bilinguals. If you have this talent in your business, make sure that your bilinguals can help build your data foundation so that they can spot the most impactful business use cases and ROI opportunities.
  4. Connect data quality to business success: The general perception of data – as a source for reporting – needs to change. As mentioned, the ability to leverage data effectively can make or break a business. By centralizing your enterprise data using the right digital technologies, your leaders will see that employees can make faster, more intelligent, data-driven decisions, positively impacting your data foundation.
  5. Make it traceable: It's essential that you can trace the way in which you push and pull data to the data foundation. When you can source transactions, it enables data lineage and reduces reconciliation time. This establishes trust and accountability for data audits.
  6. Take the opportunity to act fast: Near real-time data allows you to make decisions and act with speed. It also provides an opportunity for course corrections for faster business impact – something that's simply impossible with stale or incomplete data.
  7. Leverage the cloud: Build your data foundation in the cloud with granular transaction updates enabled by machine learning. This gives you adequate samples for machines to learn from. Then, they can deliver predictive insights and recommend the next best course of action in real time.
  8. Lead with architecture: Spend time on the architecture of your data foundation, including centralized infrastructure and controls. Keep data close to where it's needed using AI, operating systems, and analytical and digital experts. This provides efficiency, ownership, contextual knowledge, and accountability among all parties.
  9. Don't forget governance: Governance is important as new projects emerge, so data must not only be high in quality but also handled securely. This is particularly important given the need to comply with the General Data Protection Regulation (GDPR) and California Consumer Privacy Act (CCPA).
  10. A holistic approach: The data foundation isn't just a way to inform AI and analytics. Design and build your data foundation so that everything – including the data, digital platforms, products, and people – are all working in tandem toward the best business outcome.

If establishing a data foundation feels overwhelming, begin with a hard look at where your data sits today. From there, establish your goals for a data foundation – a roadmap for both your data and analytical investments. Make sure you have the right talent in place, and plan for your ideal future-state operating model.

Ultimately, if you follow these principles, you'll create a data foundation that lays the groundwork for augmented intelligence. You'll usher in a future wherein machines can uncover predictive insights and make intelligent recommendations that drive better decision-making and innovation across the enterprise.

Visit our augmented intelligence page