Cognitive Computing and the Watson Effect: A New Layer for Enterprise Systems?

Think back to that television broadcast from last year: IBM’s Watson system defeated Jeopardy champions by a considerable margin. For most business professionals watching, it looked like an impressive demonstration and little more. Then IBM started announcing what came next. Several major healthcare organizations in the United States are now running Watson as a clinical decision support tool in pilot programs. Similar experiments are underway in financial services. For enterprise technology managers in Turkey, the real question is straightforward: when does this become relevant to us, and in what form?

Cognitive computing needs a clear definition before the business implications make sense. Conventional enterprise software — ERP platforms, accounting systems, inventory modules — operates on explicit rules. A purchase order is entered, inventory is adjusted, an accounting entry is created. The system knows what to do because every step has been programmed in advance. Cognitive computing works from a different logic: it processes unstructured data — texts, reports, historical decisions — and generates probability-based recommendations rather than deterministic outputs. Watson does not ‘know’ the correct answer; it calculates the most probable answer given available evidence. That distinction matters enormously for enterprise applications.

IBM’s commercialization strategy for Watson starts with two sectors: healthcare and finance. In healthcare, the pilot application centers on giving clinicians decision support during diagnosis. The system evaluates patient symptoms alongside medical literature and ranks possible diagnoses by probability. In finance, applications are taking shape around risk analysis and regulatory compliance — scanning large transaction datasets for anomalous patterns, automating compliance reporting. What these two sectors share is that decisions carry both high complexity and high error costs, which makes the value proposition for decision support tools easier to quantify.

For enterprise decision-makers, a practical framework helps here. Watson-type cognitive systems are not positioned to replace existing ERP or accounting software; they sit on top as an analytical layer. Consider the perspective of a finance director: the ERP records and reports cash flow data. A cognitive analytical layer could evaluate that same data alongside sector conditions and historical patterns to generate early warnings about liquidity risk. The ROI calculation becomes clearer at that point — improvement in decision quality and reduction in costs from delayed responses, measured against total cost of ownership (TCO). Making that calculation with precise figures today is difficult, though, because the pilots are still producing results.

From the perspective of Turkey’s enterprise software ecosystem, the near-term operational impact of these developments looks limited. The current agenda for most local companies centers on e-Invoice and e-Ledger compliance obligations, ERP integration projects, and early cloud migration discussions. Cognitive computing remains in pilot phase even for large-scale enterprise environments. That said, these developments demonstrate clearly that analytical capacity in finance and accounting processes is becoming a strategic competitive factor. Watching how quickly the healthcare pilots translate into financial services, insurance, and corporate credit assessment — all domains with comparable decision complexity — will be an important indicator of the timeline.

The practical obstacles deserve honest assessment. Watson’s training process requires large volumes of domain-specific structured and unstructured data. In the healthcare pilots, that data came from medical literature and patient records accumulated over years. A financial institution adapting a similar system to its own processes needs both strong data quality and a mature data governance framework. In most Turkish organizations, data governance is not yet a developed discipline. Data extracted from ERP systems is frequently inconsistent, incomplete, or non-standardized. Building an analytical layer on top of that foundation does not produce reliable results — the underlying data problem has to be addressed first.

For enterprise technology managers, the appropriate response right now is not to pursue speculative investment in cognitive computing platforms. The priority is strengthening data infrastructure. Improving data quality within ERP systems, ensuring consistency across accounting and finance processes, treating the e-Invoice and e-Ledger transition as an opportunity to establish clean data practices — these steps deliver operational efficiency today and open the path for analytical layer investments later. Monitoring Watson’s commercial development and watching how IBM shapes its pricing and access models will be sufficient for this year. The real decision point is still a few years away, and the organizations that arrive there with clean, well-governed data will have a meaningful head start.

This article was originally written in Turkish by Gökhan MERCANOĞLU on February 27, 2012 and has been automatically translated into English and other languages using machine translation.


If risk management is approached only as an efficiency agenda, it remains incomplete. Customer experience, employee behavior, financial impact, and operational resilience must be evaluated together. Corporate technology changes not a single department, but the way the whole business operates.


Gökhan Mercanoğlu
Finans Yönetimi