Research on Optimization Paths for Enterprise Decision Support Systems Based on Big Data and Artificial Intelligence
Drawing on data from 108 enterprises, this study proposes a novel "technical-management dual-helix" model that integrates data fabric, explainable AI, and organizational learning mechanisms to overcome data isolation, algorithmic opacity, and structural inflexibility, thereby enhancing decision quality and response speed across manufacturing, finance, and retail sectors.
Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer
Every modern company relies on a digital nervous system to make choices, from a factory deciding how many parts to build to a bank approving a loan. For years, the assumption was that if a business bought enough powerful software and gathered enough data, its decisions would automatically become smarter. Yet, a persistent gap has emerged: companies are spending billions on technology, but their leaders often still struggle to make effective choices. The problem is not a lack of information, but a mismatch between the tools they use and the way their organizations are built. Data often sits in isolated silos, the algorithms that process it remain opaque black boxes that humans cannot trust, and the company structures themselves are too rigid to adapt to new insights. This disconnect creates a situation where technology exists, but the value it could provide remains locked away.
Researchers Yixuan Yao and Liang Chen from the Yunnan College of Business Management set out to solve this puzzle by looking at how technology and management must grow together, rather than separately. They propose that a decision support system cannot simply be an add-on tool; it must evolve in tandem with the people and processes it serves. To test this, they studied 108 enterprises across manufacturing, finance, and retail, interviewing leaders and analyzing years of operational data. Their work moves beyond the idea that better algorithms alone will fix broken decision-making. Instead, they found that success comes from a dual approach: one strand focuses on making data accessible and the logic behind AI transparent, while the other strand focuses on training the organization to trust and use these new tools. By weaving these two strands together, companies can transform their decision-making from a static, error-prone process into a dynamic, learning system that adapts to change.
The researchers discovered that the most effective path forward involves two interlocking spirals of improvement. The first spiral is technical. It begins with a "data fabric," a flexible architecture that connects scattered information from different departments into a single, coherent view, much like a unified map for a traveler. On top of this foundation, they introduced explainable artificial intelligence. Unlike traditional systems that spit out a recommendation without saying why, these new engines reveal the reasoning behind their suggestions. They show which factors influenced a decision, such as a supplier's delay or a seasonal trend, allowing human managers to understand the logic and verify the trustworthiness of the output. This transparency turns the algorithm from a mysterious oracle into a collaborative partner that managers can question and learn from.
The second spiral is organizational. Technology alone fails if the people using it are not ready. The study found that successful companies build a "context-aware" system that constantly monitors the business environment, from market shifts to internal bottlenecks. More importantly, they cultivate a culture of continuous learning. When a decision is made, the organization does not just move on; it reviews the outcome, extracts lessons, and updates its internal knowledge base. This creates a feedback loop where the organization gets smarter with every choice. The researchers measured a specific metric called the "change resistance index," which tracks how much employees fear or reject new tools. In the companies that followed their dual-spiral model, this resistance dropped significantly over time, as workers shifted from viewing the system as a threat to seeing it as a tool that enhanced their own expertise.
To prove that this approach works, the team conducted a detailed comparison between companies using their new dual-spiral model and those sticking with traditional, single-focused upgrades. In a manufacturing setting, a company that adopted the full model saw its ability to forecast demand jump from 68% accuracy to 89%, while a similar company using only traditional software improved to just 75%. The new model also slashed the time it took to make a decision from over four hours down to under two. In the financial sector, a bank using the approach reduced its risk response time from 48 hours to just 15 minutes, while simultaneously meeting strict new regulatory requirements that had previously seemed impossible to satisfy. In retail, a chain of stores shortened its product launch cycles by half and increased its marketing return on investment from 1:3.2 to 1:5.8. These improvements were not just about speed; they were about the quality of the choices being made and the trust the teams had in them.
The study also highlighted that there is no single "one-size-fits-all" solution. The path a company takes depends heavily on its starting point. A factory with messy data and complex supply chains needs to focus first on cleaning and connecting its information. A bank with strict rules needs to prioritize governance and trust-building before adding new automation. A retailer facing fast-changing trends needs an agile structure that allows for rapid testing and iteration. The researchers developed a method to diagnose a company's specific situation—looking at how complex its decisions are, how mature its data is, and how flexible its organization is—and then matched it to the right combination of technical and management steps. This ensures that resources are not wasted on the wrong fixes.
While the results are promising, the researchers are careful to note the limits of their findings. The study relied on real-world data from 85 valid responses and deep dives into specific cases, but it was not a randomized experiment where companies were randomly assigned to different groups. This means the improvements observed are strongly associated with the new model, but other factors like general market trends or internal leadership changes could also have played a role. The data comes primarily from Chinese enterprises, and the researchers acknowledge that cultural differences in how people trust technology might mean the results look different in other parts of the world. They also point out that the technology landscape is moving fast, with new developments in quantum computing and generative artificial intelligence on the horizon that could reshape these systems again.
Despite these caveats, the core message is clear: the era of buying a software package and expecting instant intelligence is over. The future of decision-making lies in the slow, deliberate work of aligning technology with human organization. It requires building systems that are transparent enough to be trusted and organizations that are flexible enough to learn. By treating technology and management as two sides of the same coin, growing together rather than apart, companies can finally unlock the true potential of their data. The path forward is not a single leap, but a continuous, spiraling climb where every step in technology is matched by a step in human capability, creating a decision-making engine that is both powerful and wise.
Drowning in papers in your field?
Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.