Business Imperatives for GenAI for Analytics
For decades, enterprises have been improving and expanding the way they use analytics and data software, commonly referred to as business intelligence (BI) software, to improve their operations. BI software enables enterprises to improve business outcomes by operating more efficiently, accelerating product development and enhancing customer service. BI software providers have made dramatic improvements to their products over the years, with highly interactive visualizations and the ability to process and display very large volumes of data quickly.
These improvements have led to efficiency improvements for data analysts, making it easier to collaborate on the development of BI reports and dashboards for senior executives and business decision-makers. Even with these improvements, there remain challenges with BI that limit engagement with BI software and cause delays and friction that can negatively impact business decision-making.
Despite the focus on data-driven decision-making in recent years, only one-third (33%) of participants in the ISG Research Analytics and Data Benchmark Research are satisfied with their enterprise’s use
of analytics and data. Perennial complaints about analytics and BI include challenges integrating with business processes, a lack of adaptability to change and complexity of accessing data sources. Additionally, preparing data for analysis and checking data quality remain too time-consuming.
Traditional reports and dashboards are often static, and actionable insight depends on the user interpreting tables and visualizations and making intelligent decisions. Exacerbating these issues is that engagement with analytics software is too often limited to those in data analyst roles rather than business decision-makers. Only 15% of participants in our research are very comfortable providing business users with self-service access to data. Almost double the number (29%) of those satisfied with their enterprise’s use of analytics and data are very comfortable providing business users with self-service access to data, however.
Artificial intelligence and machine learning (AI/ML) have the potential to accelerate business decision making by providing users beyond data analysts with automatically generated explanations, predictions and recommendations. However, traditional approaches to AI/ML require skills that are beyond the reach of many workers, and enterprises have had difficulty finding skilled resources. As a result, we expect through 2025, AI and ML approaches will remain largely independent of BI approaches, requiring three-quarters of organizations to maintain multiple, separate skill sets. The emergence of generative AI (GenAI) has the potential to further empower data analysts. GenAI tools can be used to create content, including text, digital images, audio, video or even computer programs and models. When combined with governance to improve accuracy and trustworthiness, however, GenAI can improve data democratization and lower the barriers to business users and decision-makers working directly with data and analytics.
How GenAI Empowers Data Analysts
GenAI has potential efficiency benefits for data analysts. The ISG Market Lens AI Study indicates that analytics is at the forefront of AI adoption, with 87% of participants indicating that their organization is using AI for analytics and BI, well ahead of other application areas such as customer engagement (55%) and content management (48%). Analytics software providers and users are moving quickly to take advantage of GenAI, using large language models (LLMs) to convert natural language questions into analytic queries, as well as automatically generating summarizations and recommendations from data and charts.
By 2026, three-quarters of enterprises will realize their analytics are ineffective without GenAI capabilities to guide the workforce with personalized recommended actions necessary to improve
outcomes. GenAI can also be used to automate routine tasks such as data preparation, cleansing and transformation. These are amongst the most time-consuming aspects of any analytics project, with more than two-thirds (69%) of participants in our research spending most of their analytic time preparing data for analysis, compared to only 27% who spend most of their time determining how changes impact the business.
Natural language processing (NLP) and natural language generation (NLG) are by no means new to the analytics sector and are key elements of augmented analytics that we assessed in our Analytics and Data Buyers Guide. However, pre-GenAI implementations of NLP and NLG were often complex and required much work from BI development and IT teams to model potential use cases. It was also time-consuming to create and maintain databases of synonyms required to convert natural language questions into analytic queries.
GenAI also has significant potential to unleash the value inherent in unstructured data—something that many enterprises have previously struggled to capitalize on. Unstructured data can include documentation, as well as comments in surveys or social media. While this information has huge potential value, it has traditionally been difficult and time-consuming to analyze. GenAI can create structured data from these sources that can be used in further analysis. Comments can be categorized by sentiment—positive, negative or neutral—while data can be extracted from documents or images and converted to more structured information for use in traditional analyses. Potential use cases include analysis of audio, video and images as well as sentiment analysis of social media content and interpretation, and summarization of written documents.
How GenAI Enables Business Decision-Makers
Traditional reports and dashboards provide users with data and charts to be queried. By providing natural language interfaces that are intuitive to use, GenAI facilitates data literacy and enables data democratization by enabling business decision-makers to engage directly with data using natural language search queries. GenAI also provides users with narratives and recommendations which they can more easily interpret to accelerate business decision-making. People need to be trained to interpret dashboards and charts but written and spoken language narratives and recommendations can be understood intuitively without specialist skills. GenAI tools provide consistency regardless of the user’s knowledge or skill level, and business users no longer need to be experts in query languages and analytics and BI tools to generate business value from data.
While adoption of GenAI for analytics is still in its infancy, early signs are encouraging. For key analytics use cases, 99% of participants in the ISG Market Lens AI Study have seen positive outcomes
from natural language search queries, while 97% have seen positive outcomes from the interpretation of data.
The addition of GenAI-based NLP and NLG capabilities to existing analytics and BI software products promises to automate and accelerate the work of data professionals. It does not necessarily mean that existing analytics and BI software products will instantly become suitable for use by business professionals, however. More widespread interaction with analytics is likely to be triggered by the development of entirely new products designed with GenAI interfaces as the primary means of interacting with data, supported by charts and tables, rather than GenAI interfaces being bolted onto or alongside the charts and tables delivered by traditional reports and dashboards.
While GenAI is the current focus of attention, it is not the only approach to providing business users with enhanced analytics capabilities. Automated machine learning (AutoML) is an established approach to automating some of the routine, repetitive tasks required in AI/ML model development. AutoML is most often used today to generate forecasts and to perform customer segmentation analyses, but we see AutoML capabilities expanding to support more types of analyses and produce models with improved accuracy.
Challenges of GenAI
Although GenAI has many potential benefits, enterprises should be aware of its inherent limitations and plan accordingly to ensure that they avoid potential risks. For example, LLMs generate content that is grammatically valid rather than factually accurate, resulting in potential factual inaccuracies such as fictitious data and source references. This is because foundation models only have “knowledge” of the information they are trained on. This could be enormous amounts of public information, but public LLMS do not have access to an enterprise’s private data and content. A public LLM can provide accurate responses about generic questions for which there is a large corpus of freely available information but ask it a question that requires private data that it has not been trained on—for instance, about a particular company’s latest sales figures—and it will generate text that is plausible but has no basis in factual data.
If decision-makers are to trust the information that they are provided with, they need to be able to understand the basis for the predictions and recommendations.
While natural language interfaces powered by GenAI reduce the need for technical and domain expertise to query data, they do not reduce the value of domain expertise in interpreting and providing feedback on the results. This reinforces the need to incorporate driver-based planning capabilities into decision-making processes to evaluate the requirements for and implications of recommendations in making intelligent decisions. Additionally, as more business users begin to interact with and analyze enterprise data, the greater the need for agreement on data definitions, reinforcing the importance of semantic data modeling to standardize metrics and definitions.
Trust in the content created by GenAI can be improved by augmenting foundation models with real-life data and context from enterprise information. Explainability is essential for AI-driven predictions and recommendations. If decision-makers are to trust the information that they are provided with, they need to be able to understand the basis for the predictions and recommendations. Users should also be able to verify the attributes used when converting natural language searches into queries to ensure accuracy while also being able to trace the lineage of the response to validate its accuracy. The ability to provide feedback based on the accuracy of generated content is also essential to ensure that the applications learn from and correct mistakes. Validating the output of GenAI models to ensure that it is consistent with trusted data and content is clearly potentially time-consuming, but this task can also be automated using machine learning models. Many software providers are building validation loops into their software to automatically validate the output of GenAI.
Key Considerations for Success
GenAI has the potential to change the face of BI, but business users and data analysts have different requirements that need to be enabled by different functionality and interfaces. By 2026, one-third of
enterprises will replace legacy business intelligence tools with analytic platforms that are collaborative and utilize GenAI to inform and guide business professionals. Bolting GenAI interfaces onto existing BI tools aimed at data analysts will likely enhance productivity, but democratization for business users will come from the next generation of products developed with GenAI-based NLP and NLG as the primary interfaces for accessing and analyzing data. Enterprises evaluating analytics software include GenAI-based capabilities in the evaluation criteria while also being mindful of the requirements of different users.
Good analytics software provides more than just a pretty interface. GenAI interfaces can only do so much if the foundation of underlying data processing and analytics functionality is weak. Enterprises should also be mindful of the functionality provided to enable users to verify the output of GenAI, including validating data sources and explaining the basis for predictions and recommendations. Security, privacy and regulatory requirements should also be top of mind, alongside reliability and performance considerations.
Innovation in AI is moving at an accelerated pace. While enterprises are advised to move quickly to embrace GenAI to avoid being left behind, they should also be mindful of the risks posed by making purchasing decisions in haste. Making the wrong choice could raise the total cost of ownership, lower the return on investment and hamper an enterprise’s ability to reach its potential performance. Enterprises should be cognizant of the risks involved in getting tied into a specific model or approach. There is no one-size-fits-all approach to GenAI, and organizations will need to use a variety of models and providers to ensure that they have the flexibility to adapt to changing business requirements and the rapidly evolving landscape of AI models.
