We’re still only at the beginning stages of AI, and data appears to be one of the biggest issues slowing down progress. That’s one of the revelations of a new survey from Accenture of 2,000 executives.
“Though every business may want an AI-powered edge, many companies are still struggling to advance beyond their initial AI experiments,” survey co-authors Senthil Ramani, Lan Guan, and Philippe Roussiere find. “A big reason for this is low data readiness—which arises when all types of data, especially unstructured data, are not used to the max.”
At the same time, 70% of executives acknowledged the need for a strong data foundation when trying to scale AI. Along with data issues, executives also cited “outdated IT systems, as well as workers’ lack of access to, respectively, gen AI tools, comprehensive training and clear guidance from leadership” as significant barriers.
People’s issues also slow down AI progress. “[B]uilding and maintaining multi-disciplinary teams is, on average, the greatest challenge for both front-runners and companies that are experimenting with AI, …” the survey shows. Data issues come up as companies move from experimentation to implementation. “[B] uilding an end-to-end data foundation with quality data is the greatest challenge for companies that are progressing with AI—and the second-greatest challenge for everyone else.”
Organizations reporting substantial progress with their AI initiatives are in the minority. Only 8% of companies—considered “front-runners”—are scaling AI at an enterprise level, embedding the technology into core business strategy.
They build modern AI and data infrastructure that ensures AI is scalable, secure, and seamlessly integrated across business operations. They also invest an average of 51% of their tech budgets in the cloud and AI.
Those companies right behind the front-runners—“fast-followers”—have a solid grasp of data requirements, the survey also shows. “For example, 96% are very strong in data governance, compared to 83% of front-runners. Ditto for data platforms (98% versus 90%, respectively).”
But in many other data-related practices, the leaders are far ahead of their slower-moving counterparts, the survey indicates. For example, 17% of front-runners use RAG, or retrieval-augmented generation, to enhance their large language models, while a meager 1% of fast-followers do.
Knowledge graph adoption is also much further along with the front-runners. Twenty- six percent of the leaders employ knowledge graphs to connect data to AI applications, versus only 3% of their slower counterparts.
At least 22% of the leaders focus on complete data lifecycle management, versus only 6% of the lagging organizations.
In addition, they’re more likely than fast-followers to heavily use zero-party data (44% versus 4%), second-party data (30% versus 7%), third-party data (25% versus 8%), and synthetic data (35% versus 6%).
The front-runners are seeing tangible advantages from their AI and data work. After deploying and scaling AI across their enterprise, they expect to reduce their costs by 11% and increase their productivity by 13%, on average, within 18 months.
The Accenture authors advise building a culture around data products and driving semantic consistency across data products.
This involves “[b]reaking silos, driving semantic consistency and standardizing data to enhance interoperability and governance.”
In addition, they urge “Embedding data-product thinking across functions, to democratize access to data.”
Adroit use of GenAI is also encouraged, embedding the technology into existing portfolios of applications while avoiding redundant investments.
Interestingly, the Accenture authors label investment-heavy initiatives such as AI as strategic “bets”—although these bets are based on well-considered plans and advice.
Bringing about AI and data-fueled transformation “isn’t simply a matter of deploying a few chatbots,” the co-authors emphasize. “Reinvention is about building advanced AI capabilities like agentic architecture, networks of AI agents that go beyond automating routine tasks to orchestrating entire business workflows.”
Agentic architecture is spreading fast, with one-third of the companies surveyed already using AI agents.