Many enterprise leaders who are bullish about GenAI hold the view that data cleansing and architecting must come before the technology’s rollout. But is this missing the bigger picture?
Data inputs impact analytic models. That still rings true in some cases. However, the emergence of unstructured data processing, whether via Large Language Models (LLMs) or traditional regression techniques, offers immediate opportunities that don’t require the complete overhaul of existing systems. Companies I speak to with GenAI success stories don’t have flawless data lakes or necessarily cutting-edge analytic stacks. Instead, they’re finding ways to move fast and unlock value with imperfect data environments. So, what’s their secret?
Not all use cases are equal
Some organisations are reporting huge efficiency gains and cost reductions from using GenAI while others are seeing modest ROI. More often than not, this comes down to use case selection. This is no surprise. It’s been a defining element of success in analytics for years.
The greatest challenge in the analytics process is widely viewed as this initial phase, translating business challenges into use cases. How might data analytics be used to optimise your inventory? How can data help streamline tax credits? Could you improve your customer service by being more personalised?
Currently, many organisations base their selection of GenAI use cases on risk profile. This is just one of the key factors for GenAI’s success. Use cases must align with the LLM techniques that we know to perform well. This means picking use cases that really leverage the amazing capabilities of what an LLM can do and staying away from those where LLMs will fall short.
The chatbot wave
While chatbots dominate GenAI applications due to customer service and process automation, their real value extends far beyond simple conversation. LLMs can be used to scan the news and summarise information to provide alerts. For example, you could input the cities and dates individuals at a company are traveling and create automated alerts sharing potential disruptions picked up on the internet scans. While an investment firm could use an LLM to sift through the news each day and provide succinct summaries for key news that could be used by analysts to assess against its portfolio. These are just two low-risk use cases where LLMs tend to perform well, summarising large amounts of unstructured data and providing succinct or even structured outputs that can be easily used.
Additionally, the use cases described require little data from the companies building the automation, send very little data externally, and can provide references to where the information came from so that the user can validate the sources. This is perfect for companies to ‘dip their toes’ into GenAI and serves as a great ramp to the technology with minimal risks.
Converting unstructured data into structured data
While many associate GenAI with chatbot solutions, others are finding that leveraging LLMs to convert large amounts of unstructured data into structured tables of data can prove impactful. Imagine using an LLM to scour the websites of your competitors to pull all their pricing into tables of data, which are organised in rows and columns (e.g. name of competitor, product description, current price). This leverages the magic of this new technology in a use case that most organisations would view as both safe and requiring minimal dependency on the quality of their internal data.
The challenge then becomes, how do you guide the organisation to the right use cases to start with? The answer lies in internal culture and education.
Change management
Successful GenAI adoption goes beyond merely putting the right technology into more hands. Organisations must provide education and foster an environment that embraces these new techniques. The concepts are not difficult, and learning how to apply the technology to a myriad of domains is within reach with the right mentors guiding the team.
Change management has been a longstanding requirement for organisations to achieve analytics maturity. Whether helping the organisation learn to leverage self-service data wrangling and modelling tools or applying Machine Learning (ML) techniques to problems. However, in the context of GenAI, change management becomes less of a “nice to have” and more of a non-negotiable necessity for success.
Education is critical. Companies deploying analytics tools often accompany this with one-off training. However, the most successful organisations blend practical skills (which includes the training to get them there) with foundational knowledge. Take data visualisation. While teams need to know which buttons to press, they also need to understand the principles underpinning effective visual communication. This combination of “how” and “why” creates far more impactful results than technical step-by-step guides. The same principle applies to GenAI. Organisations should have a systematic approach to bringing people on the journey using education and training, not just technology.
This can be summed up in fostering an AI literacy culture. And with this, there must also be guidance on when it’s appropriate to use the technology. GenAI can and will provide new capabilities, but not all problems are GenAI problems. It could be ML, automation, visualisations and other techniques. Organisations that understand this are far more likely to get the most out of GenAI technology.
Final thoughts
Flawless data, data readiness, and underlying infrastructure isn’t a prerequisite to GenAI success. What matters most is how organisations prepare and support their people through the transformation that the technology entails.
The good news? Critical success factors of education, knowledge sharing and change management are within the control of enterprise leaders. Companies don’t need to wait for perfect conditions to begin their GenAI journey. They can start today by building the right foundation of skills and understanding, confident in the knowledge that technology adoption is a gradual process.
Savvy organisations recognise that humans, not technical perfection, will determine whether their GenAI initiatives excel or falter. By investing in people’s ability to understand and leverage new tools effectively, they’re setting themselves up for success.
- Data & AI
