In today’s business world, intense competition drives the need to change strategy quickly, and unforeseen events—such as the COVID-19 pandemic, natural disasters or terror attacks—can require thousands of employees to shift working patterns at a moment’s notice.
In the face of this, businesses are forced to look closely at their capabilities and processes to understand what will help or hinder their adaptability and innovation. While it may not be the solution for all, agile techniques present a powerful opportunity if the conditions are right, especially in the growing area of artificial intelligence (AI) and advanced analytics.
With its roots firmly planted in software development, agile is a counterpoint to traditional linear waterfall methods of delivery. Agile is often recommended when "The problem to be solved is complex; solutions are initially unknown, and product requirements will most likely change; the work can be modularized; close collaboration with end users (and rapid feedback from them) is feasible; and creative teams will typically outperform command-and-control groups." In other words, agile is suited to the realm of AI and advanced analytics, where poorly defined solutions are best iterated in cycles of rapid discovery.
Just as the methodology can be applied beyond the bounds of software development, the agile mindset can fundamentally change the way companies measure value and productivity. When paired with design thinking and behavioral economics, this mindset garners increased traction as the basis for a new way of working that takes the principles of agile methodology—simplicity, face-to-face conversations, iterative adjustments, and customer-centric design, to name a few—and applies them within a variety of contexts.
Those that harness agile techniques are strongly positioned to reap the rewards in scaling AI: faster speed-to-market, quicker value realization, competitive advantage, the ability to ‘fail fast’ and course-correct, and better collaboration across the business.
But for many companies, agile presents its own set of challenges. Instilling change across the entire enterprise can prove difficult, expensive and time consuming. And agile is by no means a "one-for-all" solution. As a result, many companies find themselves torn between thinking agile and being agile. A common example of this is the company that aims to take a flexible approach to project management, but still expects it to be completed against a specific, linear timeline, rather than through iterative agile sprints.
So, how can businesses use agile ways of working to unlock new value through AI and advanced analytics?
Three ways to unlock the value of AI through agile AI
Embed agile strategy into your AI delivery lifecycle
At the most basic level, harnessing the value of AI comes down to an organization’s ability to take an AI project from inception through to execution in production environments that impact customer relationships and a company’s end product. However, our survey research shows that a compelling 87 percent of UK respondents struggle to move beyond the proof of concept stage for AI projects because they either lack a documented delivery strategy or try to shoehorn the process into traditional IT delivery methods, among other reasons.
The AI roadmap is a start-to-end model we use with our clients to help them realize and multiply value from their AI projects. At the heart of this model sits an agile, sprint-based approach to capturing business requirements and iteratively delivering AI models in epics and user stories.
VIEW THE INFOGRAPHIC
There are two main traps that organizations might fall into when trying to embrace agile to deliver value from AI. The first is to try and adopt an agile mindset while using traditional tools to fuel it. Businesses that fall into this camp might use spreadsheets to run AI projects and email for communication to the detriment of agile values such as "individuals and interactions over processes and tools".
Instead, tools such as Microsoft Teams™, Slack™, Trello and JIRA put collaboration and adaptability at the center of workflows by enabling high levels of visibility within and across project teams and prioritize continuous shifts to optimize delivery outputs.
The second trap is investing in agile technologies without careful consideration of how they align to business priorities. Businesses risk ending up with pockets of value because, without support from the top or with cross-functional teams and business users, advanced analytics and AI teams adopt agile technologies in siloes that aren’t equipped to scale. In order to be more than just another application in the stack, agile tools, like any applied technology, need the right business use case and buy-in from senior business stakeholders.
Accenture has successfully implemented the AI roadmap at one of the UK’s leading banks. We helped design and scale up a data innovation bank through a ‘Value-Discover-Experiment-Prove-Scale’ methodology. Via the use of repeatable use case models, over 60 innovative and high-value analytics initiatives were identified and developed. In just nine weeks, several experiments that significantly improved the Bank were conducted, including a machine learning model that predicted a 20% uplift in onboarding customers. Ultimately, the data innovation lab evolved the bank’s analytics capability to the next level with over 300% ROI delivered within 21 months of being established.
Reinvent corporate culture to be a ‘data culture’
The success of applying agility to scale AI is not only method and technology-based; it is also cultural. In an attempted quick fix, many companies try to craft an agile culture by simply hiring in teams of data scientists with agile experience. But often, they do this without a strategy in place to build an organization-wide data culture that embraces the absorption of the insights being generated by AI.
Similarly, many companies believe it is enough to just have one agile team or one agile function for scaling AI; this is not always the case. If an organization wants to adopt AI at scale, for instance, a wider data culture must be fostered and it relies on everyone fueling it: top down, bottom up, and outside in.
This was exemplified at an insurance specialist. The client acknowledged that to build a unique, robust and effective data culture, it had to cement the right behaviors and ways of working. Accenture stepped in to make this a reality by developing a value delivery model which consisted of a flexible process, methods, tools and checklists to ensure that the data and analytics team were deploying solutions that solved actual problems. We also implemented our proprietary Data Pulse assessment to measure the consumption of data for insights. As a result of these interventions, the number of unique users consuming the analytics solutions increased from 20% to over 70%, indicating a much larger scale adoption of data-driven decision-making and spurring of innovation within the company.
Leadership influence is critical in instances such as the above. Junior employees, regardless of their skills, are likely to seek guidance from the people to whom they report. To really embed the cultural shift, leaders should consider connecting with employees across all areas of the business by:
- Making agile certification a prerequisite for all new hires to introduce cultural change gradually
- Running executive bootcamps—remotely, if necessary—to train the workforce in basic technology and agility from the top down
- Offering incentives—such as extra holiday or vouchers—for self-certification in new, agile skills such as scaled agile or design thinking
Recalibrate metrics for success
Agile methods and data culture are essential considerations for successfully scaling AI, but they are futile without a shift in the way success is quantified. In other words, redefining the frequency, methods and parameters with which the value of AI is measured.
To embrace agility and harness the power of AI, an organization may look to shift gears from rigid planning cycles to continuous, iterative planning once they identify and embrace the right metrics to define success.
For example, Accenture helped to identify the right metrics and insights for a major African bank by exploring the potential of real-time analytics. An Amazon Web Services Launchpad was deployed to conduct near-real time sentiment analysis of call center agent and customer discussions. Meaningful insights were displayed on a dashboard, which meant the contact center employees could move away from backward-looking KPI reviews and devote more time to improving customer satisfaction in the moment.
Critically, metrics at the highest levels within the organization must be recalibrated with data and AI in mind. A leading European bank decided to embark on a three-year transformation with the aim of embedding digital and AI capabilities across all their lines of business. Accenture worked with the Chief Data Officer of the bank to incorporate industry-standard indicators for data maturity such as the Data Management Capability Assessment Model® (DCAM) into the board-level metrics. This emphasized the bank’s commitment to scaling data and AI and went as far as to tie the personal bonuses of the leadership team to achieving the metrics. The results were remarkable, as the bank launched three AI-powered intelligent products within the year and unlocked $60m efficiency savings per year from a single line of business.
Start small, be agile, scale fast
Surviving, and indeed thriving, in today’s business landscape requires adaptability, innovation and new ways of working that remove the blockers to scaling AI and new sources of value. To unlock this benefit, businesses should adopt agile AI delivery methods, reinvent corporate culture into a data culture, and recalibrate organization-wide metrics for success.
From IT departments, product development and marketing teams to the C-suite, the message is clear: Given the right conditions, agile has the power to transform an organization’s ability to scale and extract value from investments in data and AI, shifting the focus from technology towards people and processes.
Most importantly, experience shows that agile techniques lend themselves to starting small, being responsive, and scaling fast, all of which are key success factors for weathering uncertainties and creating new opportunities in the world of AI, automation, and machine learning.