Introduction to AI Powered Innovation: Building a business-savvy AI backlog

Artificial intelligence (AI) provides the opportunity to transform how organisations operate and engage with their customers. Those who invest intelligently and aggressively in AI backed innovation  will gain market share and shape the future of their respective industries – particularly important in an economic downturn.

Once an organisation is ready to invest, the potential value of AI can be demonstrated early in the form of proof of concepts (PoCs). Of the AI leaders who’ve managed to develop PoCs and scale one or more for production, 97% are seeing quantifiable benefits of increased sales, streamlined operations and better understanding of their customers.

In practice the biggest challenge facing AI leaders in industry, is not the deployment of AI PoCs, but scaling those PoCs in to fully operational and industrialised solutions that deliver measurable value at scale. Just 13% of organisations that have invested in AI have managed to successfully scale an AI solution across multiple business teams. Investing in the right solutions is pivotal – 51% of these organisations have had to suspend or scale-down investment in AI initiatives after making a poor investment decision. Having a dynamic and well-prioritised AI backlog is key to ensuring you invest where you’re most likely to unlock value.

The concept of a ‘product backlog’ is well defined in Agile methodology. It is a prioritised list of ‘work’ that is derived from a high-level strategic roadmap and thus crucially aligned to the business strategy and goals. A flexible and transparent AI backlog is essential to ensuring that businesses prioritise the most valuable projects in a non-biased manner. Business leaders must ensure that they populate their AI backlog with use cases that meet both business and technology needs and lay the groundwork for future investment.

The key steps to getting started with developing a prioritised AI backlog are:

1. Inspire. Educate stakeholders from across your business about the potential of AI (as well as some of its limitations) through proven case studies to demonstrate the benefits. Explore the art of the possible: stakeholders should be encouraged to put forward their own ideas, rather than being chased for them. This will naturally result in greater business engagement and investment – a key factor in achieving success.
2. Work backwards. Work backwards from the customer or business need and then assess how (and if) AI can be leveraged solve the problem. A common pitfall of AI use cases is that they start with the technology solution and not the business problem. ‘Context-aware’ AI use case have been shown to outperform other use cases by up to 27%.
3. Collaborative. Ideation involves diverse perspectives: data scientists, business stakeholders and domain specialists together in one room. The alternative is that key opportunities may be missed and divergence from business-critical needs is likely.
4. Alignment. Develop and align to a strategic vision for AI and apply this, along with the overall business strategy, as filters over your backlog. Prioritise only those ideas which align with this vision.
5. Focus. Prioritise your remaining backlog to focus on the use cases that will have the greatest impact. The key dimensions should be effort vs return on investment. Effort should be measured by the skills, resources, technical and data requirements. Data reliability and accessibility should be considered too.
6. Measure success. Determine how you will measure success and which metrics you want to use. Be wary of optimism bias – the mistaken belief that one’s chances of experiencing a positive event are higher than they are.  This can easily creep into AI projects. Bake the measurement of these metrics into the delivery of your initial projects, so these projects can later be used as strong business cases for future investment.
7. Review your backlog periodically as priorities and available information changes over time.

Once you’ve built your AI backlog, the next step is laying the groundwork for developing your first proof of concepts. To do this, we advise setting up an innovation hub: a fail-fast environment where innovation thrives and PoCs are rapidly deployed. This standardised approach to ensuring that PoCs are developed quickly and sustainably will feed a pipeline of PoCs that can be tested for suitability for becoming scaled, industrialized solutions.

Helping organisations build a bespoke AI backlog is part of Capgemini Invent’s AI Powered Innovation service offering. This post is the first of a series of blogs outlining our position on the critical success factors needed to unlock the value of AI: from how to take your first steps in your AI journey to full-scale industrialisation. The next blog in the series introduces the journey from proving the proof of concept’s value to industrializing the solution.

Anand Patel

Managing Consultant, Insight Driven Enterprise (IDE)

Anand leads IDE’s Customer and Digital services capability and has a proven track record in transforming Consumer Products and Retail organisations by enabling them to develop meaningful customer experiences through analytics and AI

Matt Rousseau

Senior Consultant, Insight Driven Enterprise (IDE)

Matt is a leader in IDE’s digital analytics community, with a background in customer insight and digital marketing. He has successfully supported clients in the retail, banking and utility sectors for nearly 7 years