Retail

5 ways artificial intelligence is transforming how companies generate customer insights

March 23, 2023 | White Paper | 16-minute read

5 ways artificial intelligence is transforming how companies generate customer insights


Amidst a slew of business transformation efforts in recent decades, the process by which brands generate customer insights (CI) has remained relatively stable: They have questions about their customers (who they are, what they want, how they prefer to interact), commission CI professionals—internally or externally—to answer those questions and then change business practices accordingly. 

 

Now, if you believed the early hype, artificial intelligence (AI) was on the cusp of revolutionizing the CI world by delivering faster, cheaper and better business insights. CI professionals (those responsible for producing intelligence and overseeing the CI process) would either be unemployed en masse or relaxing on the beach while an army of algorithms made their jobs easy.

 

But the culling of the CI profession failed to materialize because AI’s early solutions were overwhelming, shallow, complicated and unreliable. 

 

Fast forward to today, and most people working in CI have abandoned the idea that AI will fundamentally transform how they do their jobs, let alone render them obsolete. They’re wrong. AI is already transforming how companies generate consumer insights—if one knows where to look.

 

But first, what do we mean by AI?

 

Artificial intelligence (noun): Any time a computer touches any part of the CI process without direct human intervention

 

Between the hype and the disillusionment that followed, software companies have been steadily developing AI-based tools capable of transforming every step of the insights-generation process. So, if we stop thinking about AI as a singular technology and start using the definition above, then we see clearly that AI has infused every step of the CI process.

 

A new paradigm for generating customer insights



Advances in AI mean companies can generate actionable insights in dramatically less time. But not only has AI altered how companies perform each step of the insight generation process, it’s also fundamentally transformed the process itself. A new, AI-enabled insights-generation framework looks something like the figure below.

For an ever-growing set of common questions, we’re no longer gathering data; we’re leveraging data that exists. We’re not manually conducting analysis and cooking up possible implications, we’re validating outputs from tools that do that work for us. And finally, we’re not driving change, we’re jumping straight to action. For novel and more complicated questions, where AI isn’t capable of shortcutting the process (yet), it’s still compressing time-to-insight from months to days and weeks.

 

However, barriers remain. For all that AI can do, AI can only go where a human points it. Orchestrating an end-to-end process for generating insights still requires humans, which explains why the customer insights orchestration role has grown in importance and represents rich pasture for motivated CI professionals. Fear not, CI professionals: There’s still plentiful need for those with deep industry and domain expertise to identify high-value opportunities to incorporate AI into research processes and to make sure they’re vetted, funded, implemented and evaluated.

Generating AI-enabled insights requires an organizational transformation



Transforming a company’s insights generation function by incorporating AI requires active coordination across the enterprise. Like any business transformation, it won’t happen without building the right organizational capabilities. While this transformation may vary from one company to the next, there are three key steps all companies must undertake to get started.

  1. Develop an overarching vision and strategy for the insight program. Without defining a vision and roadmap for how a company’s insights program must change, simply deploying AI tools willy-nilly is bound to fail. We recommend companies start by documenting and prioritizing the information they’re seeking to learn. They’ll be aided in this endeavor by creating (or shoring up) a repository for their existing insights and identifying “quick win” opportunities to build momentum for AI-enhanced insights.
  2. Consolidate the data and technology needed to create insights. Insights live on a steady diet of data. For companies to get the most from their investments in AI, they must have a comprehensive data strategy—both for locating and linking their own data and for acquiring new sources thereof. Moreover, tools evolve quicky; companies serious about supercharging their insights using AI must first empower someone to discover and vet new technology and, second, create a more agile relationship with IT so that tools can be tested and implemented at commensurate speed.
  3. Enshrine the right people and processes required to execute efficiently. Many of the barriers impeding adoption of AI stem from traditional walls between the insights and analytics functions. To move with the requisite speed and nip ever-present turf wars, these walls should become more permeable. New roles will also need to be invented, including AI transformation lead, tool experimentation lead and insights scrum master. No longer can companies be happy asking questions in Q1 only to get what they need to take action in Q4. They’ll need to incorporate an agile sprint approach to insight generation, whereby even if action isn’t immediate, they’re learning and acting progressively. And finally, incentives must be reworked to reward trying new things, even if they don’t always work. Without this, inertia and the perceived safety of the status quo will carry forward. 

Not only is AI already here, but it’s already having a huge impact on the world of customer insights. Successful brands of the future will be those that figure out how best to harness it. If you aren’t taking steps in this direction now, you’re putting your present and future business at risk. It’s a complex process to master, but taking the three steps outlined above represent a good start.



About the author(s)