Volume 01


Dancing to your algo-rhythm

The secret to creating enterprise AI tools that employees will adopt

Consumers are ready to chat Behavioral science can help connect with consumers

Let’s consoli-data Analytics solutions that

saved thousands of hours a quarter

Data, Insights & Action In conversation with Diana Schildhouse of Colgate-Palmolive



Building a global future with AI The convergence of analytics and AI has opened new horizons and drives innovation in multiple spheres. Today, digital transformation is accelerating the way we consume goods and serve the requirements of aging people as we enter an era of population decline across countries. At the same time, sustainable innovation is now a necessity. As a society, we must act in a way that sustains the human race and improves the quality of our lives. Artificial intelligence will be the key to addressing these challenges. Enterprises will save time and money through automation, improve the speed and effectiveness of decision-making, and increase revenue by finding new opportunities. Leaders must be prepared to re-engineer their decision-making to thrive in this changing world. We equip you with information, insight, and real-work examples to support this process with ai:sight, our new bi-monthly magazine. An amalgamation of ideas, ai:sight is about elevating value creation, delivering growth, and thriving in an ever-changing economic environment. We hear from the transformational individuals redefining how organizations operate and interact with the world. As agents of change, they bring unique insights and experiences. I hope you enjoy the read.

Dancing to your algo-rhythm

Employees will eagerly adopt enterprise AI that is designed to achieve their goals

Page 5

Consumers are ready to chat

Let’s consoli-data

How unifying enterprise-wide data created analytics solutions that saved thousands of hours a quarter

Rahul Desai on using behavioral science, backed by AI, to connect with consumers

Page 9

Page 11

Data, Insight & Action Diana Schildhouse discusses how data powers key decisions at Colgate-Palmolive

Co-founder & CEO Fractal Pranay Agrawal

Page 14


Dancing to your algo-rhythm

Enterprise technology projects often falter when internal users don’t adopt the new tools. Combining design thinking with behavioral science provides a powerful framework for success.

Enterprise artificial intelligence (AI) investment is accelerating fast. For instance, in its latest CIO and Technology Executive Survey, Gartner predicts global revenue will reach $62.5 billion in 2022, increasing more than 21% from 2021. Enterprises are driving that growth, with 48% of CIOs saying they have already deployed AI and machine learning or plan to do so in the next 12 months. However, ensuring the adoption of those new tools can be another matter. Gartner also points out that while enterprises are keen to experiment with AI, many struggle to establish the technology as part of their standard operations. It lists reluctance to embrace AI, lack of trust in the technology, and difficulties delivering business value from AI investments among the causes. Behind those figures lies a fundamental reality: enterprises invest millions of dollars in developing AI business tools. Once they’ve deployed those tools, many discover that most of their people are not using them. And no tool can deliver value if people don’t use it. So, where are all those good intentions going astray? In most cases, the issue can be traced back to early assumptions about what users really need. “People often assume they know their teams well, so when developing new AI tools, they don’t research internal users’ needs as rigorously as they would for an external client,” says Benis Kumar Moses, Director, Experience team at

Fractal. “That assumption leads to early-stage mistakes which proliferate during development and are eventually built into the finished product. Poor adoption is the result. Without a full understanding of how the technology meets the needs of its users, enterprises are planning their AI projects to fail despite their good intentions.” Even when enterprises realize a problem, their assumptions about users’ needs often prevent them from resolving the issues early on. Many enterprises assume that a better user interface, training program, or awareness-raising exercise will get more people using their new AI tool – but those theories are usually wrong. Only rigorous research can reveal the root cause of poor adoption and the steps needed to resolve it. That means going right back to the project’s discovery phase and engaging with stakeholders to get a deep understanding of their needs. It can be a costly lesson for many enterprises to learn. But applying a few simple principles to AI development can help them drive internal adoption and maximize the value of their investment. Resourcing deep, early-stage research is a must. Still, while this is common practice for external-facing AI projects, it’s often overlooked when developing tools for internal users. Typically, internal enterprise AI projects are developed by the IT, HR, and administrative departments – none of which have the research team, design team, or budget to understand



user needs before they start building products fully. After all, if internal users are unhappy with the tools they’re given to do their job, that will spoil the experience for their customers. It’s also crucial to think beyond the new tool’s technical output. An enterprise that is focused entirely on developing an AI algorithm to improve customer segmentation, for example, can inadvertently blind itself to the real problem it needs to solve. Key questions about why the enterprise wants to build that tool can help to reframe the issues. For instance, if the sales team isn’t using the current segmentation tool, why is that happening? Perhaps they have their way of segmenting customers and have no interest in using a segmentation tool. If that’s the case, the segmentation algorithm isn’t the problem. Behavior change is the issue that needs to be addressed and better segmentation is just one aspect of that. By reframing the problem, the focus turns to understand what matters to the sales team. Identifying their needs and helping them improve the way they segment customers – and then translating that into actions, including a better way of defining segmentation. Ultimately, the successful adoption of enterprise AI starts with putting the human being at the heart of the problem. After all, scientific evidence indicates that people’s choices are based on emotions in the ‘hot’ state of decision- making. When the person wants to justify those decisions, rational factors come into play later. A traveling salesperson, for instance, lives a life of uncertainty because finishing a job or closing a deal all depends on their client’s response. They spend most of their day in the field, waiting outside clients’ offices with all their sales collateral in the boot of their car. While they wait, they may be worrying about how to close a deal in time for the quarterly bonus or how to juggle their work and childcare commitments. The technical output or business value of the new app they’ve been asked to use is the least of their priorities; however, if the app helps them achieve that bonus and strike that

work-life balance, they’ll pick it up. Of course, the tool needs to create value for the business too. But to do that successfully, it’s crucial first to understand the user’s context, emotions, and motivations. It’s a simple equation: if the tool makes users’ life easier, they will adopt it; if it doesn’t, they won’t. “For the technology to deliver real value, the problem it’s intended to solve must be framed at the intersection of the enterprise’s need and the user’s need,” says Francesca Passoni, Principal Consultant, Experience team at Fractal. “That will avoid the need for top-down directives such as pushing people into training programs or introducing incentives to drive adoption. These strategies are often bolted on after deployment and they don’t work in the long term. Instead, look at the success of technology like Apple’s touchscreen devices: they don’t require a training manual or awareness-raising exercises because they deliver intuitive value for users. Enterprise AI tools enable human decision- making in the enterprise, not replace it, and they should be just as simple and intuitive for users to adopt . Fundamentally, the process of design thinking and behavioral science is about looking through human eyes and putting aside any preconceptions from a business, technology, or organizational perspective. Those perspectives are important but they come later – it’s important to first look at the issue through a human lens.”

Ultimately, the successful adoption of enterprise AI

starts with putting the

human being at the heart of the problem.

3 KEYS TO INTERNAL AI ADOPTION Deep, early-stage research is a must Allocate resources for deep early-stage research into users’ needs. View the project through a human lens • Understand the emotional needs and context of the people who will use the tool.

Follow the path of least resistance

People will naturally adopt a tool that makes their lives easier – if it doesn’t, no amount of top-down edicts will force them to use it.

• Think beyond technical output – consider the business impact you want from the technology. • Fo cus development efforts at the point where business needs and users’ needs meet.




Consumers are ready to chat Behavioral science, backed by artificial intelligence, can help connect with consumers

With a complete, dynamic picture of the customer’s journey in hand, the next step is to craft a bespoke experience that will connect with them. Customer Genomics enables the agility for informed, on-the-spot decisions by using powerful, self-learning AI models that predict the customer’s intent and recommend the best intervention. These recommendations are used to orchestrate targeted customer communications using the company’s preferred sales, marketing, and CRM tools. Meanwhile, a feedback loop ensures the system continually learns from the results of each action and reports full explanations of the reasons behind its decisions. By integrating these capabilities with their existing technology investments, enterprises in any industry can provide a ‘Next Best Action’ framework for engagement. A leading healthcare insurer in the US, for example, found that smarter targeted personalization has improved patient outcomes, customer satisfaction, and engagement – while reducing the number of communications it sends. A top US retail bank uses self-teaching algorithms to democratize data and automate decision-making across the enterprise with exhaustive reporting. This helped the users ensure the models aligned with their KPIs and metrics. Meanwhile, a global asset management firm uses the platform to understand how its financial advisors behave under different market conditions and refine its conversations accordingly.

In the media, entertainment, and communications sector, technology is helping to minimize churn. Enterprises use intelligent models to identify the customers they’re at risk of losing and drive engagement that enriches their experience with the brand.

These examples highlight the growing importance of emotionally intelligent

engagement in every sector. Combining AI with behavioral science in a modular solution provides a flexible way for enterprises to drive customer interaction at speed and scale, meeting individuals at a point where the relationship can move forward.

lack of interconnected data, decision siloes, lack of insight agility, and lack of orchestration. Let’s start with interconnected data and emotional insight. Most major enterprises have grown to include several lines of business, each with its own set of customer data. For example, a bank has separate departments to manage credit cards, debit cards, auto loans, mortgages, digital experience, etc. Together, that data could provide a comprehensive view of the customer. Instead, it often sits in silos around the enterprise. In addition, traditional data models don’t capture dynamic influences – the context, emotions, preferences, and bias that drive customer decisions – so even connected data often provides an incomplete picture. Enter Customer Genomics 3.0, a Fractal designed solution that resolves these challenges by integrating data harmonization with our behavioral sciences toolkit. The harmonization layer quickly stitches data from across the organization to create a single view of each customer, including products, transactions, and browsing history. Behavioral science tools work with that data and design gamified labs to identify the emotional levers behind those decisions.

Rahul Desai Client Partner Fractal

Leads the firm’s AI practice of Algorithmic Decision Making including Customer Genomics, Fractal’s flagship solution for ‘Next Best Customer Action’. He is interested in disrupting the status quo to drive deep impact through contextual problem solving and data science at an enterprise scale.

Personalized engagement is the key to customer satisfaction and loyalty in the age of digital transformation. Enterprises like Amazon, Facebook, and Netflix have shown the world what can be achieved with intuitive content and communication that fits each customer’s habits, moods, and preferences at any point in their journey. But many Fortune 500 enterprises today face challenges in achieving those levels of proactive and personalized engagement. Blocking their way are typically four obstacles:




Let’s consoli-data How unifying enterprise-wide data saved thousands of hours a quarter

how the new solution ranks relative to others. It also identified power users who could be proactively approached for more detailed feedback regarding modifications to improve usage and usefulness. A single source of truth As a result, the solution is now the single source of truth for financial results and diagnostics. Standardized definitions and business rules ensure that users across functions and geographies align with consistent metrics. Older, ad hoc reports have now been decommissioned. Today the solution with a monthly usage rate of over 95 percent is one of the most highly adopted within the company. “The solution is cutting edge,” said the general manager of property insurance at the company. A product line leader for auto insurance shared, “Managers now have access to information they didn’t have before. This saves thousands of hours a quarter, if not a month.”

IN NUMBERS Data from over 300 reports was consolidated into 15 guided solutions Information was instantly available to 500+ users Registering over 14,000 views of the solutions in the first six months Uncovering over 100 new, untracked, unreported metrics Sustained impact through a monthly usage rate of over 95% among its core user group

Rapid prototyping to launch

A leading US-based property and casualty insurer sought to revamp its entire reporting landscape by recognizing that prompt, data- driven decision-making was integral to its future success. The company was data-rich but information poor. Multiple reports provided only a static view of the business and no insight into actionable recommendations. Moreover, reports were not always accessible or available on time and inconsistency in definitions and business

The solutions were iteratively developed with an agile framework before being operational in the client environment. From discovery to viable product, the entire process took three months. It was then launched to the insurer’s wider audience over the next three months. By bridging various information gaps, data from over 300 reports were consolidated into 15 analytics solutions made available through a visualization suite to generate faster, more descriptive, and actionable insights. This information was instantly made available to over 500 users. It had been viewed over 14,000 times in the first six months after going live. With the new solution, over 100 previously untracked metrics were now being reported. The adoption and popularity among users were its biggest marks of success. This was measured through an automated usage tracking report that collected continuous feedback. The report not just captured usage information relating to the new system but did so for all Tableau reports in the client’s ecosystem. This gave real-time feedback on usage trends and

rules led to conflicting information. This impacted teams across product

management, underwriting, pricing, claims, sales, and distribution, who required reliable information to function. To address these challenges, the insurer engaged Fractal. After reviewing key business needs and the current state of reports and dashboards, a roadmap was built to implement a range of analytical solutions. Drawing from best practices in building effective and intuitive business intelligence solutions, a comprehensive data visualization system was proposed with fast prototyping and wireframes. These gave users a clear picture of the solutions

and allowed them to provide immediate feedback while they were being built.



I N T E R V I E W Data, Insights & Action Diana Schildhouse discusses what data means and how it powers key decisions at Colgate-Palmolive

Diana Schildhouse Chief Analytics and Insights Officer at Colgate-Palmolive

quantify the value of the recommendations coming out of our work and align those with the business to determine which are feasible. By tracking that, we can focus on both business adoption as well as driving real, measurable value from analytics. Can you tell us a little about your analytics journey at

Please tell us about your primary goals for Colgate-Palmolive I’m very focused on enabling data-driven decision making for the organization, giving everyone across the business access to the tools and the data they need to make effective decisions. Enabling data is key – it’s about instilling a data culture and ensuring everyone understands what data means and how it applies to their role. A key component of that is building data literacy so that everyone speaks a common language. I’m also focused on bringing

With more than 20 years of experience in data science and advanced analytics, business intelligence, strategic planning, and consumer and market insights, and now as chief analytics and insights officer for Colgate- Palmolive, Schildhouse has learned the power of data storytelling and how this applies to the consumer-packaged goods industry. Here, she tells us more about her strategy.

Colgate-Palmolive and how you prioritize the workstreams?

We started with an outside in perspective. We took time to understand other proven success stories and familiarize ourselves with the learnings that have happened along the way in this space.

discipline to measuring the value of our analytics work. For example, we



“You also have to articulate your vision and strategy and where you’re headed very clearly and often... This means speaking in plain terms to demystify data and analytics and make it something that everyone understands.”

We then looked at how this might apply to our journey, while recognizing that a cookie-cutter approach cannot achieve successful analytics transformations. We are operating in a ‘build the plane while you fly it’ mode, so we aren’t waiting for our data to be in a perfect state before we move forward. This is where some companies stumble – by prioritizing perfection over progress, there’s a high probability that you’ll never get started. We’re also adopting a use-case-driven approach rather than a tech-first strategy. That means we

focus on the different domains across the business where analytics can drive the most value. For example, that might be revenue growth, pricing, trade promotions or marketing and media effectiveness. We choose a couple of priorities and then focus on building, embedding, and scaling. How do you organize analytics teams for success? What are the skill sets typically needed? The first t hing I’d say is that the partnership with the IT organization is critical. For us, that means having folks working on data strategy and data operations that act as

a day-to-day link with our IT experts. They make sure that we have the data we need to do the type of analytics work that we want to do. Of course, you cannot underestimate the importance of roles like data engineers and data scientists, and they need to work together in an ecosystem. We also need analytics translators for adoption to serve as a link between data scientists and the business teams. To succeed, we need to articulate what data means, how to use it, and how it solves key problems. We’re also looking to foster important attributes, such as curiosity, across our entire team. Curiosity is important

in terms of problem-solving and around best-practice approaches to learning, understanding what’s new and evolving in the analytics space, and how different approaches might set us apart. Storytelling is also crucial. It’s never been more important for everyone within an organization to connect the dots in their work and weave it together to tell a compelling story that spurs action. The most successful analytics teams can clearly articulate what their work means and how it can impact the business.

The greatest successes within analytics come from embedding that work into the rhythm of day-to-day business operations. Another is data storytelling. Of all the case studies I’ve seen around analytics transformation, as well as the transformations I’ve led myself in prior roles, the ones that are most successful are those that succeed at changing behaviors and getting those behaviors to stick. It’s about fostering a business-first approach that can be enabled through storytelling – especially as we start to link many different types of data. It’s not just about syndicated data or

What competitive advantage will your focus on analytics provide? What potential will this open for the business? There are two areas where we are focused that I believe will set us apart. The first i s that business link. Everything we do within analytics starts with a business question - rather than a tech-first approach. We have to deeply understand what decisions our business partners need to make, when they need to make them, and what priorities are most important to their success. Then, we can help to solve some of those through analytics.



market share data, but also first-party data

skepticism from the outset. It’s also effective to find the curious, interested champions and with whom you can partner for ‘quick wins’ before scaling up. You also have to articulate your vision and strategy and where you’re headed very clearly and often, not just to your team and your analytics organization but also to the entire business. This means speaking in plain terms to demystify data and analytics and make it something that

business and the analytics space will continue to evolve. We have to constantly look at what’s happening externally to make sure that we’re still taking the most relevant approach for our business. We need to stay current with emerging approaches, data sources, and technologies. We are most excited about more predictive and prescriptive analytics for the future and using AI to automate decisions that require human effort. However, it is always with the approach of tying it to a business question, then building, showing value, and scaling.

and data partnerships with retailers. Being able to mine that data, draw the relevant insights and spur action is key to real scalable success. Are there specific adoption challenges that companies like yours face? And what practical steps can be taken to overcome them? Change management can be a big challenge – it’s the case in any organization where people are comfortable working in a particular way. Building trust across the organization is incredibly important to solving adoption challenges. That starts with really taking the time to listen and understand your stakeholders’ priorities and their business problems so that we can address the most important areas for them. I invest in building relationships, whether that’s through one-on-one meetings, lunches, or informal chats. By communicating effectively and, most importantly, by listening, you can build trust from the beginning and therefore overcome a lot of

CSC:Miami Content Supply Chains must be forensic in their detail. Television broadcasters have long relied on instinct,

everyone understands. We have our analytics

strategy on a single page at Colgate, including all of our strategic pillars. It acts as a kind of north star that can be referred to constantly to communicate why we’re doing what we’re doing, and to make sure that we’re focusing

market knowledge and spreadsheets to forecast TV viewership - but instinct needs to partner with information; market knowledge is never enough; and spreadsheets are no way to excel. As witness to these challenges, Fractal undertook its own detective work. By combining AI, data engineering and user-centric design, Fractal created an industry-first TV forecasting system for Europe’s leading media and entertainment company. The result? Up to 30% improvement in forecast accuracy. Fractal: perfectly targeted and timed TV, no drama.

and staying on track with what’s most important.

What does the future hold? How do you expect your strategic objectives to evolve in the coming years? We have our strategy, and we know where we’re headed, but we also recognize that the


Copyright 2022 Fractal Analytics. All rights reserved.

Page 1 Page 2-3 Page 4-5 Page 6-7 Page 8-9 Page 10-11 Page 12-13 Page 14-15 Page 16-17 Page 18-19 Page 20

Made with FlippingBook - PDF hosting