Is Your Company Ready for Artificial Intelligence?

Overview

Companies are rushing to invest in and pursue initiatives that use artificial intelligence (AI). Some hope to find opportunity to transform their business processes and gain competitive advantage and others are concerned about falling behind the technology curve. But the reality is that many AI initiatives don’t work as planned, largely because companies are not ready for AI.

However, it is possible to leverage AI to create real business value. The key to AI success is ensuring the organization is ready by having the basics in place, particularly structured analytics and automation. Other elements of AI readiness include

  • executive engagement and support,
  • data excellence,
  • organizational capabilities,
  • and completion of AI pilots.

Key Takeaways

There is tremendous AI hype and investment. Artificial intelligence is software that can make decisions without explicit instructions for each scenario, including an ability to learn and improve over time. The term “machine learning” is often used interchangeably with AI, but machine learning is just one approach to AI, though it is currently the approach generating the most attention. Today in most business situations where AI is relevant, machine learning is likely to be employed.

The hype around AI is tremendous and has accelerated in the last few years. It is rare to read a business-related article these days that doesn’t mention AI.

The AI hype is being accompanied by massive investments from corporations (like Amazon, Google, and Uber), as well as from venture capital firms.

Because organizations often pursue AI without fully understanding it or having the basics in place, many AI initiatives fail. The AI fervor is causing companies to hurriedly pursue AI. There is a rush to capitalize on AI, but significant frustration when it comes to actually delivering AI success. AI initiatives are often pursued for the wrong reasons and many AI initiatives experience pitfalls. Some key pitfalls are:

  • Expensive partnerships between large companies and startups without results.
  • Impenetrable black box systems.
  • Open source toolkits without programmers to code.

The root cause for these failures often boils down to companies confusing three different topics:

  • automation,
  • structured analytics,
  • and artificial intelligence.

AI1

Despite the challenges, some organizations are experiencing success with AI. While the hype around AI is overblown, there are organizations having success by leveraging AI to create business value, particularly when AI is used for customer support and in the back office.

The key to AI success is first having the basics in place. In assessing AI successes and failures, the presenters drew three conclusions:

  1. There is a huge benefit from first getting the basics right: automation and structured analytics are prerequisites to AI.
  2. The benefits from AI are greater once these basics have been done right.
  3. Organizations are capable of working with AI at scale only when the basics have been done at scale.

GETTING THE BASICS RIGHT

The most important basics for AI are automation and structured analytics.

  • Automation: In most businesses there are many examples of data processes that can be automated. In many of these examples, there is no point having advanced AI if the basics are not yet automated.
  • Structured analytics means applying standard statistical techniques to well-structured data. In most companies there is huge value in getting automation and structured analytics right before getting to more complicated AI.

Examples of how businesses use structured analytics and automation include:

  • Competitor price checking. A retailer created real-time pricing intelligence by automatically scraping prices from competitors’ websites.
  • Small business cash flow lending product. Recognizing the need for small business customers to acquire loans in days, not weeks, a bank created an online lending product built on structured analytics.

BENEFITS WHEN THE BASICS ARE IN PLACE

Once the basics of structured analytics and automation are in place, organizations see more value from AI—when AI is used in specific situations.

AI2

Examples of how adding AI on top of the basics helps improve business results are:

  • New product assortment decisions. Adding AI on top of structured analytics allowed a retailer to predict the performance of new products for which there was no historic data. With this information, the retailer was able to decide whether or not to add the product to the stores.
  • Promotions forecasting. A retailer was able to improve forecasting of promotional sales using AI. Within two months of implementation, machine learning was better than the old forecasts plus the corrections made by the human forecasting team.
  • Customer churn predictions. A telephone company used AI and structured analytics to identify how to keep at-risk customers from leaving.
  • Defect detection. An aerospace manufacturer used AI to supplement human inspection and improve defect detection.

AI AT SCALE AFTER THE BASICS ARE AT SCALE

Once an organization proves it can work with automation and structured analytics at scale, it is ready for AI at scale. Readiness for AI at scale goes beyond completing a few AI pilots in defined but isolated areas of capability; the basics need to be in use across the business.

Before undertaking AI, organizations need to assess their AI readiness. To be successful, organizations need to be ready for AI. Readiness consists of multiple elements, including

  • executive engagement and support,
  • data excellence,
  • organizational capabilities,
  • and an analytical orientation.

Organizations often struggle with data excellence and organizational capabilities.

AI3

Click here to access HBR and SAS article collection

The Future of Planning Budgeting and Forecasting

The world of planning, budgeting and forecasting is changing rapidly as new technologies emerge, but the actual pace of change within the finance departments of most organizations is rather more sluggish. The progress companies have made in the year since The Future of Planning, Budgeting and Forecasting 2016 has been incremental, with a little accuracy gained but very little change to the reliance on insight-limiting technologies like spreadsheets.

That said, CFOs and senior finance executives are beginning to recognize the factors that contribute to forecasting excellence, and there is a groundswell of support for change. They’ll even make time to do it, and we all know how precious a CFOs time can be, especially when basic improvements like automation and standardization haven’t yet been implemented.

The survey shows that most PBF functions are still using relatively basic tools, but it also highlights the positive difference more advanced technology like visualization techniques and charting can make to forecasting outcomes. For the early adopters of even more experimental technologies like machine learning and artificial intelligence, there is some benefit to being at the forefront of technological change. But the survey suggests that there is still some way to go before machines take over the planning, budgeting and forecasting function.

In the meantime, senior finance executives who are already delivering a respected, inclusive and strategic PBF service need to focus on becoming more insightful, which means using smart technologies in concert with non-financial data to deliver accurate, timely, long term forecasts that add real value to the business.

Making headway

CFOs are making incremental headway in improving their planning, budgeting and forecasting processes, reforecasting more frequently to improve accuracy. But spreadsheet use remains a substantial drag on process improvements, despite organizations increasingly looking towards new technologies to progress the PBF landscape.

That said, respondents seem open to change, recognizing the importance of financial planning and analysis as a separate discipline, which will help channel resources in that direction. At the moment, a slow and steady approach is enough to remain competitive, but as more companies make increasingly substantial changes to their PBF processes to generate better insight, those that fail to speed up will find they fall behind.

Leading the debate

FSN’s insights gleaned from across the finance function shed light on the changes happening within the planning, budgeting and forecasting function, and identify the processes that make a real difference to outcomes. Senior finance executives are taking heed of these insights and making changes within the finance function. The most important one is the increasing inclusion of non-financial data into forecasting and planning processes. The Future of The Finance Function 2016 identified this as a game-changer, for the finance function as a whole, and for PBF in particular. It is starting to happen now. Companies are looking towards data from functions outside of finance, like customer relationship management systems and other non-financial data sources.

Senior executives are also finally recognizing the importance of automation and standardization as the key to building a strong PBF foundation. Last year it languished near the bottom of CFO’s priority lists, but now it is at the top. With the right foundation, PBF can start to take advantage of the new technology that will improve forecasting outcomes, particularly in the cloud.

There is increasing maturity in the recognition of cloud solution benefits, beyond just cost, towards agility and scalability. With recognition comes implementation, and it is hoped that uptake of these technologies will follow with greater momentum.

Man vs machine

Cloud computing has enabled the growth of machine learning and artificial intelligence solutions, and we see these being embedded into our daily lives, in our cars, personal digital assistants and home appliances. In the workplace, machine learning tools are being used for

  • predictive maintenance,
  • fraud detection,
  • customer personalization
  • and automating finance processes.

In the planning, budgeting and forecasting function, machine learning tools can take data over time, apply parameters to the analysis, and then learn from the outcomes to improve forecasts.

On the face of it, machine learning appears to be a game changer, adding unbiased logic and immeasurable processing power to the forecasting process, but the survey doesn’t show a substantial improvement in forecasting outcomes for organizations that use experimental technologies like these. And the CFOs and senior finance executives who responded to the survey believe there are substantial limitations to the effective of machine forecasts. As the technology matures, and finance functions become more integrated, machine learning will proliferate, but right now it remains the domain of early adopters.

Analytic tools

Many of the cloud solutions for planning, budgeting and forecasting involve advanced analytic tools, from visualization techniques to machine learning. Yet the majority of respondents still use basic spreadsheets, pivot tables and business intelligence tools to mine their data for forecasting insight. But they need to be upgrading their toolbox.

The survey identifies users of cutting edge visualization tools as the most effective forecasters. They are more likely to utilize specialist PBF systems, and have an arsenal of PBF technology they have prioritized for implementation in the next three years to improve their forecasts.

Even experimental organizations that aren’t yet able to harness the full power of machine learning and AI, are still generating better forecasts than the analytic novices.

The survey results are clear, advanced analytics must become the new baseline technology, it is no longer enough on rely on simple spreadsheets and pivot tables when your competitors are several steps ahead.

Insight – the top trump

But technology can’t operate in isolation. Cutting edge tools alone won’t provide the in-depth insight that is needed to properly compete against nimble start-ups. CFOs must ensure their PBF processes are inclusive, drawing input from outside the financial bubble to build a rounded view of the organization. This will engender respect for the PBF outcomes and align them with the strategic direction of the business.

Most importantly though, organizations need to promote an insightful planning, budgeting and forecasting function, by using advanced analytic techniques and tools, coupled with a broad data pool, to reveal unexpected insights and pathways that lead to better business performance.

As FSN stated, today’s finance organizations are looking to:

  • provide in-depth insights;
  • anticipate change and;
  • verify business opportunities before they become apparent to competitors.

But AI and machine learning technologies are still too immature. And spreadsheet-based processes don’t have the necessary functions to fill these advanced needs. While some might argue that spreadsheet-based processes could work for small businesses, they become unmanageable as companies grow.

PBF

Click here to access Wolters Kluwers FSN detailed survey report