Thought Leadership Series Issue 10.

2017

AI and the Future of Procurement

Sponsored by:
 Accenture posted a video on YouTube
earlier this year entitled, Artificial Intelligence: Delivering in the New, that
Virtual cognitive tools such as Accenture’s are starting are gain traction,
powered by artificial intelligence (AI) technologies. Indeed, AI has been a
provides a glimpse of what lies ahead for procurement organizations. It’s faint glimmer on the horizon for a while, but the technology has evolved
95 seconds of a digital procurement market intelligence advisor explaining to the point where innovative companies are operationalizing it to
how “she” can help your company buy materials and services at the lowest improve business performance and deliver cost savings and new revenue
price by supplying “accurate information on your suppliers, commodity opportunities.
pricing and access to market trends across a broad range of categories.”
In the realm of procurement, the benefits of AI can be significant. They
“I work with procurement analysts,” she says, “and help identify which include providing actionable, near-real-time insights about the cost,
suppliers we should be evaluating, who are the leading suppliers in the availability and lead times of materials, and operational risk of their
region, and what may happen that could impact future buying decisions, suppliers and the geographies they operate in. With the right leadership
such as potential market consolidation, and mergers and acquisitions. I and decision-making skills, this information can boost organizational
can search and extract relevant agility and supply-
information across thousands chain flexibility.
of digital sources and utilize
my contextual training on It’s not hyperbole
procurement market intelligence to suggest that AI
to quickly understand and sort has the potential to
information into categories.” rapidly transform
business processes
Not only is she thorough but and disrupt
she’s quick. “I can complete industries. If your
these tasks more than 10 times company is not
faster than a human expert with getting on the AI
my qualifications,” she boasts. bandwagon already,
it will be soon.
The digital advisor’s purpose
is clear: To help companies
“buy what they need to deliver
services and products at the
best price so that they can stay Accenture’s AI-enabled procurement market intelligence advisor can complete tasks “more than 10
competitive and grow.” times faster than a human expert with my qualifications.”

Thought Leadership Series 2

 The AI Imperative
A recent McKinsey & Company survey1 of 3,000 AI-aware C-level executives A recent Deloitte global survey of CPOs was a bit more sobering. It revealed
from around the world found that AI investment is growing fast, led by an acknowledgement that while their procurement organizations need
global tech giants, such as Google and Baidu as well as global automotive to become more digitally innovative, many lack staff with relevant digital
and financial services companies. Still, the McKinsey survey suggests that talent. The Deloitte survey results include:
AI adoption outside the tech sector is in the early, experimental stage of
development and few firms have deployed AI at scale. In fact, only 20% of • 60% of CPOs do not believe their teams have the skills to deliver on
respondents said they currently use any AI-related technology at scale or their procurement strategy
in a core part of their businesses. • 75% of CPOs believe that procurement’s role in delivering digital
strategy will increase in the future
Focus areas for AI span many corporate functions including the potential
to improve forecasts, optimize and automate operations, develop targeted • 65% of CPOs believe that analytics will have the largest impact on
marketing and pricing initiatives, and enhance the user experience, future organizational performance
according to McKinsey.
The most significant and immediate application of AI in procurement orga-
Hardly a surprise, the study found a correlation between tech-savvy nizations is to improve the forecasting accuracy of end-customer demand
companies that are deeply engaged in digital transformation and those for finished goods and material requirements from the supply chain. AI can
that are investing heavily in AI. The message is that companies should discern trends and patterns that are not evident using traditional forecast-
not delay jumping on the digital transformation bandwagon, including ing methods, which still rely largely on historical data. Using AI tools to ac-
the adoption of AI technologies. “Early adopters are already creating curately anticipate future sales trends, based on a variety of data sources,
competitive advantages, and the gap with the laggards looks set to grow,” it’s possible for procurement organizations to confidently order materials
McKinsey warns. to satisfy demand for as-yet-to-be-ordered finished goods.

There’s evidence that chief procurement officers (CPOs) see the value of Such accurate and repeatable projections of inventory will create a clear
stepping up to the AI challenge. A separate McKinsey survey found CPOs competitive advantage. Traditional forecasting sources are incapable of
believe that adopting digital procurement practices can deliver significant processing the vast quantity of data and analyze the number of relevant
business value, including the potential to increase annual savings by 40%, variables that AI systems can. Data such as product introductions, distri-
spend 30 to 50% less time on transactional sourcing, and reduce “value bution inventory, weather, seasonality, changes in customer perception,
leakage” by 50%. advertising campaigns, regional conflict and local labor strikes, media
coverage, and a variety of social sentiment databases can be harnessed by
AI systems to iteratively improve the performance of the supply chain.

Thought Leadership Series 3

 Assembling the AI Toolbox It’s About the Data
There are a number of technologies encapsulated by the term artificial Artificial intelligence systems are only as good as the data they consume.
intelligence. These include machine learning (ML) and deep learning A priority for procurement executives is to gain access to as many
(DL)—also called deep neural networks—natural language processing and sources of relevant data as possible to feed into their AI engines. Gartner
natural language understanding (NLP and NGU) and computer vision. AI has identified four categories of data and provides insight into how to
technology systems—the combination of various AI technologies and the maximize the value.1
essential raw material of data—include robotics, autonomous vehicles,
and virtual agents. Figure 1 captures McKinsey’s estimate of the relative Supplier data: Procurement applications and networks can be used
investment in AI technologies and systems. (Continued on page 5.) to gather supplier data efficiently by enabling suppliers to submit this
information on a self-service basis. Ideally, supplier information is shared
Machine learning received the most investment, although boundaries
across multiple buyers on a permissions-basis to avoid repetitive data
between technologies are not clear-cut entry and enable scalability for suppliers.
External investment in AI-focused companies by technology category. 20161
$billion Internal data: For spend analytics and supplier performance
management, data from internal systems—such as ERP, supply chain
Natural language
0.6—0.9
management, quality management and even CRM systems—can be
leveraged to provide data for insights and classification.

Subscription data: As reliable data becomes more valuable, credible data

aggregators becomes more relevant. Vendors such as Dun & Bradstreet
and Bureau van Dijk already provide financial information that can be
Autonomous
vehicles
Computer vision
2.5—3.5
used in supplier evaluations and scorecards. There are other types of
0.3—0.5 Machine learning
Multiuse and nonspecific applications
5.0—7.0
external data sources that should be explored, such as weather services or
commodity price indexes.

Public data: External public data is without doubt the largest and most
complex source of data. Different sites have different levels of reliability.
Smart
Even within sites, reliability can vary and the data can be unstructured.
Virtual
robotics
0.3—0.5 agents This is the area that puts the greatest requirements on the smart
0.1—0.2
procurement solutions and on the procurement analysts using them.
1 Estimates consist of annual VC investment in Al-focused companies, PE investment in AI-related
Understanding the technology is important to understand the reliability of
companies, and M&A by corporations. Includes only disclosed data available in database, and its alerts and recommendations.
assumes that all registered deals were completed within the year of transaction.

SOURCE: Capital IQ; Pitchbook; Dealogic; McKinsey Global Institute analysis 1.“Start Preparing Now for the Impact of AI on Procurement,” March 2017

Thought Leadership Series 4

 By far, the most popular and essential AI tool today is ML, of which DL devices—including internet of things (IoT) devices—come on line. Within
is a variant. As the name suggests, these are machines that can “learn” a few years, these billions of devices will connect to global networks and
without being programmed. Instead of providing explicit instructions— transmit data at ultra-high speed across 5G networks.
if X, then Y—ML algorithms process vast quantities of structured and
unstructured data to determine patterns that allows them to provide a As a consequence, annual global IP traffic is projected to almost triple,
statistically accurate answer to a question or to execute a task. DL systems reaching an annual run rate of 3.3 zettabytes per year by 2021 up from 1.2
need thousands of examples to recognize an object—a dog for example— ZB in 2016, according to June 2017 data from Cisco. That’s a compound
and must “look” at those examples hundreds of thousands or millions of annual growth rate of 24%. Not only will the quantity of data explode, but
times before getting it right, according to Gartner. the combination of M2M data and 5G’s low latency will make AI systems
such as autonomous vehicle viable, which will transform transportation,
ML and DL are enablers for natural language processing (NLP). The result is including land and sea logistics, and potentially air as well.
the evolution of sophisticated virtual cognitive expert advisors and digital
agents able to communicate conversationally with humans, which is a The Human Factor
much more pleasant user experience. Of course, applying the technology requires human leadership. Yes, the
tools of the trade can be assembled but to take full advantage of their
The more these advisors and agents engage with users and the data, potential requires leaders and managers who understand the value of the
the faster they learn and improve the quality of their responses. As they technologies and have experience applying the tools.
become more human-like in their ability to provide specific information
they become a more potent competitor of apps and websites—never mind Bottom line, procurement leaders must become AI-conversant so they can
humans—for how organizations digitally interact with their customers and onboard and direct the professionals who possess the skills to leverage
supply chain partners. With time, these advisors and agents will assume the tools. Without competent staff, it’s widely believed it will be nearly
responsibility for a variety of day-to-day functions. impossible for companies to build next-generation strategic sourcing and
procurement organizations. Perhaps the most important skillset for new
Access to the massive datasets required for AI systems is made possible hires will be data scientists and engineers who are experienced in applying
by recent advances in cloud computing, storage, and new processor AI and have domain experience.
technologies. For example, Google’s Tensor Processing Unit (and its
TensorFlow ML framework are optimized for AI applications, reportedly Not surprisingly, demand currently exceed supply for top-flight data
faster than graphics processors and CPUs, and more energy efficient. IBM’s scientists. A recent survey by Glassdoor ranked data scientists as the
Watson employs a cluster of 90 IBM Power 750 servers, powered by 2,880 “best” job in the US for the second year in a row commanding a median
3.5 GHz POWER7 eight-core processors and 16 terabytes of RAM. base salary of $110,000. Data engineers also ranked highly. Despite the
shortage of talent today, the supply of data scientists is expected to
The availability of data to feed these powerful systems will increase increase in the coming years to meet the demand.
rapidly in coming years as more and more machine-to-machine (M2M)

Thought Leadership Series 5

 At the same time, leaders need to build more agile organizations so they ACTION ITEMS
can make quick course corrections in response to increasingly dynamic Learn. Get up to speed on AI procurement tools by attending conferences,
supply chains and the emergence of new powerful digital marketplaces. studying best practices, and engaging with industry leaders and AI tool
They also face the challenge of keeping pace with the quantity and types vendors. There are plenty of open-source online resources available to
of data that they will be able to utilize in the future. explore.

In a recent interview, Graham Wright, IBM’s vice president of global Add AI to new-hire qualifications. Most procurement professionals have
procurement, identified three corporate imperatives for IBM’s own a background that does not include a concentration in computer science,
procurement transformation; two are outcome-driven and the third is an data science, or AI. To succeed in an AI-centric world, some new hires—
enabler.2 specifically procurement analysts—should have a working knowledge of
• Drive insights. “Enriching data for analytics and cognitive insights and experience with AI.
to make more informed decisions.”
Seek out new data sources. Catalog and analyze the types of procure-
• Amplify the talent. “Elevating the procurement capabilities and
ment data the organization uses now and identify new, external data
intelligence to ‘extraordinary’, and enabling people to drive the level
sources to acquire or subscribe to that will support your AI tools.
of change and influence on the business that can come from those
insights. By Bruce Rayner, Contributing Editor
• Cognify, automate and ultimately digitalize the processes.
“Try to think about those processes differently as you uncouple
them from the applications that sit on them today to truly move
things from analog to digital in a seamless responsive procurement
ABOUT THE SPONSORS
organization.”
TTI, Inc. is the world’s leading authorized distributor of
interconnect, passive, electromechanical and discrete
According to Wright, IBM is moving from reacting to customer demand
components. TTI’s product line and supply chain solutions make it
based on historic data toward predicting demand driven by dynamic fore-
the distributor of choice for customers worldwide.
casting, market insights and industry expertise. All three of his objectives
stand as worthy goals for any procurement organization that is setting out
TE Connectivity is a global technology leader offering unmatched
on the journey toward an AI-driven cognitive and digital future.
breadth in connectivity and sensor solutions.

1. “Artificial Intelligence: The Next Digital Frontier?” June 2017, McKinsey Global
Institute, McKinsey & Company.
2. Taraz Berezowsky “Q&A With Graham Wright, IBM’s VP of Global Procurement:
It’s Not the 80s Anymore,” Spend Matters, September 19, 2017.

Thought Leadership Series 6