Artificial Intelligence (AI), and particularly Machine Learning (ML), will change the game for pricing. But AI’s just the tip of the iceberg. You need first to make sure that the foundations of your business – your pricing strategy and implementation plan, data, workforce, etc. – are solid so you can introduce AI constructively, as the authors explain.  Peter Vermeire (peter.vermeire@pwc.com) is a Partner in the Management Consulting practice of PwC Belgium. Stijn Cottenie (stijn.cottenie@pwc.com) is a Manager in the Management Consulting practice of PwC Belgium. Louis de Liedekerke (louis.de.liedekerke@pwc.com) is an Associate in the Management Consulting practice of PwC Belgium. 

The Pricing Advisor, April 2019

Artificial intelligence may be a game changer for pricing, but don’t expect it to be a panacea on its own.

There is a lot of expectation surrounding artificial intelligence (AI), some consider it to be overhyped.

Nevertheless, it is widely accepted that AI technologies will be the most disruptive over the next decade. Interest in AI is reflected in our Global CEO Survey^[1] which found 85% of CEOs agreeing that AI will significantly change the way they do business in the next five years, even if AI’s penetration in companies is not yet impressive (Exhibit 1).

Exhibit 1: Corporate AI initiatives worldwide

The increasing importance of AI is due to several factors, including data proliferation and continuous technological improvements (processing power and storage), leading to the democratisation of AI and massive investments in these tools and technologies.

In our broad definition, AI is a collective term for computer systems that can sense their environment, think, learn and take action in response to what they are sensing and their objectives. Forms of AI in use today include predictive models, autonomous vehicles, cognitive computing, chatbots and smart robots, among others.

AI works in four ways:

• Automated intelligence: Automation of manual/cognitive and routine/non-routine tasks
• Assisted intelligence: Helping people perform tasks faster and better
• Augmented intelligence: Helping people make better decisions
• Autonomous intelligence: Automation of decision-making processes with no human intervention^[2]

Investments: Europe is lagging behind in terms of investments in AI ($3 to $4 billion) compared to the US ($15 to $23 billion) and Asia ($8 to $12 billion) (2016).

Business areas most affected: The biggest potential impact is on supply-chain management/manufacturing and marketing and sales. Roughly 2/3 of the entire AI opportunity.

AI will be a game changer for pricing

Rational pricing relies on data. Given the ever-increasing amounts of data available, conventional engineering solutions will soon no longer be up to the job. AI technologies allow organisations to use data to adapt to new situations and solve problems more effectively. Machine learning (ML) – an AI application – in particular has the potential to benefit almost every industry, as it offers the ability to extract certain knowledge and patterns from series of observations. ML is based on the idea that we can build machines that can process data and learn on their own, with no need for constant human supervision.

What are the applications, benefits and pitfalls for the pricing field? We identified three main pricing areas impacted by AI:

• Price Optimisation
• Discounts and rebates
• Product Portfolio

Price Optimisation

Almost all companies struggle to determine the fair price for a product or service, how customers will react to different prices and how to extract the maximum value for each customer segment in a market. This is where ML comes into play. It gives companies the opportunity to optimise prices, their pricing strategy and managers’ effectiveness.

Setting prices accurately is difficult mainly because of the complexity and proliferation of the available internal and external data that needs to be analysed. ML can accurately analyse large amounts of data in a short timeframe and derive business-guiding insights from it. It can help companies maximize profits via value-based pricing and optimal price tiering. It also allows decision makers to understand customers’ willingness to pay for a product and their reactions to different pricing strategies.

ML helps firms build a more rational global pricing strategy by gathering data from multiple sources: historical transactions, win/loss analysis, the competitions’ price levels and contextual data (e.g., what does the customer say about our product on social media, overall market trends, etc.). By looking at such prescriptive and predictive data, leaders can take better and faster decisions.

Let us take algorithmic dynamic pricing, a popular pricing strategy, as an example. It adjusts prices for each customer automatically based on data to quickly adapt to changes in the market and improve profitability. Although companies in the airline and hospitality sectors have long used machines to set their prices, much has evolved in the field of pricing systems. Rather than being rule-based programs, they now predominantly use reinforcement-learning, where the logic of deciding a product’s price is no longer within a human’s control; algorithms constantly collect and analyze customers’ feedback and reactions to pricing – often in real time – and use this as input to optimize the model.

In our experience, companies able to leverage the power of AI in price optimization can increase net sales by up to three percent on identical volumes.

Discounts and rebates: How can a firm define an efficient promotion strategy? What discount should it give to customers to improve profitability? What is the return on investment (ROI) of historical promotions? With AI, companies can optimize promotions using accurate analysis and predictions.

Though very costly, promotions represent a crucial means of increasing brand awareness and sales. Sales promotions ROI measures the impact of promotions on sales. However, ROI is a complex exercise, as you must isolate the impact of promotions based on the several factors that simultaneously influence sales (see Exhibit 2). ML can identify a sales increase/decrease by, for example, measuring the impact of both cannibalization (through category-SKU regression) and the discount. This way, AI helps leaders calculate both a promotion’s – either real or potential – ROI and penetration.

Exhibit 2: The effect on sales of a trade promotion is seen across the product portfolio and throughout time

AI can also help in building a promotion strategy. By analyzing data (customer activity, information, subscriptions, competitors, purchase and transaction history, etc.), AI algorithms can perform accurate marketing segmentation and target customers to build loyalty or identify churn behaviors. AI enables companies to deliver more targeted, profitable (fixing the ideal discount rate) and sales-driven promotions.

Product portfolio

Extracting more revenue from their existing client base is the number one lever companies should look at to grow their top line. By analyzing different data sources, AI algorithms make intelligent and personalised suggestions to sales reps during the sales process on what additional (cross selling) or more profitable (upselling) products and/or services a client could be interested in.

By being able to process data from across business units, channels and geographies, algorithms can make more accurate recommendations than any human could.

Main challenges for AI implementation

Despite its significant advantages, AI should not be seen as the golden bullet that will easily solve all the pricing problems a company is facing. Putting AI’s technological capabilities aside (platform to run ML, predictive models, etc.), embedding AI in a company leads to three major challenges related to strategy and organization, data and people.

Do not get too excited and have a look at the main challenges:

• Strategy & Governance
• Data
• People

Setting a pricing strategy and implementation plan

The most challenging problem is organizational. Though we are convinced that this new technology will become an essential support for business decisions, humans will still structure the way it is integrated and make final judgments. In that sense, for AI to deliver its full potential in the field of pricing, it is essential that a company has:

(1) A clear pricing strategy: taking into account the company’s needs and direction. As Sizing the Prize^[3] puts it, with a clear vision about its business problems, a firm will be able to more accurately identify the pain points that AI and automation could help it address.

(2) Set priorities: Instead of jumping headfirst into AI investments that might not be right for the company’s business goals, it might pay to start small. For instance, trialling AI techniques via short, six-week sprints is often a better bet than investing an enormous amount of money in extended AI projects. Taking an agile approach and adopting a test-and-learn philosophy helps glean insights from mistakes.

(3) A strong governance process: Strong data maintenance and governance processes must be put in place, focusing on both the beginning —where data’s gathered and how data efforts are organised —and what follows — where the results of AI models are implemented into workflows.

(4) Trust in AI: Trust and transparency are critical to convince people to use what AI models propose^[4].

AI’s cool, but it requires high-quality data

Experience has shown that only the best in class are able to use data to steer strategic and commercial decisions. This is mainly due to lack of accessibility and faith in data quality. To implement AI for pricing, a number of issues need to be addressed:

(1) Data gathering: Understand the data currently available and establish how it can be acquired and organised in a comprehensive way. Data must be transparent and comply with regulatory processes such as the EU’s General Data Protection Regulation (GDPR).

(2) Storage (technology): Organize data by storing and integrating it using a cloud solution, data lakes or other technologies.

(3) Data management: Once the previous points are in place, continual verification and improvements where necessary are key.

The right people on board

All too often, companies embark on AI projects only to end up with a very complex AI algorithm that is of no real use to their business. Best practice is to create three-levels of AI-savvy employees^[5] who understand AI:

(1) Citizen users: Teach people how to use the company’s AI-enhanced pricing applications, support good data governance and get expert help when needed.

(2) Citizen developers: Turn a more specialized group into line-of-business professionals who can identify use cases and data sets and work closely with AI specialists to develop new AI applications.

(3) Data scientists: Define a small but crucial group of data engineers and data scientists who will create, deploy and manage AI pricing applications.

To conclude

AI, and particularly ML, will change the game for pricing. By analysing different data sources, ML helps company leaders more effectively and efficiently set their products’ price and create and effective pricing strategy, calculate sales promotions ROI and make more accurate customer segmentation. It also allows for more intelligent and personalised suggestions as to which additional (cross selling) or more profitable (upselling) products and/or services a client could be interested in.

But AI’s just the tip of the iceberg. Most effort should be put into extracting, cleaning, normalizing and organizing data. Remember, embedding AI in an organisation takes time and effort. You need first to make sure that the foundations of your business – your pricing strategy and implementation plan, data, workforce, etc. – are solid so you can introduce AI constructively, beyond the hype.

1. https://www.pwc.com/gx/en/ceo-survey/2019/report/pwc-22nd-annual-global-ceo-survey.pdf ↑
2. http://usblogs.pwc.com/emerging-technology/briefing-ai/ ↑
3. https://www.pwc.com/gx/en/issues/analytics/assets/pwc-ai-analysis-sizing-the-prize-report.pdf ↑
4. https://www.pwc.com/us/en/services/consulting/analytics/artificial-intelligence/what-is-responsible-ai.html ↑
5. https://www.pwc.com/us/en/services/consulting/library/artificial-intelligence-predictions-2019.html#section2 ↑