Skip To Main Content
Illustration of a person holding a group of colourful geometric shapes.

Building a digitization strategy for your company

A digitization strategy is a crucial part of your organization’s success. Let us show you how to use digital technologies and data management to improve business outcomes.

This article is part of our Mastering Data series which examines the digitization underway in Canada’s economy, why it’s important, the data governance issues it creates, and how to address them. It also looks at the role you can play as a CPA in guiding your organization through the transition.

Read this article to learn:

  • what is a digitization strategy
  • why a digitization strategy is necessary
  • strategic objectives of digitization
  • considerations for a digitization strategy
  • CPAs as change agents

What is a digitization strategy?

Digitization is an enterprise-wide, ongoing effort to use artificial intelligence (AI) and other digital technologies to achieve business results. With the massive increase in digital technologies over the past decade, organizations across all sectors are embracing digital transformation.

An effective digitization strategy can ensure that trustworthy digital data and technologies are accessible in real time to support business leaders and data scientists across the organization in adding value. A digital strategy will outline the priorities, deliverables, outcomes, and timelines in your organization’s digital journey.

Why a digitization strategy is necessary

In their seminal book, The AI Ladder, IBM senior executives Robert Thomas and Paul Zikopoulos reported on learnings from over 30,000 AI engagements around the world. They assert that organizations need to build a thoughtful and well-architected approach to AI in order to succeed. The same holds true for your organization’s broader digitization effort.

A well-defined and credible digitization strategy will send the right signal to all branches in the organization, to its supply chain and to its customers, creating a common understanding of the objectives for digital initiatives. Such a signal will support the creation of trustworthy data value chains (the process of turning raw data into something of value), which are essential to digital initiatives. It will also set expectations regarding issues such as data hoarding, data complexity, talent scarcity and a lack of trust in AI systems.

Married with a solid corporate data policy, articulating the rules for data management and governance, a digitization strategy can align cross-organizational efforts.

Strategic objectives of digitization

Across organizations and economies, digitization projects have resulted in increased efficiencies, enhanced decision making and predictive capabilities, real-time reaction to events, and the creation of new digital products and services.

In 2019, the World Intellectual Patent Office reported more than 40,000 AI-related patent applications. In July 2020, the GitHub open source software development platform listed more than 34,000 AI projects underway.

Here are a few examples that could drive a discussion with your peers on the potential benefits of creating a digitization strategy

Customer service

  • increase consumer retention rates by automating call center activities with conversational assistants to handle routine calls
  • accelerate processing of claim adjustments by using algorithms that are trained on images of vehicle accidents
  • make purchase recommendations to online customers


  • improve forecasting ability through real-time data access
  • move financial planning and analytics from manual to automated processes
  • improve forecasting and scenario analysis through the use of pre-built AI planning tools that access real-time data on customer behaviours
  • use pre-built AI planning tools for HR staffing and compensation analysis, demand and inventory management, customer profitability, and planning promotions and sales

Risk and compliance

  • get real-time compliance advice on banking and tax regulations through trained AI recommendations
  • identify fraud through tracking data sources such as customer data, real-time transaction data, identification of devices processing transactions, etc.
  • identify and set aside spam emails or inappropriate comments on social media
  • manage supply chains and inventories

IT operations

  • monitor logs and detect intrusions through prebuilt applications
  • predict or detect component failure
  • predict capacity surges

Considerations for a digitization strategy

Here are some considerations when developing and implementing a digitization strategy:

Get senior management buy-in
Most would agree that a systematic approach is needed to integrate digitization into strategic plans, establish the right policies and design an appropriate information architecture across the organization. The challenge is that upwards of 80 per cent of business leaders do not understand the data and infrastructure required for successful adoption of AI technology, according to a research report by MIT Sloan.

The first step should therefore be to inform and get buy-in from senior management and the board of directors. Many organizations are already managing projects and initiatives that are in line with digitization. What is lacking is the commitment to do what is needed to set digitization as an overarching objective for the organization. 

Establish a corporate data policy
The second logical step would be for the organization to develop and implement its corporate data policy. A data management policy sets the rules and accountability regarding:

  • data ownership and tagging
  • data sharing and access
  • data residency
  • the overall accountability framework

This is the base, the foundation, that your organization needs in order to design its own digitization strategy.

Develop a budget
Many experts recommend treating AI projects as operating expenses on the understanding that most organizations will choose to train, test, and run algorithms in the cloud. This allows the organization to pay only for the computational power it needs when it needs it.

AI and machine learning teams are generally small and often use open source software when designing solutions. This, however, does not hold true for all sectors. For example, organizations engaged in the manufacturing or resource extraction sectors may require a capital budget to purchase digital technologies such as sensors, robots, human machine interface (HMI) equipment or additive manufacturing equipment.

Work on quick hits
A solid digitization strategy will often include an initially limited but growing number of small projects spanning the entire organization. Ideally, each member of your senior management team should be willing to experiment with AI in order to solve problems and improve performance in their respective units. This way, digitization can take hold in a culture of iteration and experimentation, with small agile groups working on short projects, three to five weeks at a time.

The expectation should be that many projects may fail at the beginning. The right mindset is to learn from fast but safe failures. That said, the digitization projects that succeed will likely generate noticeable outcomes for your organization.

Focus on problems that leaders want to solve
Identify problems that are aligned with business unit objectives and that can be solved with better insights through data. Examples could be to make pricing for products and services simpler; predict lifetime value of assets; or automate first-line customer service. In identifying projects, engage different business units, subject matter experts, and data scientists, to ensure that AI and/or digital technologies can contribute to a solution. Establish clear metrics to evaluate success

Create an initial inventory of data sources
Once small projects have been identified, work with business line leaders to identify data sources to be included in an initial data inventory. The inventory should include metadata for different data sources as well as sources of data residing outside the organization if datasets are not large enough.

Of course, data must be in digital form but older formats such as paper records, photographs, and x-rays are now usable. They can now be digitized, transformed into digital objects, and labelled with unique identifiers.

Don’t reinvent the wheel on new algorithms
AI is the science of teaching programs and machines to complete tasks that normally require human intelligence. This includes techniques for building intelligent software that learns from data and acts on what it has learned. You can take advantage of this by understanding the following:

  • In most cases, you will not need to make considerable investments in AI R&D in order to digitize your organization’s operations. Large AI firms and cloud service providers may be able to provide this service.
  • Open source projects generating state of the art algorithms (SOTA) are freely distributed in open source libraries. Tens of thousands of algorithms have already been developed and tested in sectors such as finance, accounting, inventories, logistics and customer interactions.
  • The most important investment to make relates to data. It is to locate, describe and grade trustworthy datasets that can be used to train algorithms to respond to your organization’s needs.
  • When it comes to testing and training algorithms, your organization may need additional computer power for short periods of time. This can be rented through cloud providers, which can be rented for very short periods of time.

Create hybrid teams
For each project on your list, create small teams composed of subject matter experts, data scientists and IT solutions. Subject matter experts are important team members because they have the experience to:

  • describe the key problems to solve
  • identify trends and outliers in data based on previous experience
  • determine whether algorithms are generating the right insights
  • point to the right data sources to use and identify weaknesses in datasets

Anecdotal feedback is that data scientists estimate they currently spend about 80 per cent of their time searching, accessing, organizing and labelling data, rather than on the high value work they are trained to do: developing and training algorithms. A strong hybrid team is the most efficient way to complete small projects in rapid succession.

Think about a flexible approach for data storage and access
Most organizations manage data in silos. In the past, data silos were often purposefully designed in order to manage data access and to control information flows. In an organization that seeks to digitize its operations, subject matter experts and data scientists need a plethora of trustworthy data in order to choose, adapt, train and deploy the right algorithm to meet business needs. A fundamental change in the organization’s approach to data management is often required.

Most organizations also have legacy systems containing mission-critical databases and datasets. It generally does not make sense to upend legacy systems to implement your digitization strategy. Instead consider this approach:

  • Investigate existing solutions to access datasets in legacy systems through data dashboards.
  • For new data sources, such as streaming data from video feeds, web clicks or internet of things (IoT) devices, consider using cloud services as opposed to making investments in new servers. Hybrid cloud solutions make sense for most organizations.
  • When addressing data access and storage, define and action appropriate access rights for each data source in order to avoid leaks or unauthorized use.

When the number of data sources exceeds your capacity to manage manually, consider purchasing a commercial data governance software. It can help you manage data access, tag data according to data policies, and provide auditability and lineage to underlying data sources. It can also inform you about the quality of data in real time through dashboards.

Be ready to explain outputs
In the world of AI, explainability is key. No compliance or risk team will accept the outputs of an algorithm if the rationale for how results are generated cannot be explained or replicated by human analysis using the same datasets. Large organizations have experimented with self-learning algorithms that generated seemingly brilliant insights that could not be explained nor replicated by humans. Ultimately, these could not be used in operational settings.

The explainability hurdle is real and needs to be addressed. It is even a requirement under the EU’s General Data Protection Regulation (GDPR). CEOs and boards will require evidence that AI solutions are accurate, fair, and unbiased. This requires close collaboration between data scientists, subject matter experts and compliance/risk teams.

Get ready to catch the next wave
Successful businesses need to find ways to transition from a state where most processes are managed by humans and supported by technology to a new state where most processes will be managed by digital technologies and AI, and overseen by humans.

A digitization strategy will facilitate the transition from the existing state to the future state and allow organizations to optimally utilize current data and get ready for the next wave of data sources coming our way:

  • Whole body communication: Increasingly, communication is morphing from textual to visual, for example from emails to visual expression platforms (that seamlessly combine words, images, video, sounds and graphics into one) and from mouse scroll and clicks to touchpads swipes and gestures. Soon, our primary interaction with technology won’t be through a keyboard but through voice commands and queries. In the factory, virtual reality equipment will be used to connect engineers with digital twins (digital replica of a living or non-living physical entity). Digitized body language movement, voice and eye movements will generate vast quantities of data.
  • Blockchain: This distributed ledger technology used for cryptocurrencies is also supporting sectors such as income share agreements, food supply chains, remittance payments, the settling of financial transactions, document exchange, and portable medical records.
  • Chatbots and AI assistants will be used by an ever-larger proportion of business-to-consumer interactions for marketing to consumer engagement. By some estimates, AI-powered bots will activate 95 per cent of all consumer service interactions by 2025, generating large amounts of data for businesses.
    • Digital assistants such as HomePod and Alexa already use AI to perform basic consumer functions.
    • Socially assistive robots, which can perform household chores, accomplish health care tasks, and potentially offer emotional support, are being deployed.
    • Robots are also being introduced in a wide variety of sectors including education, health care, and business, as well as policing, military and disaster relief operations.
  • Wearables: Increasingly associated with 5G and the upcoming 6G, personalized data generated from wearables, handheld and specialized sensors will provide a new level of granularity to human body functions and diseases. Again, this data will be collected through apps and services managed by organizations in a wide variety of sectors from health and disease prevention to well-being technologies, apparel, and sporting goods. 

CPAs as change agents

CPAs are well positioned to launch a much-needed dialogue about the need for a digitization strategy in their organizations or at their clients’ organizations. They have a detailed understanding of the business, the ability to develop credible business cases and the know-how to manage projects that will deliver a return on investment. They are trusted, well versed in processes, and understand the flow of information across an organization. Some CPAs are also IT specialists with hands-on experience with modern technologies. Overall, given their leadership roles within SMEs and other organizations, no other profession is better positioned to help Canada’s business community make this transition.  


More information is forthcoming on the critical role CPAs can play in mastering data, including the following issues: