The 4 pillars of successful digital transformation (2026 update)

Updated on October 19, 2026: Adtek rewrote the entire article with the latest data, platform features and regulations from 2025 to 2026.

A digital transformation succeeds when a digital project meets or beats the business target set when its budget was approved: more revenue, lower costs, customers served faster. Buying new software is only the part where you spend the money. In a Gartner survey of 3,186 CIOs and technology executives published in October 2024, only 48% of digital initiatives met or exceeded their business outcome targets.

This article uses the four pillars of digital transformation framework that Harvard Business Review published in January 2022, updates it with AI data from 2025 to 2026, and then applies the framework to a marketing department. I do not compare specific ERP or CRM software here.

What is the success rate of digital transformation today?

The 48% figure from Gartner (October 2024) is not an outlier. Back in 2018, a McKinsey survey found that only 16% of companies had both improved performance and sustained those changes over the long term.

Five years later, the payback was still low. Writing in Harvard Business Review (July 2023), a team of McKinsey authors reported that 89% of large companies worldwide were running digital and AI transformations, yet on average they captured only 31% of the revenue lift they expected and 25% of the expected cost savings.

Run the numbers on a small example. You pitch the board a project expected to add VND 10 billion in revenue per year. At the average rates above, you would actually collect about VND 3.1 billion. The gap usually sits in work that never made it into the budget: cleaning data, training users, redrawing processes.

Generative AI is repeating the same old story

McKinsey’s The state of AI in 2026 report (August 2026), based on 1,719 respondents, found that 89% of organizations use AI in at least one business function. Only 37% see AI contributing to earnings before interest and taxes (EBIT), almost unchanged from 2025. The leaders, who attribute 5% or more of EBIT to AI, still make up only about 6%.

BCG (September 2025) surveyed 1,250 executives and reached a similar result: 5% of companies are creating value from AI at scale, while 60% have yet to see any meaningful value at all.

AI projects fail in exactly the places where digital projects used to fail. Nobody owns the business result, the data is dirty, and the process stays the same with a new tool bolted on top.

The 4 pillars of digital transformation

The most useful idea in the original HBR article is that “digital transformation” is a catch-all label for four different kinds of work. Lump them together and it becomes easy to hand the work to the wrong person and measure it with the wrong KPI. Split them apart and each one gets its own C-suite sponsor and its own metrics.

Companies usually tackle pillars 1 and 2 first, but that order is not mandatory.

IT uplift

This is often the first step: moving systems to the cloud, pulling data into a single repository, giving employees new workplace tools. In 2026 this pillar picks up one more job, which is preparing data and access permissions so AI can actually use them. An AI reading data with duplicate customer IDs will give wrong answers and sound confident doing it. The common mistake is treating this pillar as the whole transformation. The result is a new system and the same old revenue.

Digitizing operations

At the low end, this means replacing paperwork with software. At the high end, it means redesigning the process itself. The original HBR article gives an example: when PayPal moved to instant payments, the company had to merge its payments and compliance departments into one, because the old review process no longer had time to run.

AI agents belong to this pillar. According to McKinsey (August 2026), 40% of respondents at companies with more than USD 1 billion in revenue said their organization is scaling AI agents, up from 27% the year before. Attach an agent to a process nobody has redrawn and the agent just makes mistakes faster.

Digital marketing

This pillar faces outward: finding customers, selling to them, keeping them. It needs clean customer data, measurement tools and a team that knows how to read the numbers.

According to the Gartner CMO Spend Survey 2026 (May 2026), CMOs now allocate an average of 15.3% of their marketing budgets to AI, but only 30% have AI capabilities mature enough to scale. The money went out before the foundation was finished.

New digital ventures

This is the hardest pillar: using technology to open a new line of business. The metric is no longer ROMI but unit economics, meaning how much profit each new customer brings in after subtracting what it cost to acquire them. The CEO should lead it, because this pillar needs money, the authority to decide quickly, and a team that is allowed to test things and then drop them.

What do companies that succeed at digital transformation do differently?

Research from McKinsey (2018, 2019) and Gartner (October 2024) points to five shared habits:

• Few priorities, each with a number attached. A handful of themes tied to measurable outcomes, instead of 20 initiatives running in parallel.
• Business leaders share accountability. Gartner calls this group the “Digital Vanguard”: the CIO and the heads of business units co-own delivery. In this group, 71% of initiatives hit their targets, compared with 48% overall.
• Every initiative has one owner. That owner is paired with an “integrator” who understands both the business side and the technology.
• A dedicated budget and short review cycles. Work in a test-and-learn mode and update the plan every quarter instead of every year.
• A change story employees can retell. Why things have to change, what changes, and how it will be measured. According to McKinsey (2018), companies that get this right are more than three times as likely to succeed.

McKinsey’s 2019 research on going “all in” adds a caveat: cutting costs alone rarely lifts a company into the top tier of its industry. The companies that went furthest worked on both pillar 2 and pillar 4, improving performance while shifting resources into new businesses.

Digital transformation in Vietnam: targets versus reality

Vietnam’s National Digital Transformation Program (Prime Minister’s Decision 749/QĐ-TTg, June 2020) set a goal for the digital economy to reach 20% of GDP by 2025. Using figures from the National Statistics Office as reported by Bao Dau tu (January 2026), the digital economy’s share in 2025 is estimated at 14.02% of GDP, about USD 72.1 billion, or 1.64 times the 2020 level. That leaves the country nearly 6 percentage points short of the target, while Politburo Resolution 57-NQ/TW (December 2024) has already raised the bar to at least 30% of GDP by 2030.

Consumers are moving faster. The e-Conomy SEA 2025 report by Google, Temasek and Bain (VnEconomy, November 2025) estimates Vietnam’s digital economy at USD 39 billion in GMV for 2025, up 17%, the second-fastest growth in Southeast Asia. 81% of Vietnamese users interact with AI every day, and 96% are willing to share data access with AI agents, the highest in the region.

Customers already ask AI before they buy, while many companies are still testing AI internally. Getting your brand to show up correctly in AI answers, known as AIO, is now part of the digital marketing pillar.

Adtek’s view: applying the 4 pillars to a marketing department

If you want digital transformation to succeed at the level of a marketing department, the HBR framework works as is. Each pillar has a scaled-down version, and each version needs one person accountable for it.

Take an example: a cosmetics chain with 20 stores that also sells on its website, Shopee and TikTok Shop, with a marketing budget of VND 500 million per month. This is an illustrative scenario, not a real client.

Step 1: upgrade the marketing data infrastructure (pillar 1)

The first question: where did yesterday’s orders come from? If the team needs three days and five Excel files to answer that, hold off on buying more AI tools. The job is to pull orders from stores, marketplaces and the website into one place, keyed by customer ID, using a customer data platform (CDP) or a simple data warehouse.

KPI: the share of orders that can be matched to an identified customer. With 10,000 orders a month and 4,000 of them carrying a phone number that matches a customer profile, the rate is 40%. A first-quarter target might be 60%.

Step 2: digitize how marketing work gets done (pillar 2)

Redraw the process from brief to published content and record how long each step takes. Generative AI fits best in the drafting steps: headline variations, rough visuals, report summaries. The specifics are covered in our article on using generative AI in marketing.

KPI: the number of days from brief to launch, and the production cost per piece of content that meets the standard. Do not measure the number of pieces AI produced, because that number tells you nothing about revenue.

Step 3: digital marketing with real measurement (pillar 3)

Return on marketing investment (ROMI) = (Incremental gross profit from marketing – Marketing cost) / Marketing cost. Spend VND 500 million, generate VND 800 million in incremental gross profit, and ROMI = (800 – 500) / 500 = 60%.

The hard part is the word “incremental”. Some orders arrive even when you run no ads. The way to separate them out is incrementality testing: switch ads off in one group of regions and compare it with a group where they keep running. This is the basis of data-driven marketing: decisions based on incremental numbers, not gut feeling.

Step 4: test a new digital business model (pillar 4)

For example, a paid membership program, or a subscription bundle sold through Zalo, Vietnam’s most widely used messaging app. Each test gets a capped budget, a 90-day deadline and a stop condition, for instance CAC must not exceed one third of the gross profit a customer brings in over 12 months. If a customer brings in VND 900,000 in gross profit per year, the maximum CAC is VND 300,000.

8 questions to ask before approving a digital project

1. Which pillar does the project belong to? If the answer is “all four”, split it into several projects.
2. Who is accountable for the business result, as distinct from the person accountable for installing the software?
3. What is the KPI, where does it stand today, and what is the target after 6 months?
4. Is the input data clean yet, and who checks it?
5. Which processes will be dropped or redrawn?
6. Does the budget have a separate line for training and data work?
7. If the KPI has not moved after 90 days, do you stop or change direction?
8. Are the people who will use the tool day to day involved from the start?

What not to do

• Buy AI tools before you can answer “where did yesterday’s orders come from?”
• Let IT lead a marketing project, or leave marketing to build the data infrastructure on its own.
• Report success by the number of accounts provisioned or the number of pieces AI generated.

If you need to review your marketing department against these four pillars, from data infrastructure to how you measure ROMI, contact Adtek to talk it through.

This article was rewritten and updated by Adtek in 2026, drawing on the original Harvard Business Review article (2022) and McKinsey research (2018, 2019).