"How should we use AI as an SME, and are off-the-shelf tools enough?" This question has come up especially often over the past two years, as easily accessible versions of generative AI tools have reached every employee's desk. There is a recent source to help answer it: on April 13, 2026, the OECD published the 2026 survey results of its Digital for SMEs (D4SME) initiative. According to the OECD's summary, adoption of off-the-shelf AI applications is continuing, but the strategic, targeted and secure integration of these tools into business processes remains uneven.
We covered the digitalization report card of SMEs in Türkiye, along with AI usage rates specific to Türkiye, in our article on the OECD SME Policy Index. This article focuses on a different question: how does an SME that already uses AI move beyond off-the-shelf tools?
Who Does the Survey Describe?
The D4SME survey was conducted between the last quarter of 2025 and the first quarter of 2026 and collected 2,018 responses from 12 OECD countries; 1,376 of the responses came from Japan. Türkiye is not in the sample. The participants are SMEs that are customers of digital platforms; the OECD states explicitly that the sample is not random and that the results do not represent the entire SME population of the countries.
For this reason, the survey's percentages should be read not as an answer to the question "What share of SMEs use AI?" but as a snapshot of how AI is used in relatively digitally advanced SMEs. This is where the report's real value lies: it provides a framework for the depth of use.
Widespread but Shallow: The Four-Tier Picture
The OECD reads the results through a classification it developed for the G7. The classification divides AI-using SMEs into four groups according to the complexity of the tool used and the scope of use:
- AI Novices: those who use off-the-shelf tools for individual tasks. 76% of AI users are in this group.
- AI Optimisers: those who use standard tools across multiple functions; 15.3%.
- AI Explorers: those who experiment on a limited scale with organization-tailored or advanced AI; 5%.
- AI Champions: those who have spread tailored or agentic AI across most of the business; 3.6%.
The same picture appears in the scope of use: 56.6% of participants use AI only for individual tasks, while the share using it across multiple functions or organization-wide is 19%.
Areas of use also set the two groups apart. Of those using off-the-shelf tools, 70% use them to create marketing content and 56% to draft documents and reports. In the small group using solutions tailored to the organization, the leading areas are customer demand forecasting (39%) and process automation (34%). Agentic AI, in turn, is used in multi-step workflows such as booking, invoice or campaign management. Off-the-shelf tools target individual productivity, while tailored solutions target business processes.
Returns Grow with the Breadth of Use
Of the participants, 54% say they get at least a moderate level of value from AI; however, the share reporting significant or transformative impact stays at 21%, and 44% say the impact is very small or nonexistent.
Perhaps the report's most instructive finding comes from the Japan sample. Only 2% of SMEs that use AI for individual tasks report transformative impact; the share is 4% among those using it in a single function, 11% among those using it across multiple functions and 23% among those using it organization-wide. The OECD stresses that these results rest on self-reporting rather than measured performance and do not demonstrate causality. Even so, the direction is clear: value comes less from the tool itself than from how widely it spreads across business processes.
The size gap is widening, too. According to the OECD's official statistics, AI use among small businesses rose from 7.1% in 2023 to 17.4% in 2025; however, the gap with large businesses over the same period widened from 23.4 to 34.6 percentage points.
Barriers: Time, Maintenance and Skills
When using digital tools, participants most often cite maintenance costs (39%), lack of time to devote to training (38%) and hardware costs (37%) as barriers. Skills needs, meanwhile, are mostly met through informal channels: internet search at 32%, external consultants at 29%, friends and family at 24%, and generative AI at 21%. In the Japan analysis, structured training and internal capability building turned out to be associated with higher digital maturity, whereas no such relationship is observed for informal channels.
Security is also lagging behind. Of the participants, 22% report having experienced a cyber breach, and 46% report taking no security measures at all or only minimal ones. The share that regularly reviews staff members' system access is 13%, the share that provides formal security training is 12%, and the share that carries out a security assessment with an external expert at least once a year is 7%.
From Off-the-Shelf Tool to Strategic Use: A Practical Path
The steps below are a general framework derived from the report's findings; each company needs to adapt it to its own size, sector and existing systems.
- Make current use visible. List which off-the-shelf tools employees use for which tasks. Add the embedded features in your existing software to this list as well; for details, see our article on embedded AI.
- Identify your own tier. The OECD's four-tier classification can serve as a simple self-assessment tool: are the tools used for individual tasks, across multiple functions, or inside processes?
- Move from individual productivity to a process. Choose a single pilot that connects a tool still used for individual tasks to a measurable business process (such as quote preparation, demand forecasting or document processing). The report shows that value grows as use spreads.
- Choose your acquisition route deliberately. An off-the-shelf tool, development by an external provider and an in-house team carry different cost and dependency profiles; for a comparison, see our article on acquisition routes.
- Budget for maintenance and time from the start. The most frequent barriers are maintenance cost and training time. Write the resources set aside for updates, maintenance and employee training into the pilot plan from day one.
- Keep the skills in the company. An external consultant or an internet search provides a temporary fix; structured internal training and knowledge sharing make the capability last. Add a knowledge-transfer clause to every project.
- Grow security together with usage. Review access rights regularly, tell employees what data may be entered into AI tools, and have an outside party carry out an assessment at least once a year.
- Measure the impact. Define a success criterion up front for each area of use; instead of a "that was useful" claim, track concrete indicators such as time, error rate or output.
Conclusion
The OECD's 2026 D4SME survey shows that the large majority of AI-using SMEs use off-the-shelf tools for individual tasks, and that those reporting significant returns remain a minority. Off-the-shelf tools are a good start but are not enough on their own: value grows as use spreads across processes and multiple functions. For SMEs, the way forward lies less in increasing the number of new tools than in making current use visible, choosing a process and measuring it, and growing skills and security at the same pace.
Why logistics experience matters here. The report's clearest lesson, that value comes from how widely AI spreads across processes rather than from the tool itself, is one logistics learned long ago: speeding up a single step rarely shortens end-to-end lead time, and the gain shows up only when the whole flow, from order to delivery, is redesigned and measured. Two of the pilot examples in step 3, demand forecasting and document processing, are everyday work in warehousing and distribution, and fixing a success measure before the pilot starts (step 8) is standard practice in KPI-driven operations. One of Orient Global's consultants brings more than 30 years in operations, logistics and supply chain: at a technology distribution company of roughly one billion dollars in scale, he ran logistics, warehousing, technical service and administrative operations end to end, building and leading teams of 100+ people and delivering lasting gains in operational performance through KPI-driven programs. He has also designed and directed large-scale logistics and distribution networks across leading retail, logistics and distribution organizations, managing nationwide warehouse and transport operations and network and capacity planning. An industrial engineer, he now brings that supply-chain discipline to SMEs through AI and automation. Meet our consultants.
#interimmanagement #forwarddeployedengineering