The Thailand Digital Outlook 2569 survey by the Digital Economy and Society Commission reports an average of 2.12 out of 4 and AI use at 21.18 percent. This article explains what each figure can and cannot tell you, and turns them into a five-question internal check.
An executive opens the news about Thailand's digital readiness survey and asks the team: the national average is 2.12 out of 4, so where do we stand, and if we are below average, should we buy more systems?
The question is understandable, but the figures in the news cannot answer it. This article is written from the viewpoint of a team that helps organizations decide about systems, not as an economic analysis. We describe what the survey reports as far as the news we could open states it, what each figure can and cannot tell you, and how to turn a national average into questions you can check inside your own organization.
What the survey reports
Thairath reported on 8 July 2569 (2026) that the Digital Economy and Society Commission (known in Thai as สดช.), under the Ministry of Digital Economy and Society, released the Thailand Digital Outlook 2569 survey, announced by the commission's secretary-general (see the Thairath report). The news states the following.
The sample was 834 target digital entrepreneurs.
Readiness was scored on a 0 to 4 scale. The average was 2.12, up 0.56 from the previous year, and the report describes it as a moderate level.
AI was the most-used advanced technology in the sample at 21.18 percent, followed by blockchain at 20.09 percent and big data analytics at 19.88 percent.
A caution about the source: we read one news report only. We did not open the commission's full report, so we do not know the methodology, such as what questions define each score level, when the data was collected, or who counts as a target digital entrepreneur. If you will cite these figures in an internal document, check them against the original report first.
What each figure can and cannot tell you
We suggest reading them one at a time, because each can be misused in a different way.
The average of 2.12 out of 4 says the sample as a whole sits in the middle of the scale. It does not tell you the distribution. An average of 2.12 can come from every respondent scoring close to 2, or from half scoring 4 and half scoring 0.24. The averages are similar, but the situations are very different. An organization that compares itself with the average is comparing itself with a number that does not describe anyone in particular.
The increase of 0.56 shows the direction: the sample improved compared with the previous year. Subtracting back, we calculate that last year's average was about 1.56. That figure is our own calculation, not one the news states, and it holds only if the sample and the measurement method are comparable across the two years, which we do not know. If the sample changed, a higher score may reflect a different group of respondents rather than the same organizations improving.
AI use at 21.18 percent is the share of the sample that uses the technology, not the share of all Thai businesses. The sample is target digital entrepreneurs, who probably lean toward organizations already working in digital. Businesses in general may have a different share, and we have no data on how different. The word "use" in a survey could also mean anything from trying an off-the-shelf tool to building AI into a core process, and the news does not say. One small observation, again our own calculation: 21.18 percent of 834 is not a whole number (about 176.6 respondents). That suggests the base for the percentage may not be the full 834, for example if only some respondents answered that question or if responses were weighted. It is another reason to open the original report and see what the base is.
Also note how narrow the gap is between the top three. 21.18, 20.09 and 19.88 percent differ by under two percentage points. We do not know the sampling error, so we would not read this as AI leading by a meaningful margin. The news says only that AI was highest, and that is all we can cite.
Why an average should not be your buying criterion
The executive who asks whether being below average means buying more systems is treating the average as a signal to invest. We see three reasons that path is risky.
First, the average measures the readiness of a group. It does not tell you what the problem is in your organization. One organization may score low because it has no inventory system. Another may score low because it has a system but nobody enters data consistently. These need different remedies, and neither starts with a purchase.
Second, a system bought out of fear of falling behind tends to be chosen for the features others have, not for the work the organization actually does. We wrote about the symptoms worth checking before buying a business management system in our article on ERP for business owners. The idea is to start from symptoms seen in real work, then find which part of a system answers them.
Third, a widely used technology is not necessarily right for you. AI was the most-used in the sample, but before AI touches real data in a system, the organization needs things in place: data clean enough, clear permissions, and people who check the output. We wrote about this in our article on the gate before an AI agent gets system access.
Turning the average into internal questions
Even though we do not know the criteria the commission used for each score level, an organization can build its own check from things the people who run its systems already see. We suggest five questions, answered as facts rather than impressions.
Where does the organization's core data live, and in how many places? Are customers, products and sales in one system or spread across many spreadsheets? If spread, which file is the version you trust when the numbers disagree?
Who owns each system? Name a person, and say who takes over if that person leaves tomorrow.
Do the systems talk to each other, or does someone copy data between them? If someone copies, how many hours per week, and how often does it go wrong? You can count this by observing the work for one week.
When a system goes down, how long does work stop, and who finds out first? This connects to recovery targets, which we covered in our article on RPO and RTO.
Of the digital tools already bought, do you use all the functions you pay for? Many organizations find that the problem they want a new system for already has a function in the existing one that nobody turned on.
The five answers do not add up to a score, and we chose that deliberately. A total score would become a number people compare against 2.12 even though the two measure different things. What you get is a list of identifiable problems, each with an owner and a clear next piece of work.
An example of reading the answers
Suppose a company answers question three like this: sales issues quotations in a spreadsheet, accounting re-keys them into the accounting system, and during the week of observation three quotations had mismatched figures. This does not say the company must buy an ERP. It says there is copying work that can be fixed in several ways, from adjusting the sales form so it can be imported, to connecting the two systems. Once the cost of the errors is known, you pick the route that fits the size of the problem.
This example is hypothetical. We are not claiming it is a real client case, and the figure of three quotations is only to illustrate.
If you must report to the board
Executives who bring figures from this survey to the board are often asked where the organization stands against the market. We think the more honest answer has two parts. The first is context: the survey by the commission covered a sample of 834, with an average of 2.12 out of 4, and AI was the most-used technology in that group, with the source and date given. The second is the organization's own status, measured internally, such as the number of systems whose data does not match, hours of copying per week, and the number of times a system stopped last quarter. These two parts should not be merged into one sentence saying we are below or above average, because that sentence suggests a comparison that did not actually take place.
This also lets the board ask better questions. Instead of asking why you are below average, directors can ask how many copying hours per week would fall if you fixed this point. The second question is answered with identifiable work, and if the answer is that you do not know yet, that is also something the board should hear.
Limits we should state plainly
If your organization has a few dozen people, uses a handful of off-the-shelf tools, and everyone knows where data lives, the five-question check may be more than you need. You can answer it in a ten-minute meeting without writing a document. Its value grows when the organization has several departments, several systems, and fewer people who see the whole picture than you would think.
Also, this check is our opinion, not the criteria the commission uses, and it cannot tell you where you stand relative to the sample. If you want a real comparison, you need the scoring criteria from the original report, which we have not read.
Finally, a survey is a snapshot of the group surveyed at one time. The report we read is dated 8 July 2569, the numbers may change in the next round, and we do not know whether the surveyors will use the same method.
Objections we often hear
A common objection is that competitors already use AI, and if we wait we will fall behind. There is something to this, and we do not suggest anyone stop watching new technology. The survey result that 21.18 percent of the sample use AI does say that people are doing it.
But the same figure also says that most of the sample did not report using AI, assuming the 21.18 percent is a share of the whole sample, which we cannot confirm from the news. Our point is that a technology most people have not adopted is not a reason for one organization to adopt it before it knows what problem it solves. Waiting one quarter to complete the five-question check costs less than starting a project at the wrong time.
Another objection is that a figure from the government is trustworthy and can be used to decide. A figure from a government agency is a source you can open and verify, which is better than a floating number. But trustworthy does not mean it matches your question. The survey answers how the sample is doing. You are asking what your organization should do. Those two questions need different data.
Where to start
Find the original Thailand Digital Outlook 2569 report from the commission and read the methodology, the sample and the scoring criteria before citing the figures further.
Invite one system administrator from each department to answer the five questions. It takes half a day and needs no special tools.
Observe cross-system copying for one week. Count the occurrences and the errors, and record them as figures your organization measured itself.
Rank the identifiable problems by their effect on customers and the cost of the errors, not by how exciting the technology is.
Once you have chosen one problem, work out whether the remedy is a configuration of the existing system, an integration, or a new system, and decide how you will measure the result when it is done.
Summary
The average of 2.12 out of 4 and the 21.18 percent AI use describe a sample of 834 respondents. They are not a criterion for what your organization should buy. If you take one thing from this article, answer the five questions with facts about your own systems, and choose an identifiable problem before you talk about purchasing.
If you would like to talk through the details of a project, you can reach us at 088-983-9386 or [email protected]. Our office is in Bang Kapi, Bangkok. For organizations still drafting a scope of work, we are glad to arrange a conversation to understand the problem first, without needing complete documents. And if we find that the problem is not yet at the point of needing a new system, we will say so plainly, because starting a project at the wrong time costs more than waiting another quarter.
FAQ: Frequently Asked Questions about This Article
A collection of questions and answers to help you better understand the content of this article.
Tags: