PISA 2025: A Sagecool Response

Executive Summary

1. Digital learning is mainstream, but more is not always better. Moderate educational use of digital devices is associated with higher science performance than no use or very high use, while leisure use and distraction show a much less favourable pattern. These are associations, not proof of causation. [1, 2, 6, 31, 39, 40, 43]
2. Effective digital learning depends on teachers and institutions, not just hardware. PISA measures teacher skills, preparation time, professional resources and technical support as part of digital preparedness. Better prepared systems tend to perform better, yet OECD-average preparedness fell between 2022 and 2025. [2, 25, 36, 37]
3. Live online group teaching is already significant. Seventeen per cent of OECD students report real-time online group lessons as additional science instruction, rising to 54% in Viet Nam and Indonesia. That demonstrates adoption, not that the mode causes better results. [3, 4]
4. Subject teaching can cross school boundaries. Public systems in Scotland, Newfoundland and Labrador, Texas, New South Wales, Victoria and New Zealand already use online or cross-school provision to widen subject access, form viable cohorts and solve geographic or timetable constraints. [14-20]
5. Teacher shortages are both quantitative and specialised. Around four in ten OECD students are in schools where lack of teaching staff hinders instruction, and around one quarter are in schools reporting inadequate or poorly qualified staff. The missing resource may be a particular accredited specialist, not simply another teacher. [11-13]
6. The teacher remains central. Teacher support and cognitively demanding teaching are associated with stronger learning behaviours and, within many systems, stronger performance. Research on synchronous K-12 teaching likewise emphasises pedagogy, relationships, interaction and teacher presence. [5, 21, 22, 28, 38]
7. The opportunity is better allocation, not simply more technology. Above about US$85,000 in cumulative spending per student, the relationship between spending and science performance becomes much less clear. Sagecool’s proposition is to make specialist teaching capacity more portable across school boundaries and judge the model by access, teaching quality and student outcomes. [8, 9, 30, 42]

1. PISA 2025 Is A Global Education Report

PISA matters to Sagecool because its unit of comparison is the education system, not one country. The 2025 assessment covers 91 countries and economies and combines academic results with information about schools, teachers, families, student engagement and learning environments. That scale makes it possible to distinguish international structural problems from issues created by one national policy setting. [35]

The global perspective is especially relevant because senior-secondary subject access is not neatly distributed within school boundaries. Curricula, teacher accreditation and child-safety rules remain jurisdiction-specific, but the structural problem recurs: specialist teaching capacity and individual student demand do not necessarily appear in the same school, at the same time, in a viable cohort. That is a resource-allocation problem before it is a technology problem. [14, 16, 17, 18, 19, 20]

PISA 2025 also arrives in a more difficult academic environment. Science performance was broadly stable on the OECD average compared with 2022, while reading and mathematics declined, and the performance distribution widened. The implication for any new delivery model is demanding: broader access is valuable only if it translates into learning, engagement and support. [10, 35]

2. What PISA 2025 Actually Says About Digital Learning

The OECD framework starts from an affirmative premise: digital learning technologies are fundamentally changing how people learn, enabling new forms of autonomous learning and providing opportunities for more active and authentic learning experiences. At the same time, OECD’s 2025 results reject the idea that devices themselves are educational outcomes. Access is widespread, but educational value depends on how technology is integrated into teaching and learning. [3, 7]

The distinction between learning and leisure is central. PISA reports that limited or moderate learning-related use — up to five hours per school day in the analysis — is associated with higher science performance than no use or high use. For leisure use, science performance is highest around very limited use and falls as reported time rises beyond roughly one hour per school day. These patterns remain observational. They may reflect differences in schools, students or access as well as the effects of device use itself. [2, 39, 40]

PISA also measures school preparedness for digital learning. The construct includes infrastructure and connectivity, but also whether teachers have the necessary technical and pedagogical skills, sufficient preparation time, professional resources, incentives to integrate digital tools, curriculum alignment and technical support. This is important for interpretation: “digital readiness” is partly a measure of the human and organisational system around technology. [25]

At the system level, greater preparedness is associated with higher average science, mathematics and reading scores; after GDP per capita is taken into account, cross-country differences in preparedness account for about 14% of variation in mean science performance across all participating systems. Within countries, however, the performance relationship is generally no longer observed after socio-economic profiles are considered. Principals also reported lower preparedness in 2025 than 2022 on the OECD average and in about half of systems with comparable data. These qualifications are not side notes. They are part of the OECD finding. [2, 36, 37]

3. Digital Learning, Online Learning And Synchronous Teaching Are Different Questions

PISA’s Learning in the Digital World assessment should not be used as direct evidence for synchronous online classes. OECD defines the domain around students engaging in iterative, self-regulated knowledge building and problem-solving using computational tools and practices. The assessment uses models, simulations, programming and feedback. A live senior-secondary class taught by a teacher elsewhere is a different educational mechanism. [3]

Three distinct bodies of evidence are relevant. PISA reports on digital learning generally: device use, preparedness, distraction, AI and self-regulation. It also reports a specific survey measure of real-time online group additional science instruction. Public education systems provide separate evidence that cross-school curriculum provision is operationally feasible. The three streams are complementary, but none proves the others. [2, 3, 4, 14, 16, 17, 18, 19, 20]

4. PISA Measures Real Time Online Group Teaching Directly

The most directly relevant PISA statistic for Sagecool comes from additional science instruction. Around 60% of OECD students report at least one form of additional instruction. The categories include recorded lessons or online platforms, individual tutoring on site, group tutoring on site and real-time group lessons online. Seventeen per cent report the real-time online group category; the share is about 5% in Korea and Japan and 54% in Viet Nam and Indonesia. [4]

The significance of the 17% figure is scale and adoption, not proof of educational superiority. It concerns additional science instruction, not the share of ordinary school timetable classes delivered online. Students may participate in more than one form of additional instruction, so the categories cannot be added together. PISA does not randomly assign students to the different forms, and lower-performing students may be more likely to seek additional support. [4, 41]

A material share of 15 year olds across OECD countries already experience real-time online group instruction as an additional form of science learning. PISA 2025 provides direct international evidence of that participation, but it does not establish that the online mode causes higher or lower achievement. [4]

5. Cross School Curriculum Provision Is Already Operating In Public Systems

Public systems provide a different kind of evidence from PISA. They show how online and distance provision is used in practice to solve subject-access, cohort-size, timetable and staffing problems while students remain connected to a home school or public system. [14, 16, 17, 18, 19, 20]

The examples operate at different jurisdictional levels: Scotland, Newfoundland and Labrador, Texas, New South Wales, Victoria and New Zealand. Their designs also differ. e-Sgoil is strongly synchronous; Texas and Virtual School Victoria encompass broader online delivery. The shared structural point is that a school need not physically contain every teacher contributing to a student’s curriculum. [14, 15, 16, 17, 18, 19, 20]

6. From The School As The Unit To The Student As The Unit Of Need

Educational need appears at the level of the individual student even when accountability, pastoral care and much of the curriculum remain organised through a school. Two students in the same senior cohort may need different languages, different mathematics pathways, different extension subjects or different kinds of support. A school can be adequately staffed in aggregate while still being unable to offer one student a particular subject. [35]

Cross-school provision converts part of that problem into a matching exercise. If a small number of students at several schools want the same specialist subject, each local cohort may be too small to justify a separate class. Across schools, the cohort may be viable. The resource is not created by the internet; the boundary around the market for that resource changes. [14, 15, 16, 17, 18, 19, 20]

In Sagecool’s operating model, the teacher chooses the combination and structure of the class. A class may use live teaching, student-led online sessions, preset videos or other independent work in any combination, including one component alone. Sagecool does not prescribe a repeating cycle or set proportions. Before the class starts, timetable matching applies only to teacher-scheduled live teaching sessions; student-led sessions and independent work do not require pre-start timetable coordination by Sagecool. [33]

That makes the proposition broader than a “pure synchronous platform” but still teacher-centred. The platform is infrastructure for matching and communication; educational design remains with the teacher and the relevant school or provider. [33, 34]

7. Equity And Social Diversity: What PISA Does And Does Not Establish

PISA 2025 documents large inequalities. Advantaged students outperform disadvantaged students by about 85 science points on the OECD average, and high-achieving disadvantaged students are substantially less likely than similarly high-achieving advantaged students to expect to enter higher education. Disadvantaged students also report weaker belonging in many systems. These findings are about unequal opportunity and outcomes, not a simple causal effect of school composition. [10]

School composition nevertheless matters as context. PISA reports large cross-system differences in how much performance varies between schools, and Schleicher warns that where academic segmentation compounds with social background it can amplify inequalities. The PISA questionnaire framework also draws on prior research on school diversity and includes diversity constructs. These are research foundations and measurement choices rather than new 2025 causal evidence that social diversity itself improves outcomes. [10, 23]

Additional instruction is also unequally distributed: advantaged students are more likely to receive it. Online access can therefore operate in either direction. If it is primarily private enrichment, it can reproduce advantage. If schools or systems use it to extend subjects and specialist teachers to students whose local school cannot provide them, it can broaden opportunity. PISA does not establish that merely mixing students online reproduces the benefits associated with more diverse physical schools. [24]

8. Teacher Shortages Are Global, And The Scarcity Is Often Specialised

PISA 2025 confirms that teacher supply and teacher quality remain constraints across OECD systems. Around four in ten students attend schools whose principals report that lack of teaching staff hinders instruction. Around one quarter are in schools where inadequate or poorly qualified staff are reported to hinder instruction. The measures are principal reports of constraints; they are not literal vacancy counts. [11]

UNESCO provides separate global evidence that broadens the geographic picture beyond the OECD systems covered by PISA. Its report also stresses that teacher shortages affect high-income regions through recruitment and retention, not only countries seeking to expand enrolment. [12, 13]

For Sagecool, the economically relevant scarcity can be narrower than total teacher numbers. A school may have a full staffing establishment and still lack the teacher required for Physics, an advanced language, Specialist Mathematics, an extension subject or a student with particular needs. Online delivery does not create those specialists, but it can expand the radius within which a school can search for appropriately qualified capacity. [33, 34]

9. The Evidence On Synchronous Online Teaching Is More Limited Than The Volume Of Writing Suggests

There is a great deal written about K-12 online education, but comparatively little high-quality causal evidence on teacher-led synchronous senior-secondary courses where students remain enrolled in physical schools and take only one or two subjects online. The literature combines full-time virtual schools, supplemental courses, emergency remote teaching, blended programmes, asynchronous platforms and synchronous classes. That heterogeneity makes broad claims that “online works” or “online does not work” unreliable. [21, 22]

A 2026 study by Prestridge and colleagues is unusually relevant because it examines 22 experienced Australian distance-education teachers across five sites, teaching fully online in a mainly synchronous mode. The teachers described deliberate choices around content selection and presentation, social and emotional support, relationships and online interaction. The study argues that online teaching is an intentional pedagogy rather than the simple transfer of classroom material to a screen. [21]

A 2023 systematic review of K-12 online teaching in the United States reached a compatible conclusion about conditions: prepared educators, technology access, student developmental needs and self-regulation matter, alongside connected learners, supportive environments, individualisation, active learning and timely assessment. The review also warns against importing higher-education evidence directly into K-12 settings. [22]

Neither source proves that a Sagecool-style class is equal or superior to a physical class. The appropriate conclusion is narrower: if synchronous teaching is used, quality depends on pedagogy, teacher presence, relationships, interaction, course design and student support. That is a useful proposition, but it remains different from a causal efficacy claim. [21, 22]

10. Teacher Support And Cognitive Activation Still Matter — With An Important Qualification

PISA defines teacher support through concrete behaviours: showing interest in every student’s learning, giving extra help when needed, helping students learn, continuing until students understand and allowing students to express their views. Within countries, greater teacher support is positively associated with science performance in around 56 systems after accounting for student and school socio-economic profiles. It is also consistently associated with more positive attitudes towards learning and a stronger sense of belonging. [5, 28]

The system-level evidence points in the opposite direction. Countries and economies reporting higher levels of teacher support tend to have lower average science performance, even after GDP per capita is considered. OECD says this negative association needs cautious interpretation because it may signal a compensatory mechanism: teachers provide more support where students are struggling. Changes in teacher support and performance over time also show no simple system-level relationship. [5, 38]

This is not a reason to dismiss teacher support. It is a reason to avoid cherry-picking. A variable can be positively associated with outcomes within systems while having a different correlation across systems because the underlying populations and responses differ. For a matching platform, the practical implication is that the amount and kind of support a student requires is itself part of the educational problem rather than a simple quality score. [5, 28, 38]

PISA’s work on cognitive activation points in a related direction. Teaching that asks students to explain, reason, connect ideas, interrogate evidence and reflect is associated with stronger learning attitudes and with performance. Again, these are observational relationships, but they reinforce a conception of teaching as active intellectual work rather than content transmission. [29]

11. Digital Distraction Is Not The Same Thing As Online Teaching

PISA 2025 reports that 28% of OECD students say classmates are distracted by digital devices during most or every science lesson, and greater reported distraction is associated with lower science performance in many systems. The main OECD discussion is about divided attention, leisure use and school policies for devices. [6, 31, 43]

Digital distraction and online teaching are different mechanisms. A student scrolling social media during a physical lesson and a student discussing a difficult equation with a live online teacher are both using a digital device, but the educational role of the device is opposite. The relevant distinction is between the device as distraction and the device as a communications channel for teaching. [6, 31]

12. AI Changes The Competitive Question For Online Education

PISA shows that AI chatbots are already common in schoolwork. Schleicher reports that roughly 46% of students use AI chatbots weekly or more to help them learn. Students also use AI to summarise texts, conduct preliminary research and draft assignments. The relationship with science performance is complex: non-users outperform users for some specific tasks, while moderate use for general learning can look different, and PISA repeatedly warns that these patterns do not establish causality. [9, 30, 42]

Schleicher describes generative AI as an amplifier: used well, it can explain, prompt, challenge and correct; used poorly, it can become a shortcut around the cognitive work education is intended to develop. PISA also places growing emphasis on whether students can assess the quality and credibility of AI-generated information. [9, 42]

For Sagecool, the implication is straightforward. As AI makes explanations, summaries and first-draft answers easier to obtain, the distinctive value of a human teacher shifts further towards diagnosis, judgement, feedback, motivation, discussion and accountability. AI can provide information quickly; a teacher still has to work out what a student has understood, where the misunderstanding lies, when to challenge or encourage, and how to keep the student engaged in doing the thinking rather than outsourcing it. [9, 42]

13. Spending More Is Not The Same As Spending Better

PISA 2025 finds that, across education systems, cumulative expenditure per student has a positive relationship with science performance at lower spending levels. The relationship becomes much less clear above roughly US$85,000 per student over the theoretical duration of studies. The OECD’s conclusion is not that money does not matter. It is that beyond a substantial resource threshold, additional spending does not automatically translate into higher performance. [8]

Schleicher illustrates the point with Luxembourg, which spends far more per student than most systems but does not top the science table. That comparison must be treated cautiously because cost structures differ: languages, salaries, geography, demographics and school scale all affect expenditure. The robust point is narrower — expenditure alone is an incomplete explanation of performance in well-resourced systems. [8]

For cross-school online provision, the potential productivity mechanism is not necessarily lower teacher salaries or simply larger conventional classes. It is aggregation and matching. Several schools with a handful of students seeking the same specialist subject may be able to form one viable class rather than each school attempting to create its own local class. e-Sgoil’s explicit formation of viable cohorts is a direct operating example of this mechanism. [14, 15]

14. Measurement Matters Because Teachers And Students Are Not Interchangeable

PISA 2025 does not reduce educational performance to a single test score. It measures curiosity, perseverance, adaptability, help seeking, goal setting, growth mindset, belonging and the value students place on school, as well as teacher support, digital preparedness, family support and disciplinary climate. The OECD’s measurement architecture therefore treats learning as multidimensional. [29, 35]

That has a direct but limited relevance to matching. Two teachers with the same accreditation and subject code need not be equally suitable for the same student. One may be particularly effective with extension work, another with rebuilding foundations, another with examination preparation, and another with students who need more structured support. Students differ in prior knowledge, independence, confidence, pace and the level of teacher interaction they need. [29, 33, 34]

PISA does not validate SEPI, SOMS or any Sagecool matching algorithm. It does support the narrower premise that variables beyond formal subject qualification matter to learning. The test for Sagecool is empirical: whether the platform can measure those differences reliably enough to improve access, fit and outcomes in practice. [29, 33, 34]

15. What PISA 2025 Means For Sagecool Globally

PISA documents the environment in which Sagecool operates: digital learning is embedded in modern education; moderate educational use is associated with stronger performance; preparedness matters at system level but is difficult to sustain; live online additional instruction already occurs at material scale; teacher access and staff quality remain constraints; inequalities are large; and spending alone does not determine outcomes. [1, 2, 4, 8, 10, 11]

Existing public systems demonstrate operational feasibility rather than superiority. In Scotland, Newfoundland and Labrador, Texas, New South Wales, Victoria and New Zealand, some subject provision already crosses school boundaries to deal with viable cohorts, geography, unavailable subjects or timetable constraints. The models differ, but the home school need not contain every teacher who contributes to the student’s curriculum. [14, 15, 16, 17, 18, 19, 20]

External research suggests that if teaching is delivered online, it still needs deliberate pedagogy, relationships, interaction, presence and support. The evidence base for the exact Sagecool configuration remains limited, and that limitation should be treated as a reason to measure outcomes rather than as a reason to claim equivalence in advance. [21, 22]

Sagecool’s interpretation is an allocation proposition. SEPI seeks to make appropriately qualified teaching capacity more searchable and portable across schools and geography. The teacher remains responsible for the educational structure and may use live sessions, student-led sessions, preset materials or independent work in any combination. Synchronous teaching can be central without being the only permitted mode. [33, 34]

Sagecool is teacher-controlled online curriculum delivery in which synchronous live teaching can be central while the teacher remains free to combine it with other forms of learning. This keeps the live teacher central without prescribing a single class structure. [33]

Conclusion

PISA 2025 changes the education debate because digital learning is no longer peripheral. The OECD now measures learning with digital tools as an innovative domain, examines the capacity of schools to integrate those tools and reports direct international evidence on students participating in real-time online group lessons. At the same time, the results are emphatic that technology on its own is not a learning strategy. Purposeful educational use, teacher capability, attention, support and the design of the learning environment remain central. [1, 2, 3, 4, 5, 7]

For Sagecool, the relevant question is therefore not whether all schooling should move online. It is whether the physical boundary of a school should continue to determine every teacher and every subject a student can access. Public systems already show that the boundary can be crossed. PISA shows why access, teacher quality, equity, student differences and resource allocation make that increasingly relevant. [14, 15, 16, 17, 18, 19, 20]

Sagecool’s proposition is narrower than an online-schooling claim. The school or provider remains central and the teacher controls the class design. Live teaching can be central, but it can be combined with student-led sessions, preset materials or independent work. Sagecool’s role is infrastructure: making teacher capacity and student demand easier to match across geography where curriculum, accreditation and child-safety rules allow. [33, 34]

PISA 2025 does not determine whether the Sagecool model will succeed. It does make the global problem substantially clearer and provides a strong basis for judging the model by measured access, teaching quality and student outcomes rather than by assumptions about technology alone. [1, 2, 4]