From Evidence To Decisions: A Lean Framework For Quality Improvement In Higher Education

Diego Boerchi¹ , Laura Evangelista² , Concetta Fonzo²

¹Department of Psychology, Università Cattolica del Sacro Cuore, Italy, ²Istituto Nazionale per l’Analisi delle Politiche Pubbliche – INAPP, Italy.

How to cite: Boerchi, D.; Evangelista, L.; Fonzo, C. (2026). From evidence to decisions: a lean framework for quality improvement in higher education. In: 12th International Conference on Higher Education Advances (HEAd’26). Valencia, 15–18 June 2026. https://doi.org/10.4995/HEAd26.2026.21656

Abstract

Quality work in Higher Education often produces reports and indicators without leading to clear decisions or follow-up actions. This paper proposes a lean, practice-oriented framework to reduce this evidence-to-decision gap under real workload constraints. Drawing on previous research on developmental peer review and quality improvement in vocational education and training, the framework is adapted to Higher Education through six internal functions, shared competence domains and a timeboxed 4–10-week workflow. It combines three core work products: an evidence pack, a decision log and a micro-improvement plan. A realistic worked example illustrates how limited evidence can be translated into bounded and traceable actions without adding unnecessary bureaucracy. The contribution is conceptual and design-oriented rather than an empirical evaluation.

Keywords: Quality; Peer Review; Feedback; Evidence; Improvement; Governance.

1. Introduction

Quality work in Higher Education often produces reports, indicators and feedback; however, this information does not always lead to shared decisions or concrete follow‑up actions. Many institutions face an evidence‑to‑decision gap: evidence is collected, but it is not used in a timely way to support improvement. Common reasons include fragmented responsibilities, long review cycles and unclear links between data and educational choices (Harvey & Stensaker, 2008).

A practical response is to treat quality work as an organizational capability supported by simple routines. Prior studies show that capability frameworks and organizational routines can reduce coordination costs and help teams work with evidence in a consistent way (Boyatzis, 2008; Feldman & Pentland, 2003). This is especially relevant in assessment and feedback processes, where useful information often remains disconnected across modules (Carless & Boud, 2018). Building on this rationale, the paper proposes a lean framework that uses minimal evidence, short cycles and clearly defined functions to support faster and more transparent decision-making. The contribution is conceptual and practice-oriented: it specifies a feasible decision architecture for Higher Education and identifies the organizational conditions under which it can be implemented and later evaluated.

2. Context and conceptual background

2.1. The evidence-to-decision gap in Higher Education

In Higher Education, the central problem is not the lack of evaluative information, but the difficulty of converting that information into timely and actionable decisions. Institutions routinely generate evidence through indicators, student feedback, review procedures and performance reports, yet these inputs often remain weakly connected to concrete changes in teaching, assessment or curriculum design. This disconnection, which we define as the evidence-to-decision gap, becomes especially visible when evidence circulates across multiple organizational levels without a clear mechanism for prioritization, interpretation and follow-up. As a result, quality processes may remain descriptive rather than decision-oriented, producing knowledge about problems without creating the conditions for bounded and coordinated action.

Three factors contribute to this gap. First, fragmentation: information is produced at different levels (course, programme, institution), but it is not integrated into a shared process. This makes it difficult for teams to agree on priorities or to interpret data in a consistent way. Second, unclear responsibilities: many actors are involved in quality assurance, but it is not always clear who is responsible for making decisions or how evidence should move through the organization. Third, long timelines: traditional quality procedures often follow annual or multi‑year cycles, which slow down action and reduce motivation to engage in improvement. Research shows that the effective use of evidence for educational decision-making requires clear routines, shared interpretation and specific skills such as data literacy (Mandinach & Gummer, 2016). When these conditions are missing, evidence remains descriptive and does not influence teaching, assessment or curriculum development. Similar issues appear in the field of student feedback, where information is often rich but not translated into actionable change (Carless & Boud, 2018). Taken together, these challenges indicate that institutions need lighter and more operational approaches to connect evidence with decisions (Harvey & Stensaker, 2008).

2.2. Transfer of a practical model

To address the mentioned challenges, our work draws on previous research (Evangelista & Fonzo, 2023) conducted in the field of Vocational Education and Training (VET). In that context, short improvement cycles, simple work products and clearly defined roles have helped teams use evidence in a practical and consistent way. We adapt these elements to Higher Education (Fonzo, 2022), where governance is more collegial and teaching cultures differ across disciplines.

The aim is not to replicate existing models, but to translate useful design principles, such as short cycles, minimal evidence and transparent decision‑making, into a framework that fits academic settings. This results in a practical and feasible approach that supports faster decision-making, reduced workload and more consistent follow‑up across programmes.

3. Approach and design principles

Rather than transferring the VET model directly, we adapt its most useful elements (clear roles, minimal evidence and short timeframes) to fit the collegial governance and teaching culture of Higher Education. The framework should therefore be read as a design translation rather than as a direct transfer or a validated implementation model. Its architecture was defined using four practical criteria: parsimony, traceability, feasibility under real workload constraints and role compression, so that the cycle can remain operational even when the same person covers more than one function. The design of the framework follows five core principles: 1) parsimony: only essential roles, competence areas and outputs are included to keep the process light and workable; 2) traceability: each step in the cycle creates a clear and visible link between evidence, interpretation, decision-making and follow-up; 3) timeboxing: the cycle is limited to 4–10weeks to support timely action and avoid long discussions; 4) separation of evidence and interpretation: the evidence pack contains only descriptive information, while the decision log records the team’s judgments and choices; 5) burden minimization: the model encourages teams to reuse existing information, limit meetings and request only the evidence needed for the next decision.

Taken together, these principles create a framework that is easy to use in practice and sufficiently structured to support transparent and reliable educational decisions.

4. The proposed model

4.1. Internal roles and key functions

Internal roles are a key element of the framework, as they make the use of evidence and peer feedback more consistent across modules and programmes. Prior work suggests that clear functions and simple transitions help teams convert feedback into decisions, even under real workload constraints (Boerchi et al., 2026; Campion et al., 2011). For this reason, the framework defines six essential functions, each supporting the short-cycle workflow in different ways:

• Academic Sponsor – provides legitimacy to the cycle, sets priorities and supports follow‑up at programme or school level;

• Quality Coordinator - organizes the 4–10-week cycle, keeps the process lean and maintains core outputs such as the decision log;

• Evidence Lead - prepares the evidence pack, checks data quality and supports teams in interpreting evidence;

• Peer Review Facilitator - guides the review discussion and ensures that feedback is specific, fair and aligned with the focus question;

• Peers (faculty, students and external societal actors) - contribute independent perspectives, help validate interpretations of evidence and support teams in assessing the relevance and feasibility of proposed improvements;

• Improvement Designer - translates decisions into feasible actions and helps align changes with curriculum and assessment requirements.

These six functions should not be interpreted as six mandatory appointments. In small programmes, two to four people may cover multiple functions, provided that evidence preparation, facilitation and decision ownership remain distinguishable. For instance, the Quality Coordinator may also serve as Academic Sponsor, while the Evidence Lead may additionally act as Peer Review Facilitator. This functional interpretation is intended to reduce administrative burden and to preserve the lean character of the cycle.

4.2. A 3×3 matrix for quality decisions: levels, objects and outputs

The 3×3 matrix makes decision‑making operational by indicating, for each organizational level (macro, meso and micro) and quality area, where a decision should be made, the minimum evidence required and the expected output. The three quality areas are “Governance” (roles, responsibilities and coordination rules that support decision‑making), “Curriculum” (the structure of learning activities and their alignment with learning outcomes), and “Assessment & Feedback” (grading practices, feedback routines and the use of evidence on student work). The purpose of the matrix is to provide teams with a simple structure for linking evidence to concrete actions within a 4–10-week cycle.

Table 1. 3×3 Matrix: typical decisions, minimum evidence and outputs

Level \ Object	Governance	Curriculum	Assessment & Feedback
Macro (institution/school)	Decision: priorities & review focus Minimum evidence: KPI trends, external review points Output: annual priority note	Decision: programme portfolio rules Minimum evidence: demand/enrolment, completion/attrition, coherence risks Output: portfolio decision log	Decision: assessment policy guardrails Minimum evidence: grade patterns, complaints/appeals signals Output: policy update + risk note
Meso (programme)	Decision: roles & cycle calendar Minimum evidence: workload map, process gaps Output: cycle plan + RACI snapshot	Decision: curriculum change package Minimum evidence: outcomes-to-courses map, student feedback Output: micro-improvement plan	Decision: assessment strategy alignment Minimum evidence: assessment map, rubric/reliability signals Output: assessment action list
Micro (course/module)	Decision: local action ownership Minimum evidence: last-cycle actions, local constraints Output: task list in decision log	Decision: lesson/activity redesign Minimum evidence: learning evidence, recurrent learner difficulties Output: change note + next check point	Decision: feedback and grading adjustments Minimum evidence: samples of work, feedback uptake signs Output: rubric/feedback tweak record

The matrix helps programmes keep documentation concise and focused, avoid unnecessary data collection and connect decisions across organizational levels through shared work products such as the evidence pack, decision log and micro‑improvement plan.

4.3. Short‑cycle workflow (4‑10-weeks) and core outputs

The short-cycle workflow translates evidence and peer feedback into a small set of decisions within a 4–10-week cycle. It relies on three core outputs: the evidence pack (descriptive information), the decision log (interpretations and choices) and the micro-improvement plan (actions and checkpoints). The workflow consists of six simple steps: defining the focus question, preparing minimum evidence, conducting a short peer review, agreeing on decisions, implementing small-scale actions and checking early signals at the end of the cycle. The 4–10-weeks range should be read as a bounded operational target for small-scope decisions, especially at micro or meso level, rather than as a universal duration for all quality issues.

A worked example, presented as a hypothetical but realistic scenario, shows how the short‑cycle workflow can be applied at the micro level. The example focuses on a single psychology module that introduces a basic model of work performance, explaining how individual competences contribute to performance when moderated by motivation and role clarity. Students are also expected to interpret short cases describing situations of poor performance and to identify whether the main issue relates to competence, motivation or unclear role expectations. The evidence pack includes three small pieces of information:

1. a short activity pattern from the Learning Management System showing uneven preparation across weeks;

2. a brief three‑item student pulse survey on difficulties in case interpretation;

3. a small sample of recent case‑analysis submissions with tutor comments.

During the time‑boxed peer‑review meeting, the team separates evidence from interpretation and agrees on three bounded actions aligned with the micro‑level objects of the matrix:

• Governance: clarifying the weekly responsibility for monitoring early signals within the module team and defining when the evidence pack should be updated;

• Curriculum: redesigning a learning activity by adding an explicit worked example that shows how to link competence, motivation and role clarity in a simple case analysis;

• Assessment & Feedback: simplifying one rubric criterion (from three indicators to one) and adding a short feedback prompt to guide students in justifying their interpretation.

The micro‑improvement plan records owners, deadlines and specific early signals to be observed after four weeks, such as a reduction in “misclassification” errors in short case tasks or more consistent preparation patterns. The decision log documents the rationale behind the decisions, as well as the uncertainties and limitations of the available evidence. The scenario is not intended as empirical validation, but rather as a transparent illustration of the minimum information required to apply the model in a realistic module context.

5. Implementation guidance and boundary conditions

The model is designed for use under realistic conditions, but it still requires a small set of prerequisites. First, a minimal dataset must be available without creating additional burden: basic participation and completion patterns, student feedback (even in the form of short pulse surveys), an assessment map at programme level, and, where feasible, small samples of student work for calibration purposes. To remain lean, most of this information should come from routine institutional systems or from light-touch local collection. Where possible, data extraction and formatting should be supported by a programme administrator, quality office or institutional analytics function rather than added to faculty workload. When such support is unavailable, the cycle should be restricted to a narrowly bounded question and to evidence that can be assembled within a clearly protected time budget. Second, the cycle requires a clear mandate: the scope of decisions must be defined (e.g., what can be changed at course level, what requires programme approval and what is constrained by external regulations). Third, there must be an academic sponsor who protects time, legitimizes decisions and removes barriers; without such sponsorship, routines tend to remain optional and fragile (Harvey & Stensaker, 2008). Finally, teams need access to evidence-related expertise to ensure that data are interpreted according to basic quality standards and that their limitations are explicitly acknowledged (Mandinach & Gummer, 2016). Several risks may appear during the cycle, such as turning the process into additional bureaucracy, overloading staff, selecting only convenient evidence or relying on weak data that are not aligned with the decision at hand. These risks can be mitigated by working with only the minimum necessary evidence, using small and focused data samples, assigning responsibility for checking evidence quality and keeping all steps strictly time-boxed to maintain focus and limit unnecessary tasks.

The model is designed for developmental improvement rather than causal evaluation. It works best when the goal is to keep decisions visible, timely and easy to review, and when leadership supports iterative change rather than attempting to produce a single “perfect” redesign. For this reason, its first test in Higher Education should concern feasibility and adoption before outcome claims: for example, the time required to prepare the evidence pack, the number of functions effectively covered, the completion of the three core outputs, the perceived burden for staff and the proportion of decisions that reach implementation and review. Educational effects should be examined only in a second step and with designs aligned to the scope of the intervention and the nature of the available data.

6. Discussion

The proposed model complements “classic” quality assurance approaches in Higher Education, such as programme reviews, periodic self-evaluation reports and audit-oriented cycles. These approaches are valuable for accountability and system coherence, but they often operate on long timelines and may privilege documentation over learning in practice (Harvey & Stensaker, 2008). The model presented here is not intended as a replacement; rather, it acts as a lean layer that aims to improve how evidence and peer feedback are translated into decisions between major review events. The first distinguishing dimension is speed. Programme reviews and audits typically run on annual or multi-year cycles. While they can identify issues, response times are often slow because decisions are bundled into large packages and must pass through multiple governance stages. By contrast, the proposed micro-workflow is intentionally timeboxed (4‑10-weeks) and focuses on “small-batch” change: one focus question, a limited evidence set and a small number of actions that can be implemented and reviewed quickly. This supports iterative improvement under real workload constraints, rather than deferring action until the next formal cycle. The second dimension is traceability. Traditional quality processes can produce rich reports, but the link between evidence, discussion, decisions and follow-up is not always visible in a concise and integrated way. The model’s core work products (the evidence pack, decision log, and micro-improvement plan) make this link explicit. They enhance transparency by separating descriptive evidence from interpretation and decisions, and by documenting responsibilities, timelines and checkpoints. In this sense, traceability is achieved through concise artefacts rather than extended narrative reports. The third dimension concerns the translation of evidence into action. Reviews and audits often result in recommendations that are reasonable in principle but difficult to implement or prioritize at programme and course levels. By defining internal roles and routines, the model provides a structured pathway from peer feedback to concrete decisions, with responsibilities distributed across leadership, coordination, evidence expertise, facilitation and course owners. This role-and-routine architecture is intended to reduce the evidence-to-decision gap by lowering coordination costs and protecting time for follow-up (Feldman & Pentland, 2003; Mandinach & Gummer, 2016). Beyond these three dimensions, the model also supports a more practice-oriented culture of quality work. Many institutions report that improvement activities are often slow or disconnected from everyday teaching practices. The short cycles proposed here enable teams to work with evidence in a concrete and focused way, reducing the distance between problems, decisions and visible actions. The worked example illustrates how a small, low‑cost intervention can generate early monitoring signals within a few weeks and motivate continued engagement. This suggests that the model can function as a flexible layer that complements existing procedures without replacing them. It also offers a shared language that facilitates coordination across modules and programmes. In this sense, the approach may help academic leaders build a more stable routine for decision‑making under real workload constraints. Finally, the scope of the claims must be clearly stated. The model is not a measurement framework, and it does not estimate causal impacts on learning outcomes. Its aim is organizational, to make quality work feasible, transparent and learnable, and to support timely improvement decisions that can later be evaluated using appropriate research designs when needed.

At the same time, the framework has clear limits. It may fail when local teams do not have protected time, when evidence infrastructures are weak, or when the decision scope exceeds what can be handled meaningfully in a short cycle. It also carries the risk of privileging easily available evidence over more important but harder-to-assemble information, and of producing local optimization without sufficient alignment at programme or institutional level. These limits do not invalidate the framework, but they indicate that its value depends on careful scoping, explicit sponsorship and modest claims about what short-cycle routines can achieve.

7. Conclusion

This paper provides a lean, design-oriented framework for developmental quality work in Higher Education, including a function-based architecture, a 3×3 decision matrix and a timeboxed 4–10-week workflow with core work products. The framework is most credible when institutions operate under a clear mandate, have limited but usable evidence and can rely on leadership support and at least minimal data assistance, particularly when the goal is iterative improvement rather than compliance reporting alone. Its main contribution is to clarify how quality work can be organized so that evidence is more likely to inform bounded decisions under real workload constraints. Future research should therefore test feasibility, adoption and outcomes through pilot implementations across different programme contexts and institutional settings. Where appropriate, evaluation designs should be aligned with the scope of decisions and the nature of available data, ensuring both methodological rigor and practical relevance.

References

Boerchi, D., Evangelista, L., & Fonzo, C. (2026). Beyond compliance: Societal external review as a driver of curriculum renewal for the twin transition. European Journal of Education. https://doi.org/10.1111/ejed.70440

Boyatzis, R. E. (2008). Competencies in the 21st century. Journal of Management Development, 27(1), 5–12. https://doi.org/10.1108/02621710810840730

Campion, M. A., Fink, A. A., Ruggeberg, B. J., Carr, L., Phillips, G. M., & Odman, R. B. (2011). Doing competencies well: Best practices in competency modeling. Personnel Psychology, 64(1), 225–262. https://doi.org/10.1111/j.1744-6570.2010.01207.x

Carless, D., & Boud, D. (2018). The development of student feedback literacy: Enabling uptake of feedback. Assessment & Evaluation in Higher Education, 43(8), 1315–1325. https://doi.org/10.1080/02602938.2018.1463354

Evangelista, L., & Fonzo, C. (2023). La metodologia europea della Peer Review: prima sperimentazione tra istituti scolastici e Centri di Formazione Professionale in Rassegna CNOS. Problemi, esperienze, prospettive per l’istruzione e la formazione professionale, 39.

Feldman, M. S., & Pentland, B. T. (2003). Reconceptualizing organizational routines as a source of flexibility and change. Administrative Science Quarterly, 48(1), 94–118. https://doi.org/10.2307/3556620

Fonzo, C. (2022). Università: uno sguardo verso il futuro. Quaderni di Comunità: persone, educazione e welfare nella società 5.0: 2, 2022, 35-40.

Harvey, L., & Stensaker, B. (2008). Quality culture: Understandings, boundaries and linkages. European Journal of Education, 43(4), 427–442. https://doi.org/10.1111/j.1465-3435.2008.00367.x

Mandinach, E. B., & Gummer, E. S. (2016). What does it mean for teachers to be data literate: Laying out the skills, knowledge, and dispositions. Teaching and Teacher Education, 60, 366–376. https://doi.org/10.1016/j.tate.2016.07.011

Schippmann, J. S., Ash, R. A., Battista, M., Carr, L., Eyde, L. D., Hesketh, B., Kehoe, J., Pearlman, K., Prien, E. P., & Sanchez, J. I. (2000). The practice of competency modeling. Personnel Psychology, 53(3), 703–740. https://doi.org/10.1111/j.1744-6570.2000.tb00220.x