Designing Coaching for School Improvement: Cost, Impact and Evidence

Reflecting on PSA’s experience over the past seventeen years of designing school improvement programmes, coaching models stand out as a major design consideration for two reasons.

Firstly, they can be highly impactful. Numerous studies suggest that training alone often results in relatively low levels of sustained practice change, with implementation rates typically below 10-20%. Where training is followed by sustained coaching support however, implementation of new practices can increase dramatically, often reaching levels above 70-80%.

Secondly, coaching can have a major impact on programme costs. In many large-scale initiatives, coaching engagements, particularly when travel, accommodation and logistical support are included, can account for more than 50-60% of the total implementation budget.

For organisations working at scale, the choice of coaching model therefore requires careful and intentional design, balancing the potential impact of coaching against the financial realities of delivering it.

Approaches to Coaching

Broadly speaking and in our experience, two approaches to coaching can be identified, each with its own strengths and limitations:

• Unstructured coaching – here the coach engages a school leader or teacher in a conversation largely driven by the participant’s immediate challenges. These may involve personal leadership issues, staff management concerns, or broader institutional difficulties. The coach’s role is primarily to guide reflection and help the participant think through possible responses.

• Structured coaching – this usually follows formal training workshops. The training introduces specific practices or methods that participants are expected to implement in their schools. The role of the coach is then to support the implementation of those practices. Coaching sessions therefore focus on whether the practices are being applied, what difficulties are being encountered, and how implementation can be improved.

  
  

PSA is increasingly moving towards hybrid engagements in which both approaches are combined. Here, coaching supports the implementation of specific practices introduced through training while also allowing space for school leaders or teachers to raise contextual challenges that they may be dealing with. This allows coaching to remain grounded in concrete practices while still responding to the realities of changing school environments.

Cost Drivers in Coaching Design

Once the purpose of coaching has been established, programme designers need to consider the factors that determine the cost of delivering coaching. Three cost drivers are particularly important.

The first is the Mode of coaching delivery. Coaching may take place face-to-face within the school, online through digital platforms, or through a blended combination of the two.

Face-to-face coaching allows coaches to observe practices directly, engage with the broader contextual factors and build stronger relational trust with school leaders and teachers. Online coaching, on the other hand, is more flexible and significantly reduces travel time and logistical costs. Its effectiveness, however, depends heavily on the digital maturity of the school(s) and their supporting infrastructure and it can limit the coach’s ability to observe real-time practice and organisational dynamics.

The second cost driver is Configuration which relates to with whom and how coaching is conducted.

In one-to-one coaching, a coach works directly with an individual school leader or teacher. This approach allows for deep engagement with a participant’s specific context and challenges and can be particularly valuable when supporting principals or senior leaders dealing with complex organisational issues.

In group coaching, a coach works with several participants simultaneously. This may involve groups of teachers within a school or clusters of leaders across schools. Group coaching allows programmes to reach larger numbers of participants at lower cost and can also encourage peer learning and shared problem solving.

A more complex form of group coaching - embedded coaching - occurs when the coach participates directly in existing organisational interactions such as School Management Team meetings, phase meetings, or discussions between Department Heads and teachers. This form of coaching can be particularly impactful and cost effective but requires a high level of coaching competency.

The third cost driver is Frequency or Dosage, which refers to the regularity and duration of coaching engagements over time. In our experience, coaching engagements often work well at around two to three hours per month, typically in one or two sessions. More intensive engagement can begin to interfere with teaching time and may lead to participants becoming overly dependent on the coach, while engagements that occur less than once a month can result in a loss of continuity. Dosage can also be adjusted according to school progress - schools that have already embedded key practices generally require less frequent support, while schools that are still struggling may benefit from more intensive coaching.

Coaching Impact

Looking beyond cost drivers, the question of coaching Impact is critical. In practice, many programmes simply do not measure the impact of coaching at all. Some may point to improvements in learner outcomes as evidence of success, however, the link between coaching and learner performance is often indirect and difficult to establish clearly.

Where specific professional practices can be shown to be associated with improved learner outcomes, it makes far more sense to measure coaching impact at the level of practice change.

One way of measuring this is through the use of coaching rubrics that describe observable leadership or teaching practices and track improvements in those practices over time.

Below is an example of a simple rubric reflecting an aggregate score per school of the practices of all Grade R teachers. These scores are based on more detailed rubric criteria which aggregate into three categories: Emotions, Engagement and Evaluation. Data from every coaching engagement is captured and stored in a central data repository, making it possible to track progress longitudinally.

Red (0–24%) = Not in place. Yellow (25–49%) = Partly evident. Orange (50–74%) = Growing Evidence. Green (75–100%) = Consistently Demonstrated

Rubric-based coaching allows programmes to track incremental improvements in practice and gives coaches and participants a shared framework for discussing progress. Coaching discussions can therefore focus on specific observable behaviours rather than general impressions. Importantly, this approach also makes it possible to correlate improvements in professional practices with learner assessment outcomes over time.

Coach Capability

The quality of a coaching intervention ultimately depends heavily on the experience and competence of the coaches selected. Even well-designed coaching systems can fail if coaches lack the credibility, judgement or organisational awareness required to work effectively within schools. In our experience three elements appear particularly important:

• Credibility and relevant experience - coaches need professional experience that school leaders and teachers recognise as relevant. Without this credibility, coaching relationships struggle to gain traction.

• Consistency in interpreting programme practices - Where coaching programmes rely on defined practices or rubrics, coaches must interpret these in a consistent way. If different coaches interpret rubric indicators differently or give conflicting advice, the programme quickly loses coherence. Regular training and calibration are therefore necessary.

• Ability to navigate organisational dynamics - coaching takes place within complex leadership environments. Coaches must be able to recognise and respond to contextual factors such as leadership tensions, transitions and organisational disruptions that can undermine improvement efforts. Where such issues arise, coaches may need to support leadership alignment or use group-based engagements to strengthen collaboration within leadership teams.

Towards a Coaching Model

The discussion above suggests that effective coaching models emerge from a sequence of deliberate design choices.

The starting point is clarity about the purpose of the coaching intervention - whether it is primarily intended to support the implementation of defined practices following training (structured coaching), to provide reflective support for leaders and teachers facing contextual challenges (unstructured coaching), or to combine both approaches in a hybrid model.

Programme designers should then consider the cost drivers that shape how coaching will be delivered. These include the mode of delivery, the configuration of coaching interactions and the frequency of engagement. Together these factors largely determine the financial sustainability of the coaching programme.

At the same time, coaching engagements must remain focused on measurable improvements in professional practice, ensuring that the impact of coaching can be observed and tracked over time.

Finally, the quality of delivery and capability of the coach remains central. Credible, well-prepared coaches who can interpret programme priorities consistently and work effectively within complex school environments are essential for coaching interventions to succeed.

Final Remarks

Designing an effective coaching programme requires a deliberate and structured process which will be significantly strengthened if it is supported by good historical data.

Once a programme is launched, programme leaders need good data on how often coaching engagements take place, who participates, what practices are being addressed and whether those practices are changing over time. Without this level of operational data costs cannot be controlled, impact cannot be assessed and quality cannot be ensured.

Coaching then risks becoming an activity sustained by assumption and belief rather than evidence - an act of blind faith rather than a data-driven decision.