The question the September session opened
The session distinguished a protected space for co-learning from consultancy commissioned around a predetermined output. It then asked whether learning could itself be a deliverable. The discussion raised the question; it did not settle a contracting or evaluation method. What follows is a proposed development to try in practice.
‘We will learn together’ is too loose on its own. It can conceal an absence of purpose, evidence, or accountability. Yet promising a solution before inquiry begins can be equally misleading. A useful contract makes the inquiry, the client’s participation, and the decisions that learning will inform explicit.
The central test is: what will the client be better able to notice, judge, decide, or do, and what evidence would justify that claim?
Separate four things
| Element | What can be agreed | What would count as evidence |
|---|---|---|
| Contracted work and outputs | Questions, participation, access, inquiry activities, records, review meetings, and an agreed decision process. | The work took place to the agreed standard; records show what was examined, including contrary evidence. |
| Learning | Changes in understanding, assumptions, options, or judgement that matter to the purpose. | People can explain what changed, why, and what difference it makes to a live decision. |
| Capability and practice | The ability to keep inquiring and acting after the consultant leaves. | A team repeats the review process, tests a new assumption, or changes a routine without consultant prompting. |
| Wider outcomes | Effects on the service, organisation, or people whose lives are affected. | Appropriate outcome evidence over time, with attention to other influences and unequal effects. |
You can commit to carrying out a disciplined inquiry and producing usable evidence. You can create and agree conditions for learning. You cannot guarantee that people will change their understanding, or that a short assignment will cause a complex public outcome. Agree the ambition and the limits without making the contract empty.
A practical learning agreement
| Agree | Make it concrete |
|---|---|
| Purpose and decision | Name the live question and the decision the learning must inform. Who has authority to make it? |
| Starting account | Record current explanations, evidence, assumptions, disagreements, and known gaps before beginning. |
| Participation and permission | Identify who must contribute, including people affected. Agree access, time, support, and permission to surface unwelcome information. |
| Inquiry and evidence | Agree a proportionate mix of observation, existing information, conversations, and bounded tests. Specify what would challenge the starting account. |
| Responsibilities | Consultant: design, facilitation, analysis, candour, and records. Client: access, participation, decisions, resources, and follow-through. Tailor this split explicitly. |
| Review points | Put dates in diaries. At each review choose continue, adapt, extend, stop, or transfer. State who can make each choice. |
| Acceptance | Define what makes the inquiry and its outputs adequate. Do not make acceptance depend on a preferred finding or favourable story. |
| Resources and boundaries | Set time and cost limits, constraints on tests, information handling, and a process for scope changes. |
| Transfer | Agree who maintains the learning routine and what evidence will show that it can continue without you. |
A staged agreement can hold purpose and resource limits steady while allowing the method to change. For example, contract first for a bounded discovery phase, then use an explicit decision to commission further work. Avoid an open-ended promise of emergence with no review or stopping rule.
Worked example: a six-week inquiry
Illustrative case, not a reported client engagement. A public-service team believes repeated hand-offs are delaying help. It wants to decide whether to change a referral rule. Staff disagree about the causes, and people using the service experience the hand-offs differently.
| When | Work and evidence | Decision |
|---|---|---|
| Before the start | Sponsor names the decision owner; team protects time; participants agree how accounts will be used. Record current explanations, a small baseline, and the proposed rule change. | Is there enough access and authority to begin? If not, narrow or pause the work. |
| Weeks 1-2 | Trace a small, explicitly bounded set of journeys with staff and people affected. Compare cases that went well and badly. Record limits of the sample and competing explanations. | At review 1, select a test or decide that more inquiry is needed. Do not imply the sample represents every case. |
| Weeks 3-4 | Try a reversible change within agreed authority and safeguards. Record what happened, including workload shifts, exclusions, and unexpected effects. | At review 2, continue, adjust, or stop the test using the agreed evidence. |
| Weeks 5-6 | Compare the starting account with the evidence. Practise the review process with the team leading it. Prepare a decision record, remaining questions, and ownership for the next cycle. | The authorised client decides whether to adopt, extend, redesign, or reject the proposed rule change. |
The contracted outputs are a shared starting account, an evidence and assumptions record, a bounded test where authorised, two interim reviews, a final decision record, and a transfer exercise. The hoped-for outcome is more useful and timely help. Six weeks cannot by itself establish that outcome or attribute it to the consultant.
A decision not to change the rule can be a successful inquiry if the evidence supports it. Missing access, unsupported conclusions, or a review that never informs a decision are shortcomings to address, not successes to relabel as learning.
How would we recognise learning?
| Look for | Useful evidence | Insufficient on its own |
|---|---|---|
| A changed explanation | Before-and-after accounts identify an assumption that changed, the evidence, and the implications. | ‘We had some good conversations.’ |
| A changed decision | A decision record shows how evidence affected a choice, including a justified no-go. | A report was accepted or a workshop scored highly. |
| A changed practice | Observe the new routine in use; examine cases where it breaks down or excludes people. | A commitment or an action list. |
| Retained capability | The client leads another inquiry or review without relying on the consultant. | People attended training or received a toolkit. |
| A plausible contribution to wider outcomes | Trace the reasoning between action and effect, look for other explanations, and check who benefited. | A favourable indicator treated as proof of causation. |
Agree a small number of useful indicators with the people doing and experiencing the work. Include disconfirming evidence, effort and cost, and consequences for people whose voices are easily missed. Counting ‘lessons learned’ rewards volume; it says little about whether judgement improved.
Possible wording for a scope of work
‘The purpose of this phase is to enable [decision owner] to decide [live question] on a better-evidenced basis. We will jointly examine [questions], test [assumptions], and record evidence, disagreements, and remaining uncertainty. The consultant will provide [contribution]. The client will provide [access, participation, authority, and resources]. At [review dates], we will decide whether to continue, adapt, stop, or agree a further phase. Acceptance concerns the quality and completeness of this agreed work, including evidence that challenges the starting proposal; it does not require a predetermined finding.’
This is practice wording to adapt with the client and the organisation’s contracting requirements. It is not a ready-made legal clause.
Where transformative outcomes can help
The Transformative Innovation Policy Consortium’s formative evaluation approach connects learning objectives, emerging outcomes, and revisable theories of change. Its setting is innovation policy and systemic transformation. It offers a useful adjacent resource for asking whether experiments are changing the conditions for further change, while addressing timing and attribution difficulties.
My suggestion here is narrower: use a revisable account of how the work might contribute, and review that account against experience. Do not import a transformative-outcomes vocabulary merely to make a small assignment sound ambitious. Read TIPC’s methodology and Molas-Gallart, Boni, Giachi, and Schot’s 2021 paper.
An invitation to test this
Take a real brief that asks for a report, workshop, or recommendation. Keep the useful output. Add one learning question, one live decision, one way of testing the starting account, and one review at which changing direction is allowed. Then ask the client what would make that agreement credible.
The unresolved practical question is often power: can the people learning actually change anything? If their answer is no, the contract needs a route to someone who can decide, or a more modest purpose. The five core practices help connect that agreement to the organisation’s everyday conversations, clarity, and learning routines.
Download the contracting-for-learning note · Use the editable worksheet