Follow step two of the route and run a build task before anyone talks about the system: Lego Serious Play or an equivalent exercise where the group makes something together. The page puts the warm up there because the group is making shared objects before it starts describing a system, and that sets the register for everything after it. The order matters for the rest of the route: the warm up comes first, and the personas walk the pathway afterwards, at step three. So the room makes something together, and only then starts naming the people who will walk the pathway.

The route exists so that the logic model is defensible. This page says the method ends in a logic model and that everything before it exists to make that model traceable: each row should be traceable back to something a person said in the room, or to an established framework. Skipping coding removes the link to the Theoretical Domains Framework, to mechanisms of action and to CFIR at the organisational level. Skipping the causal loop diagram and the leverage points removes the step where the group sees which nodes do the most work and where a change might spread. Technique selection then rests on facilitator instinct rather than on the evidence, and the logic model loses the traceability the route exists to provide.

A workable statement: this map covers one adult's route from first contact with the general practice, through the community team, to hospital admission and back home, over the three months around that admission, which is the point in time the map is bounded by. That puts the person, the organisations and the time window into one sentence the group can check. What it leaves out is everything outside that edge: housing and employment, which shape the route but sit outside the boundary the team agreed. Boundary setting is a decision rather than a discovery, normally made by the team plus whoever is paying, and this page asks for it to be written down and agreed with the group in the first ten minutes. Recording it lets later readers see the map is partial.

The map is a set of claims about the service, made by particular people on a particular day, not a description of the service itself. Three limits apply. It is partial, bounded by the knowledge and conditions of the people who made it at one point in time. It is not reproducible, because a different group would have drawn a different map, and that is not a design fault. It flattens difference, because relations often work differently for different groups, so inequalities are hard to show, which is why the method relies on diverse participation to add lived experience context. The map should be presented as a hypothesis to test, and the team should say so rather than letting the board treat the picture as the service.

Choose a causal loop diagram. The aim here is explanation, not simulation: a causal loop diagram is a qualitative picture of structure, with factors linked by positive and negative connections forming reinforcing and balancing loops, which matches a question about how the system produces its outcomes. It has no data requirements, so the missing dataset is not a barrier, and it suits a group that can work visually. It is also the most used method in population health, so there are examples to copy. The plausible alternative, a systems dynamics model, would need quantifiable data, specialist software and mathematical specification that this project does not have. Pick the simplest method that answers the question.

Rigour is not the ranking this Framework uses: the page says method choice is not a quality ranking and that you should pick the simplest method that answers your question. A Bayesian belief network needs quantifiable data, so the first choosing question has to be answered with states and conditional probabilities that the project may not have, and it is typically acyclic, so it cannot carry the feedback loops that a causal loop diagram is built from. It also sits at the far end of the capacity gradient: extensive capacity building and skills, statistical work, and it is hard to update live. Unless the output must be probable causes or outcomes under uncertainty, the colleague is buying effort the project cannot use.

Before the workshop: is there a written boundary and a stated purpose? A map without a boundary cannot be finished, only abandoned. Before coding: does every touch point have a location, a person and a quote? Coding needs a source to point at. Before the causal loop diagram: has the journey material been through step five, the missing personas? A diagram built from incomplete journeys quietly reproduces the gaps. Before the leverage analysis: is every variable a quantity that can rise or fall? People and activities are not variables. Before strategy selection: has every determinant been given an actor and a behaviour? A framework label on its own is not a finding. Before anything is used: can every claim be traced to a quote or a framework term? That traceability check is the one to keep if you keep only one.

The method does not describe the service. It describes how the people in the room understand the service, and the gap between the two is data, not error. It does not estimate effect: nothing in the route produces an effect size, a cost or a probability of success. It is not designed to be reproducible: the guidance is explicit that a different group would draw a different map, and that this is not a fault, because diversity of the group is what protects you rather than repeatability. It has no gold standard for leverage, since there is no agreed method of identifying leverage points. That fourth limit obliges you to say which route you used: counting connectivity, network analysis, or expert judgement.

10.00 Welcome, purpose, boundary, and what happens to the material. Scribes write the boundary as stated. 10.15 Build-based warm-up in small groups: build the journey someone has to make to get help, no words on bricks. Scribes write photographs and the group's words. 11.00 Break. 11.15 Persona walkthrough, one persona at a time, whole room. Scribes write the steps in the group's words, who does what, and what the persona must do. 12.30 Lunch. 13.30 Journey mapping in small groups, then play-back. Scribes write contradictions between groups, quoted both ways. 14.30 Touch point pass: where each persona feels something, and what. Scribes write one card per point, with a quote. 15.15 Missing personas: who does this journey imply but not name. Scribes write the names put forward and by whom. 15.45 Close: what happens next, and when they will hear from you. Scribes write the commitment and the date.

Card 1. Persona and step: parent, referral. What happened: watched the health visitor write the referral with no confirmation. What was felt: uncertainty, mild. Quote, parent: 'I did not know whether anything would happen.' Level: person. Source: parent, touch point pass, session date. Uncertainty: low. Card 2. Persona and step: parent, waiting. What happened: three weeks with no contact. What was felt: not knowing. Quote, parent: 'I heard nothing for three weeks.' Level: person. Source: parent, touch point pass. Uncertainty: low. Card 3. Persona and step: parent, ringing the service. What happened: rang four times and was told to wait for a letter. What was felt: like a nuisance. Quote, parent: 'I felt like a nuisance for chasing.' Level: person. Source: parent, touch point pass. Uncertainty: low. Card 4. Persona and step: receptionist, letters. What happened: only staffed to send letters weekly, with no way to tell a caller when a letter will come. What was felt: not recorded, because this is a service account rather than the receptionist's feeling. Quote, receptionist: 'We are only staffed to send letters weekly, and I have no way to tell a caller when the letter will come.' Level: organisation. Source: receptionist, touch point pass. Uncertainty: high, resting on one staff account; a rota and the letter turnaround would resolve it. The receptionist's material stays at organisation level, so a service condition is not coded as the parent's psychology.

Quote 1, the parent ringing: TDF domain Environmental Context and Resources, with the actor and the behaviour named as the parent chasing a letter, confidence high because the quote is explicit, and a source location so the judgement can be traced. Level: person. Quote 2, the receptionist: the service is only staffed to send letters weekly, which is a service condition, so it stays at system level and is coded to CFIR as an organisational barrier rather than reduced to a parent's psychology. Level: organisation. Quote 3, the parent: Emotion, because not knowing is the state the touch point work records. Confidence medium rather than high, with the working shown: the quote states that nothing was known rather than naming a feeling word, so the code rests on the touch point reading of it. Level: person. Record overlaps rather than forcing an item into one domain, and check that a sceptical reader could reconstruct each judgement from the quote alone.

The Theory and Techniques Tool links behaviour change techniques to mechanisms of action, built from a literature synthesis and an expert consensus exercise, with the agreement between the two shown for each link. Read plainly, it shows that experts and the published literature associate a given technique with a given mechanism. It does not show that the technique works, and it does not show that it would work in your service, with your staff, at this step of your pathway. A cell with no link is an answer: report no linkage in the tool, not a plausible substitute. The CFIR-ERIC tool encodes expert consensus about which strategies address which barriers, not trial evidence, so a level two strategy must never be presented as level one. Where a recommended strategy is not feasible here, write that down and say why.

Variables: hours of staff overtime per week; staff exhaustion; clarity of the written information sent to patients; number of times a carer has to chase information. Links: hours of overtime affects staff exhaustion, positive, with a delay. Staff exhaustion affects clarity of the written information, negative. Clarity of the written information affects number of chases, negative. Number of chases affects hours of overtime, positive, because every chase consumes staff time. The four links close one loop that contains two negative links. Two of these links are the team's reasoning rather than something the material states: overtime to exhaustion, and chasing back to overtime. Treat both as weak. Two is even, so the loop is reinforcing: exhaustion erodes clarity, unclear information creates chasing, and chasing adds overtime, which deepens exhaustion.

The ranking is missed appointments, then staff morale, then waiting time, then clinic rooms. The count is a fact about the diagram and can be done by hand, and a variable sitting on several loops does more work than one sitting on none, so the ranking tells you where to look and what the top variables connect. It does not license calling the top variable a leverage point. A high count can be an artefact of how the map was drawn, since splitting one messy topic into four tidy variables or merging four steps into one changes it, so check whether the rank survives a redraw. Two people working from the same notes who get different counts have not yet produced a finding. There is no gold standard method for locating leverage points.

Run the traceability check over the list. For each technique, quote the tool extract row it came from and state whether the tool says it directly, implies it, or does not support it; anything the tool does not support moves to the unsupported list. Then check each number and label against the taxonomy itself: the behaviour change technique taxonomy version 1 lists 93 techniques in 16 clusters, so a number or label that is not in that list is reported as a framework term that is not in the framework. Record the omissions too: where the tool has no entry for a determinant, the correct output is "no linkage in the tool", and the row stops there rather than being filled with a plausible technique.

Add these instructions. For each judgement, quote the material it came from and say where the quote sits. Label anything the model added as INFERRED and say what would confirm it. Where the material is silent, write NO EVIDENCE and leave the row empty, because a gap is an answer and filling it with a plausible invention is the main failure mode. Where a framework or tool has no entry for something, say that it has no entry and stop, and never claim a linkage the tool does not contain. End with a list of what could not be determined. Before anyone relies on the output, run the traceability check over it.

Inputs: staff time for reception training, the text messaging system, and the chasing work the journeys revealed, the last labelled DERIVED FROM THE MAP where the journeys recorded it. Activities: the reminder text and the explanation at booking, each named exactly as in its taxonomy, labelled ASSUMED because the plan is the team's reasoning. Short-term outcome: the determinant the material was coded to, named with its framework, labelled DERIVED FROM THE MAP, and stated as a wish because it has no measure yet. Medium-term outcome: patients attend the appointments they booked, measured as attendance recorded at every appointment, labelled MEASURED IN THIS PROJECT. Impact: capacity released for other patients, reached by the chain from attending, labelled ASSUMED and collected with the other assumptions at the end.

The matching tool encodes expert agreement about which ERIC strategies fit which CFIR barriers. It is not trial evidence that a strategy works, in this service or in any other, and it supplies no causal chain from strategy to outcome, so any mechanism given here is our reasoning and is labelled as such. Each recommendation is reported with its tier: level 1 where more than half the expert panel ranked it in their top seven for that barrier, level 2 where 20 to 50 per cent did, and a level 2 strategy is never presented as though it were level 1. Strategies the service cannot deliver are named with the reason rather than dropped. The tool is a consensus product revised over time, so the version or access date is recorded.

A determinant is a factor that influences the target behaviour, located in an established framework: a Theoretical Domains Framework domain at the individual level, a CFIR construct at the organisational level. A mechanism of action is the process through which a technique produces change. It is not a separate framework: the list of 26 used by the Theory and Techniques Tool is the 14 TDF domains plus 12 further constructs from a set of 83 theories of behaviour change, so a determinant coded to a domain usually has a matching mechanism entry. A behaviour change technique is a discrete, observable component of an intervention, named and numbered in behaviour change technique taxonomy version 1, which holds 93 techniques in 16 clusters. An implementation strategy is a discrete method for adopting, sustaining or scaling an intervention, named in the ERIC taxonomy, which holds 73 strategies in nine clusters. Both are named from their taxonomy so that someone outside the team can recognise what was done.

The method guidance is the participatory systems mapping guidance (Blake and Rigby et al. 2024), cited with the version used. The barrier-to-strategy source is the CFIR-ERIC matching tool, which maps 39 CFIR barriers to ERIC strategies and tiers each pairing by expert endorsement; because it is a consensus product revised over time, record the version or access date and the tier of every recommendation you report. The link from mechanisms of action to behaviour change techniques comes from the Theory and Techniques Tool, with the technique numbers and labels taken from behaviour change technique taxonomy version 1. The guidance, the taxonomy and ERIC are cited with the version used.