Read enough incident reports and a pattern surfaces. The person at the centre of a near-miss almost always knew the rule. They could have recited the isolation procedure word for word. What went wrong wasn’t recall — it was the call they made in the thirty seconds before the rule was relevant, under time pressure, with a supervisor waiting and a job half-finished. The gap between knowing the rule and making the right call when it counts is where most safety failures actually live.
A linear course can’t see that gap. It walks the learner through content, asks them to recognise the correct answer from four options, and records a pass. Recognition is a real skill, but it isn’t judgement. Branching scenarios are how Lattice Learn tests the second thing.
What a branching scenario actually does
A branching scenario presents a situation, gives the learner the context they’d have in the field, and then asks them to decide — not to pick the textbook answer, but to choose what they’d do. The path responds to that choice. Pick the shortcut and the scenario plays out the consequence of the shortcut; pick the careful option and you see where that leads. The route through the material is determined by the decisions the learner makes, not by a fixed page order.
This isn’t a quiz wearing a costume. In a quiz, a wrong answer is marked and the learner moves on having learned, at most, that they were wrong. In a branching scenario, a wrong call takes you somewhere. You arrive at the point where the corner you cut becomes the problem you now have to manage, and the lesson lands as experience rather than as a red cross in a results table.
Underneath, the chosen path is auto-determined — the engine reads the learner’s choice and routes them — and every one of those decisions is written to the automated-decisions log. That last part matters more than it first appears. When the scenario is finished, you don’t just have a pass or a fail; you have the sequence of calls the learner made to get there. You can see that someone reached the right outcome on the second attempt after initially choosing to proceed without isolating, or that they took the conservative path at every fork. A completion stops being a single bit of information and becomes a record of judgement under the conditions you designed.
Authoring on a canvas, not in a spreadsheet
Scenarios are built on a visual canvas. Each node holds content — text, a video, an image, a document, framing for the decision — and offers a set of choices, and each choice draws an edge to the next node. You’re laying out a map of how a situation can unfold, and you see the whole shape at once: where the paths diverge, where they rejoin, which branches end well and which end in the consequence you want the learner to sit with.
Branching is one block type among several in the same block-structured editor that produces the rest of the course library — text, video, image, document, quiz, and SCORM blocks live alongside it. So a course can open with a video, move through some reading, drop into a branching scenario for the part that needs judgement, and close with an auto-graded check. The branching section isn’t a separate tool you bolt on; it’s a block you add where the material calls for a decision instead of a fact.
And because scenarios are courses, they inherit everything courses already have. Copy-on-write versioning, so a published scenario is immutable and editing it produces a new version rather than silently rewriting history. The author to reviewer to publisher approval chain, so no scenario reaches a learner without someone signing off on the branches. And the accessibility publish gate — a scenario cannot be published until it clears WCAG 2.2 AA, the same bar as every other piece of content in Lattice Learn.
The UX problem: keep the learner in flow
Here’s the part I care about most as the person responsible for how this feels. The whole value of a branching scenario is that the learner is in the situation. The moment you break that immersion — pop a modal, interrupt with a “Question 3 of 8” header, switch the visual language to test-mode — you’ve told the learner this is an exam, and they revert from deciding to answering. They start hunting for the response the system wants rather than the response they’d actually make. The judgement you were trying to observe evaporates.
So the design rule is: content and a clear choice, never a quiz interruption. The decision point reads as a continuation of the story, not a break from it. The options are presented as actions a person would take, phrased the way the learner would think about them, not as labelled A/B/C/D answers. There’s no score ticking in the corner, no countdown, nothing that signals you are being marked right now. The grading and the logging happen behind the scenes; the experience in front of the learner is just a situation and a decision, then the next situation.
That experience has to hold up on a phone. Field-based staff, contractors, anyone doing their training between jobs — they’re on mobile, often standing up, often on a patchy connection. Scenarios are designed mobile-first: the content and the choices are legible and tappable on a small screen, the branches resume cleanly from the My Learning portal if someone steps away mid-scenario, and the whole flow meets WCAG 2.2 AA so it works with a screen reader, with keyboard navigation, and at the contrast ratios people with low vision actually need. Accessibility here isn’t a checkbox we satisfy and forget — it’s the difference between a scenario everyone on the roster can complete and one that quietly excludes part of the workforce.
Why the logging is the point
It’s worth circling back to where this connects to the rest of the platform. Lattice Learn doesn’t treat a finished scenario as an export to be reconciled elsewhere. The completion is a compliance input in the same schema, so it does the same downstream work any completion does — feed the live compliance score, materialise a skills record, satisfy a mandatory rule — described in full in what changes when a course is finished.
But branching adds something the other block types can’t. A multiple-choice pass tells you the learner could recognise the right answer. A branching record tells you the calls they made when the situation didn’t announce which answer was correct. For the work that gets people hurt — the judgement calls under pressure that no checklist fully covers — that’s the evidence worth keeping.