Online care platforms change their administrative flows regularly: how a person books a slot, how reminders are sent, how intake forms are collected, how follow-up messages arrive. None of these changes is clinical, yet each can confuse real people if it is rolled out carelessly. This piece compares three common ways of planning such a change, with particular attention to the pre-mortem, an exercise in which a team imagines the new booking flow has already confused users and lists the reasons why. It is a discussion of administrative planning only. Nothing here is medical advice, and nothing here concerns diagnosis or treatment.
The three approaches in brief
- Launch and learn: release the change and fix problems as complaints arrive.
- Checklist-led: work through a standard list of launch tasks and tick them off.
- Pre-mortem-led: imagine the change failed, collect the reasons, then adapt the plan.
Most teams blend them, but the balance matters. The comparison below looks at five dimensions: what each approach catches, what it costs, who participates, how it behaves under time pressure and what it leaves behind.
Dimension 1: What each approach is good at catching
Launch and learn catches everything eventually, since real users find real problems. The drawback is that those users are the ones inconvenienced while it happens. Checklists catch known, repeatable issues, such as whether links work, whether confirmation emails send and whether the form fields save. They are weaker on problems nobody anticipated because a list only contains what someone previously thought of. The pre-mortem is aimed at the unknown-but-plausible: the misunderstandings, the mismatched expectations and the awkward moments that surface when real people meet a new flow.
Dimension 2: Cost and effort
Launch and learn looks cheapest up front and tends to cost most in rework and goodwill. Checklists are inexpensive and repeatable once written. A pre-mortem takes a short meeting, often less than an hour, and no tooling. Its main cost is attention from the people who normally prefer to be building. In most organisations that trade is favourable, because an hour spent imagining failure typically replaces many hours of reactive fixing.
Dimension 3: Who takes part
Launch and learn involves the product team and later the support team. Checklists belong to whoever owns the launch. The pre-mortem benefits from a wider circle: the front-desk or support staff who answer questions, the person who writes the reminder messages, a representative of the groups who will use the service, and someone who has never seen the design. Different people imagine different failures. For platforms serving people who are unwell or anxious, it is worth including those who know how users feel on first contact. Pieces such as a reassuring explanation of how remote counselling sessions function and an introduction to understanding online ear, nose and throat services show how much explanation people expect from the experience itself.
Another useful voice is the person who handles follow-up after the first contact, as in an account of how remote guidance and follow-up are arranged for sleep concerns. Follow-up messages are administrative moments too, and a pre-mortem should walk through them as carefully as the first booking. Ask what a person sees on the second visit, not only the first, and what happens when they return after a long gap.
Dimension 4: Behaviour under time pressure
When deadlines tighten, checklists survive because they are quick, and launch and learn becomes the default because it needs no preparation. The pre-mortem is the first thing cut, which is a pity, because a short version works. Ten minutes with three people can still surface the top few failures. The important point is that scarcity of time argues for a shorter pre-mortem, not for none.
Dimension 5: What each leaves behind
Launch and learn leaves a bug list and some frustrated users. A checklist leaves a completed form, which is useful as evidence of diligence but says little about what was missed. A pre-mortem leaves a ranked list of plausible failures, each with an owner and an early-warning sign. That list is reusable: the next change inherits the questions from the last one.
A side-by-side summary
- Best for known tasks: checklist.
- Best for unknown friction: pre-mortem.
- Best for catching real-world surprises you could not imagine: launch and learn, which is why monitoring after release still matters.
- Cheapest to prepare: launch and learn, at the expense of users.
- Most reusable output: pre-mortem.
Combining them in practice
The strongest approach is sequential. Run a short pre-mortem early, while the design can still change. Turn its findings into extra lines on the launch checklist. Release with monitoring and a fast route for questions. Then let genuine feedback refine the plan. Each method covers the gaps of the others.
What a pre-mortem might surface for a booking change
To make the comparison concrete, here are imagined reasons a new booking flow could confuse people. They are illustrative.
- The confirmation message used wording that did not say whether the booking was final.
- Time zones or time formats were unclear to users in different places.
- People with older phones could not complete a step, with no alternative offered.
- Reminder messages arrived at inconvenient hours or not at all.
- Support staff were not briefed and gave contradictory answers.
- Rescheduling was easy to start and hard to finish.
Every item is administrative, and every one can be fixed before launch at low cost.
A note on method from the software side
Software teams that look after their release machinery think in the same terms, and the blog at DevSecOps Now, aimed at practitioners who protect continuous integration and delivery pipelines, hints at how routine imagined failure is in that discipline. Health platforms can borrow the attitude without borrowing any technical detail.
Verdict
If you can run only one approach, run the checklist. If you can run two, add the pre-mortem, because it asks the questions a checklist does not know to ask. Keep launch and learn as your safety net rather than your plan. For any service people use at a vulnerable moment, a confusing administrative step is a real cost, and the effort to avoid it is modest. As always with health platforms, keep clinical guidance with qualified professionals and keep the planning discussion to the administrative layer around it.