Pick up almost any health conference programme and read it as if you were a finance minister with one question: after these three days, what will be different? Most programmes can’t answer that. The session titles are hopeful and wide. “The Future of Primary Care.” “Innovation in Health Financing.” “Partnerships for Impact.” Each has five speakers, forty-five minutes, and a moderator whose job is mostly to keep time.
Nobody walks out of “The Future of Primary Care” with a decision. Nobody expected to.
This matters more than it seems. A large health conference gathers hundreds or thousands of people who, between them, control budgets, guidelines, procurement and research agendas. For a few days they sit in one building. The agenda decides what they spend that time on. When the agenda is built for visibility, the room spends its most expensive hours watching people be visible.
The argument here is simple, and I think underappreciated: a conference agenda is not a list of topics. It is a set of choices about what kind of disagreement is allowed in public. Most agendas choose, without saying so, to allow none. That single design decision explains most of what people complain about afterwards.
The agenda is a market, and the buyers are not the audience
Start with where sessions come from, because the answer is rarely “someone asked what the audience needs to decide.”
Sessions come from sponsors who have paid for a slot. From partner organisations that expect a stage in return for promotion. From abstract submissions, which reward whoever wrote an abstract, not whoever holds the most useful problem. From the organisers’ need to have recognisable names on the poster, because names sell registrations. And from the diplomatic arithmetic of making sure every region, agency and constituency feels included.
None of this is corrupt. Conferences cost money, and sponsors and partners pay a lot of it. But notice who the customers are. The sponsor is buying exposure. The partner is buying legitimacy. The speaker is buying a line on a CV and a photograph. The organiser is buying next year’s registrations. The delegate, the person whose decisions the conference supposedly informs, is the one party not directly buying the session design. They buy a ticket to the whole thing and accept whatever sessions the other buyers have produced.
So the agenda optimises for the people who pay for slots. Exposure is best served by many speakers and uncontroversial titles. Legitimacy is best served by being on the same stage as other legitimate organisations. None of those buyers benefits from a session that ends with a hard choice, because hard choices produce losers, and nobody pays to lose in public.
Panel inflation: five speakers, zero arguments
The clearest symptom is the swollen panel. Five or six speakers, each with a prepared opening statement of “just a few minutes,” which always runs long. By the time the moderator reaches the second question, twenty-five minutes have gone. Audience questions get the last seven.
Panel inflation happens for reasons that are rational from the organiser’s side. Every extra seat on a panel is another organisation that promotes the event, another logo, another relationship maintained. A five-person panel costs the same room time as a two-person debate and satisfies two and a half times as many stakeholders.
The cost is invisible but large. With five speakers, no single argument can be developed past its headline. Speakers learn, sensibly, to use their slot for the message their organisation wants on the record. When the moderator does ask a pointed question, it goes to one person, who can deflect it knowing the next speaker will change the subject. The format all but guarantees that nobody is ever pressed twice on the same point.
There is a quieter effect too. Big panels are selected for representativeness: one ministry, one multilateral, one NGO, one private-sector voice, one academic. That looks balanced. But representation of institutions is not the same as representation of positions. You can put five organisations on stage who all hold the same view on the question at hand, and the audience will leave believing the matter has been examined from every angle.
The manel was a symptom of the same disease
For years the most visible criticism of conference panels was that they were all male. The criticism was fair and it had effect. In June 2019 the then director of the US National Institutes of Health, Francis Collins, said publicly that he would decline to speak at events where attention to inclusiveness was not evident in the agenda, and asked other scientific leaders to do the same. Plenty of organisers changed their habits.
But the “manel” was never only about gender. It was the easiest-to-see sign that panels were being assembled by habit and network: the organiser calls the people they already know, who are the people who have always been on stage. The same mechanism produces panels that are all from high-income countries discussing low-income health systems, panels with no one who has run a district hospital discussing hospital reform, and panels of funders discussing what implementers should do.
Fixing the gender balance of a panel while keeping the same assembly method is progress, and it is not the whole job. The deeper question is whether the people on stage are there because they hold a position the audience needs to hear tested, or because they were easy to invite.
No disagreement on stage, by design
Here is the uncomfortable centre of it. Most health conference sessions are engineered to avoid disagreement, and everybody involved prefers it that way.
Speakers prefer it because public disagreement with a funder, a ministry or a peer organisation has real costs and almost no benefits. Organisers prefer it because a speaker who has been embarrassed on stage does not come back, and neither does their sponsor. Moderators prefer it because they are often from the same small professional world and will see the speakers again next month. Even the audience, if honest, often prefers it, because agreement is pleasant and clapping is easy.
The result is a genre of session where everyone agrees that primary care is important, that data should be better, that partnerships matter and that communities must be at the centre. All of these are true. None of them is a decision. The real disagreements in global health, about whether to fund vertical disease programmes or general health systems, about how much to charge patients, about which services to leave out of a benefit package, about whether a particular pilot deserves national money, are exactly the ones that never make the stage. They happen in corridors, at dinners and in donor meetings, among people who already agree to disagree privately.
A conference that keeps real disagreements offstage isn’t neutral. It tells every junior person in the room that these questions are settled or unspeakable, when they are neither.
Sessions designed for visibility, not decisions
Put the pieces together and you get the dominant session type: a visibility session. Its success is measured in attendance, livestream views, social media posts and whether the sponsor was pleased. A decision session would be measured differently: did anyone in the room leave committed to something they were not committed to before?
You can usually tell which is which from the title. Visibility sessions have titles made of abstract nouns: future, innovation, resilience, transformation. Decision sessions have titles that could be answered yes or no. “Should low-income countries pay for new vaccines before donor co-financing ends?” is a question. “Sustainable Vaccine Financing in a Changing World” is a mood.
What a decision-grade agenda looks like
A decision-grade agenda is not a radically different event. It is a set of design choices that make trade-offs impossible to avoid.
Every session names the decision it serves. Before a session is accepted, someone writes one sentence: “At the end of this session, a [specific actor] should be better able to decide [specific thing].” If nobody can write that sentence, the session is a keynote, and there should be very few keynotes.
Formats that force trade-offs. A few that work:
- Structured debates with two speakers, a stated motion and an audience vote before and after. The shift in the vote is the result. It is crude, and it is far more informative than applause.
- Budget exercises. Give a panel a fixed hypothetical budget and a list of real, costed options, and make them allocate it live. Watching experienced people refuse to cut anything is itself instructive.
- Red-team sessions, where a group presents a programme or policy and a second group, briefed in advance, presents the strongest case that it will fail.
- Pre-mortems: “It is five years from now and this reform has collapsed. Why?”
Pre-reads. If a session is meant to support a decision, the audience should arrive knowing the options. A two-page brief sent a week ahead, with the options, the evidence for each and the cost, lets the session start where most panels end.
Fewer speakers, longer exchanges. Two or three speakers with sharply different positions, and a moderator whose job is to press, not to host. A moderator should be briefed on where the real disagreement lies and allowed, even expected, to return to the same question until it is answered.
Named outputs. The session produces something: a vote, a list of the three objections nobody could answer, a statement of where the speakers actually disagreed. It gets published. If sessions knew their disagreements would be written down, speakers would prepare differently.
Sponsorship separated from the argument. Sponsors can fund a track, a room or a format without choosing the speakers or the motion. This is how serious journals handle advertising, and it is not an impossible standard.
The strongest objection
The best counterargument is that conferences aren’t for decisions, and pretending otherwise misunderstands them. Their real value is network: the corridor conversation, the dinner, the meeting you arranged in advance. The sessions are the pretext that gets everyone into the building. Asking them to produce decisions is like asking a trade fair to write legislation.
There is also a fair point about safety. Health professionals from fragile political contexts may be unable to disagree publicly with their own government. Structured debate can put them at risk. And a conference that becomes known for confrontation may lose exactly the senior officials whose presence makes it valuable.
Both points hold partly. The network argument explains why people attend despite weak sessions, not why the sessions should stay weak. If the building is full of people with decision authority, it is wasteful to spend most of their time on content designed for nobody in particular. And the safety concern is an argument for careful design, including Chatham House rules for some sessions, anonymous audience voting and debates where speakers are assigned sides, not for no disagreement at all. Assigned-side debates in particular protect speakers, because arguing a position is not the same as holding it.
A worked example (hypothetical)
Imagine a regional health meeting with a session originally titled “Strengthening the Health Workforce for Universal Health Coverage.” Six speakers: a ministry official, a WHO representative, a nursing association, a medical school dean, a private hospital group and a donor. Forty-five minutes. Predictable outcome: agreement that the workforce is important, underpaid and poorly distributed.
Now redesign it around a decision. The motion: “Newly qualified doctors should be required to serve two years in rural districts before receiving a full licence.” Two speakers, one for and one against, each with twelve minutes. A one-page pre-read setting out what is known about compulsory rural service schemes, what they cost to administer and the main objections. An audience vote before and after. Fifteen minutes of questions, with the moderator allowed to hold each speaker on a point until it is answered. Afterwards, a half-page summary: the vote shift, the strongest argument on each side, and the question neither side could answer.
The second session has fewer logos. It is harder to sponsor. It is also the only one of the two that a ministry official could take back to a policy meeting.
What each actor can do
Organisers can require a one-sentence decision statement for every session, cap panels at three speakers by default and set aside a fixed share of the programme for formats that force trade-offs. Publish the outputs.
Sponsors can buy formats and rooms rather than speaking slots. The sponsor of a well-run debate gets more credibility than the sponsor of a sixth seat on a panel.
Speakers can decline panels of more than four people, ask the organiser what decision the session is supposed to inform and use their minutes to state a position someone could disagree with.
Moderators can prepare the disagreement, not just the introductions, and treat follow-up questions as the core of the job.
Delegates can choose sessions by title: go to the ones that could be answered yes or no, and say so in feedback forms, which organisers do read.
The takeaway
Most conference agendas fail not because the speakers are weak but because the design removes the one thing that would make the sessions useful: a real choice, argued in public, by people who disagree. Fixing it does not need a new kind of event. It needs organisers willing to write down, for each session, what someone should be able to decide afterwards, and to cut the sessions where that sentence cannot be written.
What is the best conference session you have attended that actually changed a decision, and what made it different? Tell us in the comments.
