The pattern is familiar to anyone who has administered a corporate travel policy since 2020.

The policy goes in. Routine travel drops sharply, and the first year of reporting looks excellent. Internal reviews, supplier check-ins, team offsites and the majority of short-haul flying move onto video and mostly stay there.

Then the curve flattens. And the flights that come back are not the routine ones. They are the negotiations, the board meetings, the annual planning sessions, the first meeting with a new partner, the difficult conversation with an underperforming regional team.

In other words: the meetings that actually determine what the business does.

Guidance on the category is clear that the tools available work. Combined interventions across virtual conferencing, rail substitution, cabin-class policy and internal carbon pricing are credited with cutting business travel emissions by somewhere between 50 and 90 per cent where they are applied seriously. The question worth asking is why so many programmes stall well short of that, always at the same point.

The stated reason and the real one

Ask the executives who insisted on flying and you get relationship-building. Which is true, partly, and unfalsifiable, entirely.

The more useful answer surfaces when the question is asked differently: what went wrong the last time an important decision was taken on a video call?

The answers cluster. People agreed to something and then acted on different understandings of it. A quiet participant turned out to have serious reservations that never made it into the room. The decision was reopened three weeks later because nobody could establish what had actually been settled. Somebody’s contribution was lost and they concluded, reasonably, that attending remotely means being outranked by whoever is physically present.

None of that is a complaint about video quality. It is a complaint about what survived the meeting.

Travel persists on high-stakes meetings not because presence transmits information better, though it does somewhat, but because presence is how organisations currently compensate for meetings that do not produce a reliable record. Flying six people to Frankfurt is an expensive way to make sure everyone leaves with the same understanding, and it is the method most firms still trust.

What actually degrades

Three specific losses, worth separating because they have different fixes.

Turn-taking collapses. Remote conversation has worse interruption dynamics, and the effect falls hardest on participants who are junior, working in a second language, or on a poor connection. Their input drops out and nobody notices, because absence of contribution looks identical to agreement.

Attention is unverifiable. In a room you can see who has stopped listening. On a call you cannot, so decisions get taken with an unknown proportion of the participants having disengaged.

The record is thinner. This is the underrated one. Physical meetings generate corridor conversations, whiteboard photographs, and the informal alignment that happens over lunch. Remote meetings generate a calendar entry and whatever somebody typed while also talking.

The first two are behavioural and respond to facilitation discipline. The third is a tooling problem, and it is the one that has changed most in the last two years.

The documentation lever

A meeting that produces a complete, searchable, verifiable record is a materially different object from one that produces somebody’s notes.

Transcription tooling has become good enough that this is now a policy choice rather than a technical constraint. Products in this category, the Vomo transcription app among them, record the session and return the full text with speakers separated, alongside a structured summary covering decisions, action items and open questions. Coverage extends to roughly 50 languages, which matters for any organisation whose meetings are not conducted in the first language of most participants.

Three consequences follow, and only the third bears on travel.

Participants who could not follow a fast exchange in a second language can read it afterwards at their own pace, and raise a correction while it still costs nothing to raise. Disputes about what was agreed become searchable questions rather than contests of recollection. And the meeting stops depending on the memory and note-taking capacity of whoever was least busy talking.

The travel-relevant claim is narrow and worth stating precisely: a remote meeting with a reliable record is more decision-durable than a remote meeting without one, and decision durability is the specific thing executives are buying when they book the flight.

That is not an emissions intervention. It is a removal of one common objection to an emissions intervention that the organisation has already decided to pursue.

What this is not

Three things this article is deliberately not claiming, because the category attracts overstatement.

Transcription does not reduce emissions. There is no causal chain from a meeting record to a tonne of CO2e that survives scrutiny. The chain runs through a human decision to hold a meeting remotely, and that decision has many inputs. Anyone selling meeting software on a carbon reduction basis should be treated with suspicion.

None of this belongs in carbon accounting. Scope 3 Category 6 is calculated from travel that occurred. Travel that did not occur because a policy worked shows up as a lower figure, not as a credit, and no reduction may be claimed against tooling. Firms should resist any vendor proposing otherwise, and several are.

AI transcription has its own footprint. Inference consumes energy and the models were expensive to train. The per-meeting quantity is small relative to a short-haul flight, but small is not zero, and a sustainability function that adopts AI tooling without acknowledging that has weakened its own credibility on everything else.

The honest framing is unglamorous. This is an operational improvement that happens to remove friction from a travel policy. It is worth adopting on its operational merits, and the travel effect is a secondary benefit that should be described as such rather than quantified.

A policy change that fits on one page

For organisations whose virtual-first policy has plateaued, the sequence that tends to work looks like this.

Identify the meeting categories where travel persists. Most firms find between four and six, and they are almost always high-stakes, multi-party and cross-border.

For those categories specifically, mandate the record rather than the presence. A named facilitator, an agenda circulated in advance, a full transcript and a structured summary distributed within 24 hours, and a defined window for participants to correct it before the decision is treated as final.

Then apply the travel test in the other direction. Rather than asking whether this meeting justifies a flight, ask which failure mode the flight is insuring against, and whether the record now covers it. Sometimes it will not, and the flight is correct. The point is that the question gets answered explicitly rather than by seniority.

Set the retention and consent policy before the first recording, not after. Recording employees carries obligations that vary considerably by jurisdiction, and in several European countries requires consultation rather than notification.

Why the objection is usually about status

There is a version of this problem that no amount of tooling addresses, and sustainability functions should be prepared for it because it arrives in every programme.

Attending in person is a status signal. Being in the room where a decision is taken carries weight that dialling in does not, and everybody involved understands this even where nobody says it. When a policy restricts travel, the people who lose most are those whose influence depended partly on physical presence, and their resistance is genuine even when the stated reason is not.

This produces a predictable pattern. Travel policies hold firmly for junior staff and erode at senior levels, which is precisely the inverse of where the emissions sit, since senior travel skews long-haul and premium cabin. A programme that reports strong compliance overall while its executive population continues flying has not achieved much, and the reporting will eventually make that visible.

The workable responses are structural rather than exhortative. Some organisations apply an all-remote rule to specific meeting types, on the reasoning that if nobody is in the room then nobody is disadvantaged by not being in it. Others attach internal carbon pricing to the business unit rather than to the traveller, which moves the decision to the person who owns the budget.

What does not work is a policy that permits exceptions on seniority. That is not a travel policy. It is a travel policy for people whose travel was not the problem.

The measurement problem

One difficulty worth naming, since sustainability functions will encounter it immediately.

The benefit here is a counterfactual. Nobody can demonstrate that a specific flight did not happen because a specific meeting produced a good record, and any attempt to quantify it will produce a number that does not survive assurance.

What can be measured is proximate and still useful: travel volume by meeting category over time, the rate at which decisions from remote meetings are reopened, and participation distribution across locations. Firms that measure and actively manage the category report reductions averaging around a quarter within two years, which is a policy outcome rather than a tooling outcome.

Attribute the reduction to the policy. Treat the tooling as one of the things that made the policy survivable, which is a smaller claim, and the only one that is true.