The scarcest resource is the decision-maker's attention

Over several weeks in a transformation programme I work in, a pattern showed up in decisions that on the surface have nothing to do with each other: a set of campaign numbers deliberately not shared with the wider group, a standing meeting renamed from alignment to update, an external approach routed back through the team instead of into an executive calendar, and a notification setting trimmed to exactly two events. None of these were efficiency measures. They were all the same measure, applied to the one input the programme cannot buy more of.

Four decisions with one shape

Notifications first, because it is the clearest case. The executive layer set its project notifications to two events: a task assigned to them, and a decision required from them. Status changes without a required action are off. That is not a preference about tooling. It is a statement that a status change consumed attention and returned nothing.

The numbers came next. A set of campaign results was deliberately kept on the operating layer instead of going into the wide group chat, with the option for a single person to share it bilaterally if they judged it useful. The reasoning was not secrecy. It was that a number sent to fifteen people generates fifteen interpretations and some number of replies, and none of that was needed to run the campaign.

The third was a word. Recurring meetings with a wider circle were renamed from alignment to update. One word, and it changes what the calendar entry promises: not a decision forum, a report. The decisions stay with the two people accountable for them.

The fourth was routing. An external provider reached a senior calendar directly by email and got as far as a proposed introductory call. The team cancelled the call — not because the topic was worthless, but because it was already owned internally, and because an unfiltered external approach bypasses the very system being built. The rule they wrote from it: external impulses run through the team, not into the calendar.

Why this is architecture, not comfort

The reflex reading is that senior people want to be left alone. That is not what is happening. In a programme of this kind, the throughput limit is not the number of automations, not the tooling budget, and usually not the delivery capacity. It is how many decisions the two accountable people can take per week with enough context to take them well.

Once you accept that as the binding constraint, everything that spends it becomes a design object. A notification without a required action is a withdrawal. A meeting that promises participation but only delivers information is a standing order against the same account. An unrouted external conversation is an overdraft nobody authorised.

The uncomfortable consequence for automation

Most automation work is sold and measured in the opposite direction. More visibility, more dashboards, more alerts, more status. The intuitive success metric is how much information the system now produces.

That metric is wrong for the layer above a certain level of accountability. The useful metric is how few things had to reach a person for the work to proceed correctly. A pipeline that surfaces every state change to an executive has not automated the work; it has moved the filtering step from the machine to the most expensive human in the process.

What to check on Monday

Take the automations you run that end in a message to a person, and mark every one with the action it requires from that specific recipient. Then delete, downgrade to a weekly digest, or route elsewhere everything with an empty mark.

Expect the count to surprise you. In most stacks that grew over a year, the majority of outbound messages exist because somebody once said they wanted to be kept in the loop — which is a sentence about trust, not a specification. The right answer to it is usually not more messages. It is one report that arrives when something actually needs deciding.