The short version
- A single directly responsible individual works well in one time zone and creates a bottleneck across three.
- The model assumes the DRI is reachable. Across seams they are asleep for a third of the questions addressed to them.
- Separate the two things a DRI provides: decisions, which must wait for them, and context, which should not.
- Only decisions should wait. Everything else the DRI knows should be answerable from their record.
The directly responsible individual is one of the better ideas in operational management. One name against one thing, no committees, no ambiguity about who decides. It also has a hidden dependency that only shows up when the team spans time zones.
"The DRI is in another time zone and the question can't wait."
Why the DRI model works
It works because it removes the most expensive form of delay: nobody knowing who decides. A named individual means a question has a destination, decisions have an owner, and accountability is legible. Compared with committee ownership it is faster, clearer, and produces better decisions, because someone bears the consequences.
None of that is in dispute. The problem is not the model, it is the unexamined assumption underneath it.
The assumption it makes
The model assumes the DRI is available when the question arises. In one office that is nearly always true: they are down the corridor, and if they are in a meeting it is an hour's delay.
Across three zones it is false for most of the day. A DRI in Amsterdam is unreachable for the entire working day of a colleague in Tokyo and most of one in San Francisco. So a model designed to eliminate delay introduces a structural one, and the team's throughput is capped by one person's waking hours.
Two bad adaptations follow. Teams either appoint a DRI per zone, which reintroduces the ambiguity the model existed to remove, or the DRI stays reachable outside their hours, which is the always-on pattern.
Two things a DRI provides
| What is asked of them | Share | Must it wait for them? |
|---|---|---|
| Context: what was decided, where it stands, what was tried | The majority | No |
| Decisions: a new call that has not been made | A minority | Yes |
Conflating these two is what turns a DRI into a bottleneck. The model's value is entirely in the second row, and the delay it causes is almost entirely in the first.
Who answers when they are off
Split the two explicitly, and the model survives distribution intact.
The DRI's StandIn answers from their record: each day they confirm a ninety-second brief covering current state, decisions made and why, open questions and next actions. While they are off, it answers questions from that in their words, with a source under every answer, clearly labelled, and it never guesses. When the question is a new decision rather than a retrieval, it says so and names the DRI, which is the correct outcome, because that is the thing the DRI is for.
So only decisions wait for them. Everything else resolves during the asker's own working day, and the DRI does not have to be reachable at 23:00 to keep the model honest. See working across time zones.
The deputy DRI, and its limits
A deputy in another zone helps with the decisions that genuinely cannot wait, and it works only with written limits. "The deputy decides anything reversible within the current sprint; anything else waits" is a usable rule. "The deputy covers when I am away" is not, because the deputy will not know which calls are theirs and will default to deferring.
Be honest about the cost, too. Every decision a deputy makes is a decision the DRI did not, which dilutes the single-owner clarity the model was chosen for. Keep the deputy's scope narrow, review what they decided, and do not use a deputy as a substitute for making context retrievable, because most of what reaches them is context rather than decisions.
Common Questions
Does the DRI model work across time zones?
Yes, if you separate decisions from context. The model's value is a single decision owner, and its cost is delay on everything else routed to them. Make the context answerable and only the decisions wait.
Should we have one DRI per time zone?
No, that reintroduces the ambiguity the model removes. Keep one DRI per area and give them a narrow deputy with written limits for decisions that genuinely cannot wait a night.
What if the DRI is a bottleneck even in one time zone?
Then they are being asked for context rather than decisions, or their decision scope is too broad. Sort a week of their incoming questions before adding people, because the answer is usually retrieval rather than capacity.
How much should a DRI be able to decide alone?
Everything reversible within their area, and nothing that commits the company externally. Write the boundary down, because an unwritten one is negotiated during a disagreement, which is the worst time.
When you're off, your StandIn is on.
It answers your teammates' questions from work you've already done, in your words, with a source under every answer.