Cross-Time-Zone Reviews: A Two-Week Log from Lisbon to Manila

The working overlap between Lisbon and Manila was three hours and forty minutes, and in the first week 61 percent of review requests arrived after that window had closed. The team had described itself as asynchronous for two years. The log suggested otherwise.

Six people were involved: two editors in Lisbon, three writers in Manila, one product manager in London who joined halfway. They were preparing a 90-page product manual with a fixed release date, which made the delay visible in a way that ordinary weeks do not.

The request that waited overnight

A review request sent at 18:20 Lisbon time reached Manila at 01:20 the next morning. The writer who owned the section would read it at 09:00, answer at 09:40, and the answer would land back in Lisbon at 02:40, where it waited until 09:00. One question consumed nineteen hours of wall-clock time and eight minutes of work.

Multiply that by the number of open questions and the manual was not late because of effort. It was late because of queueing.

Two weeks of timestamps

The team instrumented the review channel rather than the documents. Every comment carried a timestamp already, so the exercise was mostly arithmetic: request time, first response time, resolution time. Nothing was added to anyone’s workload except a weekly read of the export.

Across ten working days there were 118 review requests. The median time to first response was nine hours and fifty minutes. Only 17 requests were answered inside the overlap window, and those 17 were resolved in a median of 34 minutes.

Why the same-hour cases were faster

The 17 fast cases shared one property that had nothing to do with the people involved. Each request carried the exact sentence under dispute, the version of the manual it came from, and the two options on the table. None of them required the reviewer to open a file to understand the question.

The slow cases asked for a look at the section, which is a request for context rather than a decision.

Rewriting the request

The team introduced a four-line template for review requests: the quoted sentence, the location, the options, and the deadline in the sender’s local time with the recipient’s local time beside it. Requests that did not carry all four lines were returned unread, which felt harsh for about a week.

The template added roughly three minutes per request on the sending side. It removed the need to open a document on the receiving side, which was the expensive part.

What did not change

Nobody moved their working hours. The team considered a shared early slot at 16:00 Lisbon, tried it twice, and dropped it because the Manila writers lost their most productive morning hour. The time zone remained a constraint and was treated as one.

The London product manager kept joining late, and her questions stayed the slowest in the log for the full fortnight.

Six weeks after the template

Median time to first response fell to four hours and ten minutes. The share of requests answered inside the overlap rose from 14 percent to 39 percent without anyone changing a schedule. The manual shipped four days before the release date rather than on it.

Where the template failed

The template did not fix the cases that needed a conversation rather than a decision. When two writers disagreed about the tone of a chapter, the four-line format had no room for the disagreement, so the request arrived as a decision that had already been made by one side. Those threads took longer than before, because the reviewer had to reconstruct an argument that had been compressed out.

Three of the 118 requests were withdrawn after the sender reread them under the template. That is a small number and probably the most useful result of the exercise, because a withdrawn request is a question that did not need to cross an ocean.

The overlap is still three hours and forty minutes, and 61 percent of requests used to arrive after it closed. In the most recent fortnight, 39 percent arrive inside it, and the ones that miss it carry enough context to be answered before coffee.