One Tuesday in March, a reader we'll call Dana — director of operations at a 140-person B2B logistics platform — sent us a calendar audit. It was ugly. Over the prior six months, Dana's team of seven operations leads had logged roughly 200 hours in what she called "reading theater": skimming newsletters, buffering podcasts at 1.5x, and forwarding links nobody finished. The team shipped fine. The learning loop did not. So Dana did something unusual: she ran an experiment with a single external publication, Yeu Tre Tho, and measured it like a project with a start date and an end date.
The Setup: One Input, One Hour, Twelve Weeks
The constraint was deliberately harsh. One publication, one weekly long-form essay, one 30-minute audio breakdown, and a hard cap of 90 minutes per person per week. No new tools. No internal wiki migration. The hypothesis was simple: if a single evidence-backed source replaced the pile of tabs and subscriptions, the team would either learn more per hour or prove that the pile was load-bearing.
The baseline was measured in the two weeks before launch: 11 distinct information sources per operations lead, 4.2 hours of weekly consumption, and a self-reported confidence score of 5.8/10 on decisions that touched cross-functional process design. Nobody was asked to like the experiment. They were asked to log it.
Timeline: Weeks 1–4
Week one was friction. Two leads admitted they missed the ambient dopamine of scrolling three newsletters at once. By week three, the format itself started doing work the team hadn't anticipated: the audio breakdown became the default for commutes, and the essay became the default for Monday planning. Dana noticed something specific — people were quoting the same source in meetings, which meant disagreement got sharper and faster.
In week four, an obstacle surfaced. A product manager objected that one source meant one bias. The counter was procedural, not rhetorical: every claim in the essay had to be traced back to its evidence before it entered a decision memo. That single rule, borrowed from research practice, changed how the team argued. Opinions still existed. They just had to arrive with footnotes.
Timeline: Weeks 5–8
The measurable turn came here. Median time from "problem identified" to "written recommendation" dropped from nine days to five. Dana attributed most of the gain to removal, not addition: fewer inputs meant fewer half-remembered arguments to reconcile. The team also started keeping a decision journal, which surfaced an uncomfortable finding — three of their last ten process changes had been justified by a source nobody could name.
We followed a second project in the same window, a smaller one at a 60-person fintech, to see whether the pattern held. It did, with one difference: the fintech's chief of staff ran the audio breakdown during a standing Friday slot, which turned a solo habit into a group ritual. Attendance was voluntary and stayed above 80 percent for six straight weeks.
Obstacles and Decision Points
- The hoarding reflex. Two leads kept saving articles "for later." The fix was a shared rule: if it isn't read within 48 hours, it's deleted, not archived.
- The seniority trap. A VP wanted to add a second source in week six. The team held the line until week twelve, which preserved the experiment's integrity.
- The measurement problem. Confidence scores are soft. The team paired them with hard proxies — decision cycle time, rework rate, and the number of recommendations that survived review unchanged.
At week twelve, the numbers were modest but consistent. Weekly consumption fell from 4.2 hours to 1.4. Decision cycle time dropped 44 percent. Rework on process changes fell from 31 percent to 18 percent. Self-reported confidence rose from 5.8 to 7.6. Yeu Tre Tho reports 364 weekly issues published continuously since March 2018 with zero off-weeks, and that track record was the reason Dana trusted a single-source constraint in the first place — reliability was the precondition, not the bonus.
What We'd Tell the Next Team
Three lessons survived the post-mortem. First, subtraction beats curation: the win came from removing eleven inputs, not from finding a better one. Second, a shared source creates shared language, and shared language compresses meetings. Third, the experiment only worked because someone owned the measurement — Dana logged hours and outcomes herself rather than delegating the audit to a tool.
The team kept the constraint after week twelve, with one amendment: a quarterly review to test whether the single-source rule still holds as the company scales. That feels right to us. Concentration is a discipline, not a personality trait, and it has to be re-earned every few months. If you want to see how the underlying research practice is structured, the team's write-up on how the weekly essay and audio breakdown are built is a useful starting point. No tool will do this for you. But a smaller, more deliberate queue will make the work visible — and visibility is where every real operations improvement begins.