What Project Oxygen taught us about 1:1s — and what it didn't.

Project Oxygen is the most-cited piece of internal research on management in the last two decades. It’s also routinely overread. Here’s what Google actually measured, what the study concluded, and — the part most summaries get wrong — what it didn’t.

What Project Oxygen actually did

Between 2008 and 2010, Google’s People Analytics team analysed performance reviews, feedback surveys, and productivity metrics across Google managers. The goal was to answer a specific question: did managers matter to team outcomes? The answer was yes, and the team identified eight behaviours that separated effective Google managers from ineffective ones. In 2018, two more were added, bringing the list to ten (Google re:Work · Manager research).

The most-cited finding comes from David Garvin’s 2013 HBR writeup: “The starting point was shocking: the least important of the eight projected attributes was technical expertise.” Ranked first: Is a good coach.

What “good coach” meant in the data

Google’s definition is in the re:Work guide: a good coach gives timely, specific feedback; balances positive and constructive; holds regular 1:1s; and tailors communication to the individual. This is where the inference about 1:1s enters the picture — Google’s own behavioural description includes “holds regular 1:1s” as a coaching behaviour.

What the study did notdo: compare 1:1 frequency against team outcomes, measure optimal 1:1 length, or quantify the incremental return of a weekly 1:1 versus a bi-weekly one. Those claims, when you see them, are extrapolations. The study supports the direction; it doesn’t specify the parameters.

The Aristotle companion finding

Project Aristotle (2012–2016) asked a different question: what makes teams effective? Google analysed 180+ active teams. The surprise was that composition mattered less than dynamics. From the re:Work guide: “Who is on a team matters less than how the team members interact, structure their work, and view their contributions.”

Of the five dynamics that predicted team effectiveness, psychological safety ranked first — people feeling safe to take risks and be vulnerable in front of each other. The other four: dependability, structure and clarity, meaning, impact.

Psychological safety isn’t a meeting format. But it’s the thing meetings either build or erode. A 1:1 run well builds it; a status meeting run poorly destroys it.

How to actually run a 1:1, synthesised from the research

Oxygen identifies coaching as the top behaviour; Aristotle identifies psychological safety as the top team dynamic; Ben Horowitz (The Hard Thing About Hard Things) provides the only clear operating guidance across all of them — in a 1:1, the manager talks 10% of the time and listens 90%. The employee owns the agenda. The manager’s job is to hear what’s not working, remove blockers, and coach.

Cadence: weekly, 30 minutes, with direct reports. A shared doc updated before the meeting does most of the work — the live time is for the things that need real-time back-and-forth.

The research supports the direction. The specifics are practitioner judgement. Don’t let anyone tell you Google has proven that a weekly 30-minute 1:1 is optimal — they haven’t. They’ve shown that coaching matters, and 1:1s are where coaching happens. That’s the honest version.