EOS Rocks are the 3 to 7 most important things your company must accomplish in the next 90 days, drawn from the Entrepreneurial Operating System created by Gino Wickman and detailed in his book Traction. The concept comes from a simple story: if you fill a jar with big rocks first, then pebbles, then sand, everything fits. If you start with sand, the rocks never fit. Rocks are the big things. They go in first.
That is the whole concept. And yet most teams running EOS, or something like it, lose their Rocks within six weeks.
This post covers what EOS Rocks are, how to set them correctly, the difference between company Rocks and individual Rocks, and how to build a tracking rhythm that keeps them alive through a full 90-day quarter.
What EOS Rocks are (and the rock sand pebble story)
The rock sand pebble story comes from a classroom demonstration that Wickman adopted to explain quarterly priority-setting. A professor fills a glass jar with large rocks. He asks the class if the jar is full. They say yes. He pours in pebbles. They settle between the rocks. Full now? Maybe. He pours in sand. It fills the remaining space. Full? The class concedes yes. He pours in water. More space fills.
The lesson is not that you can always fit more in. The lesson is that sequence determines capacity. Put the big things in first, or they never go in.
In EOS, the Rocks are the big things. The pebbles are your weekly priorities. The sand and water are everything else, the daily tasks and small fires that expand to fill whatever time you leave them. If you run your quarter sand-first (reacting to what lands in front of you), the Rocks never get done. They stay on a list somewhere, labeled "strategic," and the quarter closes with no meaningful progress.
Setting Rocks forces you to decide, before the quarter starts, what the big things actually are.
Quarterly rocks: the 90-day structure
EOS runs on a 90-day world. Not a yearly plan broken into quarters, a fresh commitment cycle every 90 days. Each quarter, a leadership team asks: given where we want to be in one year (and in three, and in ten), what are the most important things we must complete in the next 90 days?
The answer is the quarterly rocks list. For most companies, that is 3 to 7 Rocks at the company level, plus 1 to 3 Rocks for each seat on the leadership team.
Ninety days is the right horizon for Rocks because it is short enough to feel real and long enough to do something meaningful. Twelve months is too long. People do not get serious about a twelve-month goal until month ten. Four weeks is too short. You cannot complete anything structural in a month. Ninety days is the window where a team can commit to something hard, break it into weekly steps, and finish it with time left to reflect.
Each Rock should be a specific, completable outcome. Not "improve sales process" but "document and train the full 8-step sales SOP and test it on five deals by September 30." If you cannot answer "did we complete this or not" at the end of 90 days, it is not a Rock. It is a goal masquerading as a Rock.
Company rocks vs individual rocks
EOS distinguishes between company Rocks and individual Rocks, though in practice they operate together.
Company Rocks are the highest-priority outcomes for the business as a whole. They belong to the company and typically to the leadership team as a group. Everyone on the leadership team is expected to contribute to company Rocks, even if one person is listed as the owner.
Individual Rocks belong to a specific seat on the accountability chart. They represent the most important outcomes that person must drive from their seat over the next quarter. A sales leader's Rock might be "book 20 qualified discovery calls." An operations leader's Rock might be "migrate all client onboarding to the new SOP." These are seat-specific.
The distinction matters because it changes who is accountable for reporting progress. Company Rocks surface in the weekly Level 10 Meeting (the L10, a term from EOS Worldwide) for the leadership group. Individual Rocks surface when that person's row comes up on the meeting scorecard.
At Sneeze It, our rocks live on the same chart where Bogdan's seat and Radar's seat both have rows. That includes a quarterly rock review section, which means neither humans nor agents get a pass on the question: is your rock on track or off track?
How to set rocks the right way
Setting Rocks poorly is common. Teams write them too vague, write too many, or write them top-down without building the context that makes them real.
Here is the process that works:
Start with the one-year goal. What must be true one year from today for this year to count as a success? Write that down in a few sentences. Now ask: given that, what are the three most important things we must complete in the next 90 days to stay on track? If you cannot answer that question, you do not yet have a clear one-year picture. Fix that first.
Write Rocks as outcomes, not activities. "Complete the sales hiring process and bring one new AE on payroll" is an outcome. "Work on sales hiring" is an activity. Activities do not have a done state. Rocks need a done state.
Limit the list. Most leadership teams try to set too many Rocks. Seven is the ceiling. Three to five is better. When everything is a priority, nothing is. The hardest part of setting Rocks is removing things from the list, not adding them.
Assign an owner to each Rock. One seat owns each Rock. Not a committee. One person. If a Rock does not have a single owner, it will not get done.
Check each Rock against the question: can we complete this in 90 days? If the answer is "probably not," break it down or defer it to next quarter.
For a deeper look at how this connects to the weekly meeting rhythm where Rocks get tracked, see our post on running a Level 10 Meeting. And for how AI agents plug into accountability structures like this, our post on putting agents on the org chart covers the underlying seat design.
90-day priorities: keeping rocks alive through the quarter
Setting Rocks is the easier half. Keeping them alive for 90 days is harder.
Most teams set Rocks in a quarterly planning session, update them for two weeks, then stop. By week eight, nobody can say with confidence which Rocks are on track. The quarter closes with a retrospective that is mostly guessing.
The discipline that prevents this is simple: every week, in the L10, each Rock owner states whether their Rock is on track or off track. That is all. On track or off track. No detailed status report. No explaining. If it is off track, it goes to the issues list and gets discussed. If it is on track, the team moves on.
This review takes less than five minutes when the list is short and the owners are prepared. It takes twenty minutes when nobody knows where their Rock stands.
The key is that "on track or off track" is a forcing function. It requires the Rock owner to have checked on their Rock since last week. If you cannot answer the question, you have not been working the Rock. The weekly check-in makes that visible before the quarter is lost.
Secondary tools help too: a simple tracker that shows each Rock, its owner, its target completion date, and its current status. Red, yellow, or green. Nothing more complicated than that. Teams that try to build elaborate project plans for Rocks usually spend more time maintaining the plan than moving the Rock.
At Sneeze It, we track Rocks in OTP, which means our agents, Tally and Dash, can read Rock status the same way a human teammate can. When the weekly meeting runs, Rocks show up on the same surface as KPI rows and issues. Nothing lives in a separate doc that nobody opens.
Frequently asked questions
What is the difference between a Rock and a goal in EOS?
A goal is directional. A Rock is a specific, completable 90-day priority with a clear owner and a binary done state. In EOS, goals belong to the one-year and three-year vision. Rocks are what you commit to doing this quarter to move toward those goals. If you cannot say "yes, we completed it" or "no, we did not" at the end of 90 days, it is a goal, not a Rock.
How many Rocks should a company have per quarter?
EOS recommends 3 to 7 company Rocks per quarter. Most experienced teams run 3 to 5. Fewer Rocks with full commitment beats a long list with scattered attention. Individual seats on the leadership team typically carry 1 to 3 personal Rocks on top of the company list.
Who owns a company Rock?
Each Rock has exactly one owner, a single seat on the accountability chart who is responsible for seeing the Rock completed. In EOS, ownership does not mean doing all the work. It means being accountable for the outcome and reporting status weekly.
What happens if a Rock is off track at the weekly meeting?
In the Level 10 Meeting format from EOS Worldwide, an off-track Rock gets added to the issues list. The team uses the Identify, Discuss, Solve (IDS) process to work through what is blocking the Rock and what must change. The Rock owner leaves with a clear action. If the Rock is consistently off track, the team may decide to rewrite it or push it to next quarter.
Can AI agents have Rocks?
Yes. If an agent holds a seat with real accountability, that seat can carry a quarterly Rock the same way a human seat does. At Sneeze It, agent seats participate in the weekly Rock review. The seat owner (a human) reports on the agent's Rock status. The discipline is identical to how we treat human Rocks.
Run your operating system in OTP
OTP is a team chart where humans and agents share scorecards, Rocks, and weekly meetings on one surface. Rocks belong to seats, not to a separate doc, which means the weekly review is built into the same rhythm as your KPIs and issues list.
In Claude Desktop or Cursor or any MCP client, add this block:
"otp": {
"command": "npx",
"args": ["-y", "@orgtp/mcp-server"]
}
Restart the client. Then ask: "Use OTP to show me the current Rock status for each seat on our chart."