Rocks, issues, and to-dos. Turning numbers into action.

Three containers, three time horizons

The three are easy to keep straight, because the real distinction is just timeframe and size.

Rocks are quarterly goals — ideally SMART goals, though OKRs work just as well if you want more detail and milestones. They're the higher-level priorities that take about a quarter to complete, and they're set before the quarter starts. Issues — what EOS calls the IDS list, for Identify, Discuss, Solve — are the discussion topics a team works during the quarter, usually surfacing right in the weekly meeting. An issue can be a genuine problem to solve or simply information to share. To-dos are the small task items that come up in a meeting or while working an issue, and they take a week or less. If you're solving an issue and realize you need last year's data to decide, that's a to-do: someone owns it, and it's reviewed next week when the issue comes back around.

Because the buckets are defined by time horizon — a quarter, the current quarter's discussion, a single week — people rarely confuse them. A rock is set in advance and is strategic; an issue arises and gets discussed; a to-do falls out of that discussion and gets done by next week. Each kind of work has an obvious home.

Rocks: 3–5, strategic, tied to the year

We set rocks every quarter at our QBR, usually starting with a brainstorming session. Each member of the leadership team would pitch ideas for the quarter's rocks, and then the team decided together what to prioritize. We landed on three to five company-level rocks, and up to about five per leadership team member. Crucially, they weren't free-floating — every rock connected back to the company's annual goals, which we set in December before the calendar year began. The quarter's rocks are how you make ninety days of progress against the year's plan.

A couple of real ones: "Launch a new de-duplication feature for our CRM product," and, on the engineering side, "Create a performance load-testing and implementation plan." Concrete, strategic, finishable in a quarter. We reviewed every rock at the weekly L10 in a quick readout — on-track or off-track, sometimes a percent complete — so a rock quietly slipping got caught in week three, not at the end of the quarter when it's too late to recover.

The flow: how a red number becomes action

Here's the part that ties it to the scorecard. A red metric doesn't go straight into a rock. It follows a path.

Concretely: a metric goes red, so it lands on the issues list. The team discusses it in the meeting, and that discussion usually produces a task or two — pull the data, test the change, talk to the customer — each owned and due within the week. More often than not, the issue resolves over a couple of weeks of that. It only escalates into a rock if it doesn't resolve — if a number like Demos Completed stays red for many weeks and the team decides it deserves a dedicated, quarter-long focus. So the containers form a natural escalation ladder by persistence: most reds are handled as an issue plus a few to-dos; only the ones that won't die become a rock. That ordering keeps you from turning every bad week into a ninety-day project.

To-dos: small, owned, due next week

To-dos lived on a team task list with an owner and an expected completion date, and the standing expectation was that about 80% of them were done each week. That number matters more than it looks: a team that completes most of its to-dos every week is a team that trusts the list, and a team that completes a third of them has a list that's quietly become decoration. The ones that didn't get done either stayed on to be finished or got removed deliberately — never just silently ignored.

The discipline that keeps to-dos working is a size limit: a to-do has to be achievable in a week or two, full stop. The moment something will take longer, it isn't a to-do anymore — it's a project, which means it belongs in a rock, or it needs to be broken into smaller to-dos that each fit inside a week. Most rotting task lists are really just lists of projects mislabeled as tasks.

Where it breaks: vague rocks and the BAU trap

All the usual failure modes are real — an issues list that grows forever and never closes, to-dos that roll week after week, rocks nobody can define. Most of them come down to grooming: if whoever runs the meeting keeps the lists tended, manageable, and honest, they don't spiral. But the single biggest problem I saw was with the rocks themselves, and it's worth naming precisely.

Vague, non-SMART rocks are one failure; "BAU rocks" — business-as-usual work dressed up as a quarterly priority — are the more insidious one. Improving the FAQs, keeping the docs current, answering tickets faster: those are ongoing responsibilities, not strategic bets. When they fill your rock list, the quarter looks busy and moves the business nowhere. Rocks should be the handful of strategic things that, if you pull them off, the year goes differently.

The rule for each container

If you're setting this up, here's the short version of what I'd hold to for each.

Rocks: make them SMART (or OKRs, for the extra milestone detail), make them strategic and tied to the annual goals, and don't try to boil the ocean. Ten company rocks with seven to ten per leader is wildly too many; three to five at the company level and about the same per leader is the right load.

Issues: always know the actual problem you're trying to solve, and be aggressive about generating issues — pull them straight off the red numbers on the scorecard every week. And don't fear friction: there should be some real contention in working a good issue, or you're not getting to the bottom of it.

To-dos: keep the list manageable and make sure each one is genuinely achievable in a week or two. If it isn't, it's a project, not a task.

Underneath all three is one idea. The scorecard's only job is to surface what needs attention. Rocks, issues, and to-dos are what convert that attention into action with a name and a date on it. Skip the containers and a scorecard is just a prettier dashboard. Use them well, and the weekly number on the board reliably becomes a different number a month later — because something actually got done about it.