Accuracy Has Cost

Planning is not free diligence

Geoff Charles, who runs product at Eric Glyman's company Ramp, states the axiom plainly: "any second you spend planning is a second you don't spend doing."¹ The corollary he draws is not "never plan" but a targeting rule. Because accuracy carries a cost paid in velocity, he argues a team should only raise the accuracy of its planning for the things where accuracy itself has high value. Everywhere else, the time that would go into making a plan precise is better spent shipping.

Charles draws the line at coordination. Accuracy pays at large market moments, the launches where product, marketing, and sales must move together, which in practice tracks the marketing calendar and recurs roughly once a quarter or once every six months. It does not pay at the feature level: "you don't need a lot of accuracy and confidence on when specifically certain features will be live." His pods are described as extremely clear internally while looking chaotic from the outside, which he treats as correct by design rather than a defect to plan away.

The 33 percent that got cut

The concrete evidence Charles gives is a measured failure. Ramp ran quarterly OKRs with financial goals and quasi-quotas, and the exercise of making metrics both right and achievable turned planning, in his words, very political and very annoying, consuming roughly one month in every three, a third of the company's time. The R&D org, he recounts, effectively revolted: "we're just going to execute on the roadmap, screw the OKRs." The replacement stack keeps the plan but shrinks its surface: a biannual one-pager of company priorities, a financial plan where every lever has a named owner, and the roadmap itself as the product organization's contract back to sales and marketing. OKRs survive only for genuinely cross-functional objectives such as winning a new market.

The boundary is strategy, not scope

The part Charles is careful to distinguish is that low-accuracy planning is not weak strategy. Ramp's goals and vision layer, he says, has been extremely consistent from the start, with the annual strategy note often landing close to what happens. What is allowed to churn is specifics, timing, and quarterly scope. The failure he names as the one to avoid is thrash, people "waking up and feeling like they're working at a different company." New joiners are told the trade explicitly, that they are signing an implicit contract of velocity over almost everything and that things will sometimes ship broken. In this framing the pattern is the planning-layer sibling of context over control: the expensive coordination is batched into a few planned moments so that pods can be left autonomous, and even chaotic, in between, which is also why the process versus bureaucracy question resolves toward less written process rather than more.

Where it may not transfer

Charles is candid that the approach leans on conditions Ramp happens to have. It works partly because Ramp's market supplies goal clarity through revenue and comparables without needing a heavy planning process to manufacture it; a company without obvious comparables may need more of the accuracy it is being told not to buy. The 33 percent figure is also self-reported and folds in the political cost of the old OKR cycle rather than pure calendar time. And the line between an objective that is cross-functional enough to deserve an OKR and one that is product-only and therefore roadmap-only is left to judgment, not codified, which is the seam where the discipline can quietly slip back toward the process it replaced.