'Yeah, got it' is not evidence of anything
Say this sentence out loud, slowly, and actually listen to what it commits to: "Yeah, got it." Notice that it commits to almost nothing. It does not repeat back what was understood. It does not name what "it" refers to. It is, grammatically and functionally, the shortest possible sound two people can make to agree that a conversation is over. And a standup is absolutely stuffed with it — "sounds good," "makes sense," "cool, got it," "yep" — treated, every time, as proof that understanding occurred, when it is actually proof of something much smaller and much cheaper.
What conversation is actually trying to do
Herbert Clark and Susan Brennan, in their influential chapter "Grounding in Communication," laid out a theory that reframes what's happening every time two people talk: conversation isn't a pipe that transmits meaning from one head to another. It's a joint activity, where the two parties collaboratively build up "common ground" — a shared, mutually-confirmed understanding — through an ongoing sequence of contributions and acknowledgments. Crucially, Clark and Brennan argue that grounding requires evidence, not just an acknowledgment token. A nod, an "mm-hm," a "got it" — these are the cheapest possible form of evidence, signaling roughly "I am still here and did not object," and they are frequently treated as if they were the strongest form, when they are in fact the weakest one the theory describes.
Clark had made a closely related point five years earlier with Deanna Wilkes-Gibbs, in "Referring as a Collaborative Process": when two people establish what a phrase refers to in conversation, they don't aim for perfect, complete mutual understanding. They aim for what the theory calls "sufficient for current purposes" — good enough to keep the conversation moving, not necessarily good enough to survive being checked later. That phrase, sufficient for current purposes, should make you slightly uneasy the next time you hear "got it" in a standup, because the current purpose of a standup — ending the update so the next person can talk — has almost nothing to do with the purpose that will actually matter three weeks from now, which is whether the listener understood the claim well enough to notice if it turned out to be wrong.
Clark and Wilkes-Gibbs demonstrated this with a simple, elegant task: pairs of participants had to describe abstract shapes to each other well enough to sort them into a matching order, over repeated rounds. What they found was that descriptions got shorter and more compressed each round — "the shape that looks like a person ice skating" became, three rounds later, just "the ice skater" — and both parties treated the shrinking description as evidence of shared understanding. It was, for the narrow purpose of that task. But notice what that process actually optimizes for: efficiency of reference between two specific people, in one specific context, not portability of meaning to anyone else, or durability of meaning after the context fades. "The ice skater" means nothing to a third person who joins the task midway. "Fixed the auth thing" means nothing to anyone who wasn't in the room when the phrase first got compressed down from something more specific.
The grounding ladder, and where "got it" actually sits
Clark and Brennan's framework implies something like a ladder of grounding evidence, cheapest at the bottom, strongest at the top. It's worth laying out explicitly, because standups almost never climb past the second rung:
- Continued attention. The listener keeps looking at you, doesn't interrupt. Weakest possible evidence — indistinguishable from politely waiting for you to finish.
- Acknowledgment tokens. "Got it," "mm-hm," "cool," "yep." Slightly stronger — confirms a turn was received — but confirms nothing about its content.
- Paraphrase. "So you're saying the retry logic is the actual cause, not the queue?" Real evidence — the listener has reconstructed the claim in their own words, which at least demonstrates they built something, even if (per two parts ago) the paraphrase itself may drift.
- Display through action. The listener does something that only makes sense if they understood correctly — asks a specific follow-up question, flags a related risk, references the exact function later. The strongest evidence, and the rarest, because it takes time a fifteen-minute meeting rarely has.
Standups overwhelmingly operate on rung one and two. "Got it" is rung two doing the job people mistake for rung three or four. It is real — Clark and Brennan wouldn't call it meaningless — but it is real evidence of a very small thing: that the listener registered a turn happened. It is not evidence they built the same mental model of the claim that the speaker was holding.
Grounding theory doesn't say "got it" is worthless. It says "got it" is the cheapest rung on a ladder most standups never climb any higher than — and cheap evidence, mistaken for strong evidence, is exactly how two people leave a room both confident and wrong about what they just agreed on.
Why the ladder collapses to rung two, every time
It's worth asking why standups don't naturally climb higher, because the answer isn't laziness — it's the same economics Clark and Brennan built directly into the theory. Grounding, in their model, isn't free; both parties are implicitly weighing the cost of establishing stronger evidence against the cost of just moving on, and a standup's entire format is optimized to make moving on the cheap option. Asking a clarifying question, in a room where eight people are waiting their turn, has a visible social cost — it takes airtime that isn't yours, it can read as slowing the meeting down, it occasionally reads as not having been paying attention in the first place, which is a strange but real disincentive. "Got it" has none of those costs. It ends your turn as the listener as cleanly as the speaker just ended theirs. Two low-cost moves, chained together, and the meeting keeps its fifteen-minute promise — at the price of every claim inside it being grounded only as far as rung two ever gets you.
None of this is an argument that standups are badly run. It's the opposite: they're economically rational, tuned correctly for their actual purpose, which was never "produce verified mutual understanding of every claim" — it was "keep everyone roughly oriented in under fifteen minutes." Clark and Brennan's own framework predicts exactly this outcome for exactly this kind of low-stakes, high-frequency exchange. The mistake isn't holding a cheap grounding conversation. The mistake is later treating its cheap output as if it had been expensive.
A small, uncomfortable thought experiment
Picture a standup where an engineer says: "I'm going to refactor the pricing calculator — should be safe, nothing else calls it directly." Three teammates say "got it," "sounds good," "cool" in quick succession. The meeting moves on. Two weeks later, the refactor ships, and a downstream billing job — which nobody in the room was thinking about, because nobody actually checked, they just acknowledged — breaks in a way that traces straight back to that "should be safe" claim.
Nobody lied. Nobody was inattentive in any way that would show up on a performance review. Three people gave entirely genuine, socially appropriate rung-two acknowledgments to a claim that needed rung-four scrutiny — someone actually checking the call graph — and got rung two instead, because rung two is what a fifteen-minute meeting has time and appetite for. The "got it"s weren't false. They were honest evidence of the wrong thing.
What actually closes the gap
Clark and Brennan's own answer is not "make people paraphrase everything," which would turn every standup into an hour-long ritual nobody could sustain — grounding theory is explicit that speakers and listeners calibrate the amount of grounding effort to what the situation actually needs, and a standup, by its nature, needs to be cheap. The honest fix isn't forcing expensive grounding into a format built for fast turns. It's not pretending the cheap grounding a standup can afford is stronger than it is. "Got it" should never be treated, anywhere downstream, as confirmation that a claim was understood, agreed with, or reviewed — because Clark and Brennan's own framework says it was never designed to carry that weight. The claim itself — the actual sentence, with its subject — is what eventually gets checked against reality. The "got it" was just three people agreeing the meeting could move on. That's all it ever promised to be.