You agreed it on a call. Nobody wrote it down
TL;DR
A decision that is not written down is not a decision. It is a memory, and memories disagree.
Knowledge workers lose about 1.8 hours a day searching for information that already exists somewhere.
Asana found 209 hours a year go to duplicative work, the same task done twice because nobody knew the other had it.
The fix is one line per decision: what was decided, who decided it, and the date.
Write it in the same place every time, or it is as lost as the call was.
The client approved the design on a call. You heard it clearly. Your team heard it clearly. Two weeks later the client looks at the finished work and says, "I never agreed to that."
You know what was said. You cannot prove it. There is no message, no email, no note. There is only your memory against theirs, and both of you remember it differently.
This is not a rare disaster. It is a normal Tuesday. The decision was real. It just never became a record.
Why does a decision made on a call disappear?
Because a call leaves nothing behind. The words happen, and then they are gone. Nobody takes notes, because taking notes feels slow. Nobody writes a summary, because the call felt clear at the time. Everyone walks away sure they understood the same thing.
They did not. They understood their own version of it.
The client heard "yes, that direction works." You heard "yes, build exactly this." Your designer heard "yes, and keep it close to the old one." Three people, three decisions, one call. And the first time anyone finds out is when the work is already built.
Voice notes are worse. A two minute voice note feels like a record, because you can play it back. But nobody plays it back. It sits in a chat thread with forty other voice notes, and when the dispute arrives, nobody can find it, and nobody wants to listen to it twice to settle a point.
This is the same shape as the answer is somewhere in the chat. The information exists. It just cannot be found when it matters. A decision you cannot find is a decision you never made.
What does an unwritten decision actually cost?
It costs the work twice. First when you build it, then when you rebuild it. The team builds the version they remember. The client rejects it. The team builds the version the client remembers. Nobody is lazy. Nobody is lying. The decision was simply never written down, so every version of it is equally true.
The numbers behind this are old, and they still get quoted, because nothing since has made them smaller. The McKinsey Global Institute estimated in 2012 that knowledge workers spend nearly 20 percent of their week, about 1.8 hours a day or 9.3 hours a week, searching for and gathering information (source). A large part of that search is people looking for things someone already decided, somewhere, in a call nobody recorded.
Asana's Anatomy of Work Index, which surveyed more than 10,000 knowledge workers, found the average person spends 60 percent of their day on work about work, not the skilled job they were hired to do (source). The same index found 209 hours a year per worker go to duplicative work, the same task done twice because two people did not know the other had it. And 88 percent of workers agreed that time sensitive projects fall behind because of the volume of tasks on their plate. An unwritten decision is the purest form of duplicative work. You decide it once on a call, and then you decide it again in the argument, and then you decide it a third time in the rebuild.
Construction is the one industry that measures this properly, because its rework is physical and billable. PlanGrid and FMI surveyed nearly 600 construction professionals in 2018 and tied more than 31 billion dollars of annual US rework to poor communication and missing project information. Miscommunication drove 26 percent of all rework, and bad or inaccurate data drove another 14 to 22 percent (source). Your business does not pour concrete. But it rebuilds work for the same reason, and nobody sends you a bill for it. You just pay in weeks.
| What the research found | Number | Source |
| Time spent searching for information, per day | 1.8 hours | McKinsey Global Institute, 2012 |
| Share of the week spent searching and gathering info | About 20 percent | McKinsey Global Institute, 2012 |
| Time spent on work about work, not skilled work | 60 percent | Asana Anatomy of Work Index, 2019/2021 |
| Hours a year lost to duplicative work, per worker | 209 hours | Asana Anatomy of Work Index |
| Workers who say time sensitive work falls behind | 88 percent | Asana Anatomy of Work Index |
| Rework driven by miscommunication | 26 percent | PlanGrid and FMI, 2018 |
| Annual US rework tied to poor communication and missing info | Over $31 billion | PlanGrid and FMI, 2018 |
| Hours a week waiting for info or recreating knowledge that existed | 5.3 hours | Panopto and YouGov |
Read the table as a founder with a team of five. If each person loses even an hour a week to a decision nobody can find, that is five hours a week, every week, spent re-deciding things you already decided. That is the cost of a call that left no record.
Why do clients deny approvals they gave?
Most of the time, they are not lying. They genuinely remember it differently.
Memory is not a recording. It is a reconstruction, and it rebuilds itself around what the person wants to be true now. The client who approved a design two weeks ago is now looking at a finished page with fresh eyes, and the fresh eyes do not like it. So the memory shifts. "I said try it" becomes "I never approved it." Both versions feel honest to the person saying them.
That is why arguing about what was said never works. You cannot win a memory contest with a client. The moment you say "but you said," you have already lost, because the conversation is now about who remembers better, and nobody wins that conversation.
The only thing that settles it is a record written at the time, in words both sides saw. A short message after the call. "As agreed: we go with option B, delivery Friday, price as quoted." If the client disagrees, they disagree in the moment, when it costs nothing to fix. If they do not reply, the silence is the record. And when the dispute comes two weeks later, there is no contest. There is a message, with a date on it.
This is why the brief changed in chat hurts so much. The change was real, and the record of it was buried. The same applies to approvals. The decision was real, and the record of it never existed.
What does a decision record look like?
It is one line. Not a document. Not a meeting minute. One line, written the same way every time.
What was decided. Who decided it. The date.
"Client approved homepage design B. Decided by Priya for the client, 7 Nov. Delivery Friday."
That is the whole thing. Three facts, in one place, where anyone can find them. If the decision has a number in it, write the number. If it has a date in it, write the date. The price agreed on a call is the classic one. Two weeks later the client remembers a lower number, and you have no message with the real one in it. A one line message after the call ends that argument before it starts.
The place matters as much as the line. One place, for every decision, every time. A shared doc, a channel, a running note, it does not matter what it is. What matters is that nobody has to ask where the decision lives. The moment decisions live in five places, they live in none of them, and you are back to searching. That is how you end up as the reminder system of your own business, carrying every open decision in your head because it was never written anywhere else.
How do you start writing decisions down without slowing everything down?
You do not start with a system. You start with one habit, after every call.
Before you open the next thing, send one message. In the same thread the work lives in. Three lines at most. What was decided, who decided it, when it is due. Thirty seconds. If the call had no decision in it, say so in one line. "No decision. Next step is mine, by Thursday." That is also a record, and it stops the call from becoming a second call later, which is how a meeting ended with nothing decided in the first place.
The second habit is the harder one. When someone asks you to decide something in a call, and it is not written down yet, do not decide it. Say, "Send it to me in writing and I will decide by tomorrow." This feels slow. It is not slow. It is the difference between deciding once and deciding three times. A decision made in writing can be found, forwarded, and pointed to. A decision made on a call can only be remembered, and memory is the least reliable system in your business.
Within a month, two things change. First, the arguments stop. Not all of them, but the ones about what was agreed. Second, you stop being the only person who knows what was decided. Your team can find the decision without asking you, which means they stop asking you, which means every small decision stops waiting for you. The record does the remembering. You get to do the work.
The decision is not the hard part. Founders decide things all day. The hard part is making the decision survive the call it was made on. Write it down, in one place, in one line, before you move on. Thirty seconds now, or a rebuild later. That is the trade, every single time.
Frequently Asked Questions
Why do verbal approvals cause so many disputes?
Because memory is a reconstruction, not a recording. Two people can leave the same call honestly remembering different things, and the difference only shows up when the work is built. A written record made at the time, with the decision and the date, ends the argument before it starts.
How long should a decision record be?
One line. What was decided, who decided it, and the date. If the decision has a number or a deadline in it, include that too. Anything longer than three lines will not get written, and a record that does not get written is no record at all.
Where should decisions be recorded?
In one place, every time. A shared doc, a dedicated channel, or a running note all work. What matters is that everyone knows where to look, and nobody has to ask. Decisions spread across five places are as lost as decisions never written.
What if the client will not confirm decisions in writing?
Send the summary anyway, in the same thread the work lives in. "As agreed: option B, delivery Friday." If they disagree, they will say so now, when it is cheap to fix. If they do not reply, the message stands as the record, with a date on it.
Is this not just more admin work for a small team?
It is thirty seconds after a call, against hours of rebuilding work that was built on a misunderstood decision. Asana found 209 hours a year per worker go to duplicative work. One line per decision is the cheapest way to get some of those hours back.