On 8 July I wrote:
"we had discussed these tiers before but has been lost. can you go through all of our conversations and see. i had said free basic and pro or something and convinced me to use basic, but a better word, i don’t recall now"
Read that back. I am asking a machine to remember a decision that the two of us made together, that I had agreed with at the time, and that neither of us can now find.
The pricing tiers for a product I was building. Discussed properly. Argued about, mildly. Settled. And then gone, because the conversation it lived in had ended and nothing outside that conversation recorded it.
This is the actual limitation of working this way, and it is not the one people write about. The model is clever enough. It is not persistent. Every session begins as a stranger who happens to be very good at their job.
For a while I tried to solve that with better handover documents, and they help, which is a later piece. But the thing that actually fixed it was realising I already had a system whose entire purpose is remembering what was decided.
The ticket tracker is not admin overhead. It is the memory.
Linear, in my case, though the argument holds for Jira or anything else with a durable record and an API. What matters is that it sits outside the conversation, it persists, and both of us can read and write to it.
Once you see it that way a few things follow.
The backlog is the state of the product. Not a to-do list. The current, authoritative answer to what this thing is and what it is meant to become. When I start a session with "get up to speed", the tracker is where it goes.
Projects and epics are how you keep it navigable. On 3 July I noticed Timelanes was using projects properly and the yoga app was not, and asked for the yoga app to be restructured to match. That is not tidiness. A flat list of two hundred tickets is technically a memory and practically a fog.
A decision that is not in the tracker did not happen. This is the hard one and I still fail at it. If it was agreed in a conversation and not written down, it will be lost, and you will re-litigate it in six weeks with less context than you have now.
The test is one question. If your current session ended right now and a completely new one started, how much of what you know would survive?
For most people the honest answer is very little, and they are carrying the whole product in their head without noticing. That is fine until you have a bad week, or a busy month, or an operation.
Next: the part of the ticket that matters most, which is not the acceptance criteria.