What Carries Over Between Talks
🗒️ Description
An audit of how I actually prepare talks, done across a series of them rather than for one. Conference Talk Craft is the researched playbook — what the field knows. This note is the other half: what my own record shows I do, and the one defect that showed up when the talks were put side by side.
It is uncomfortable in a useful way, so it is written down rather than absorbed.
☘️ The finding — a lesson without an artefact is re-solved, not reused
The chain, stripped to its shape:
- Talk one. The opening drifted despite having a plan. A predictable hard question from the room got an answer that came out acceptable rather than good.
- Retrospective. Written up honestly a few days later, with named action items: write a reusable opening template, write a standing answer to that question.
- Those action items became tracked tasks a few months later — and then sat.
- Talk two. The task file itself named this talk as the occasion to use the new template. It was not used. The preparation notes for that talk are almost entirely toolchain — slide CSS, image alpha channels, QR error correction, demo-environment reset. Nothing about how the thing would be delivered.
- Talk three. The opening is written out word for word, in four beats, with a two-second pause marked. The hard question has a prepared five-point answer. Both are genuinely good. Both live inside that one talk’s notes. The original action items are still open.
So the lesson was solved twice from scratch and never banked. Talk four would have paid the same cost a third time.
The mechanism is not laziness. The retrospective was written, the tasks were filed, the fix was eventually built. What never happened is the step between: moving the fix out of the folder belonging to one event and into something a future talk would find. A per-talk notes file is a terrible home for a recurring lesson — nobody opens last event’s folder while preparing the next one.
The second detail worth noticing: the trigger for finally fixing it was an outside research pass, not the task system. Tasks with a start date came and went untouched. Anything that depends on me noticing an old task on my own has already failed once.
🧩 Where preparation effort actually goes
Counting lines across preparation notes was more honest than my impression of them. The deck and the toolchain take almost everything; delivery craft appears only when something external forces it onto the page.
That is not irrational — it is a legibility problem. A slide that renders wrong is visible, reproducible and finishable. “The opening is shaky” is none of those. So the tractable work crowds out the work that decides how the talk actually lands, and it does it while feeling productive. Conference Talk Craft makes the same point from the other side: the structural material can quietly displace the thing that moves a room.
🧩 What good preparation looked like when it did happen
Worth naming, because these are the patterns to keep rather than rediscover:
- Cumulative time checkpoints plus a named cut rule — not “watch the clock”, but “if the demo has not started by this exact time, these two slides collapse to one sentence”. Cuts land at block boundaries instead of as panic-rushing. This is expansion joints, done properly.
- Refusing to state a number the room can contradict. Checking a claim about the day’s other sessions, finding my own earlier figure was guessed, and deriving a standing rule from it: say no count from the stage at all. Someone sat through everything and will do the arithmetic.
- Recorded demo running, live environment in reserve on a second desktop, entering only if someone asks. Keeps the clock and keeps the credibility — and it costs nothing extra, because the demo environment has to exist to make the recording anyway.
- Capping the self-introduction at twenty seconds, with the reason stated: talking about yourself has no natural end, so it is the first place a talk slips.
- “Know every record in the demo by heart” written down as a precondition, not a hope.
The quality of preparation was never the problem. Portability was.
🔧 What I changed
Two standing artefacts now exist outside any single event’s folder: a reusable opening template with the hard-won rules attached, and a standing answer to the question that returns at every talk and every sales call. Each is written to be filled in, not read and admired.
The third piece is still missing: a post-talk retrospective with an actual rubric, run while the talk is fresh, whose output is an artefact rather than a list of intentions. That is an outstanding research item, and the honest test of it will be whether talk four starts from something instead of from nothing.
📖 Further reading
- Conference Talk Craft — the researched playbook, with the evidence graded
- Public Speaking — hub note for the whole cluster
- The Art of Public Speaking — Carnegie on the drills; the emphasis exercise is the portable one
- TED Talks — Anderson on scripting being a per-speaker question, not a rule
- Atomic habits — the same “make it obvious” failure, applied to systems rather than habits
Template: knowledge_note_info