Skip to main content
one meeting is noise. the point of training mode is the trend across weeks. this page is a practical workflow for measuring whether you’re actually getting better.

​
pick one metric

don’t try to fix everything at once. pick one metric for the next 2-4 weeks. good candidates:
  • fillers per minute - easiest to move, visible to you and listeners
  • talk ratio - best if you’ve been told “you talk too much” or “you’re too quiet”
  • longest monologue - best if calls feel one-sided
  • pace - best if people ask you to slow down or speak up
if you’re unsure, start with fillers. it’s the most measurable and the hardest to fake.

​
baseline before you change anything

  1. record 5-10 meetings without trying to improve
  2. note the metric for each in a list
  3. compare similar meeting types only - don’t compare a solo demo to a 5-person standup
  4. take the median, not the average. one noisy meeting shouldn’t set your baseline

​
how to structure a coaching cycle

​
week 1 - baseline

  • don’t change behaviour. record normally
  • at the end of the week, pick a target: “fillers/minute from 6 to 3” or “longest monologue under 250 words”
  • write the target down (notes panel on any meeting works)

​
weeks 2-3 - deliberate practice

  • one behaviour change per week. see specifics in filler words and delivery metrics
  • check the metric after each relevant meeting
  • expect regression - the metric often gets worse before it gets better because you’re self-conscious

​
week 4 - evaluate

  • compare your median from week 4 to your week 1 baseline
  • if the metric moved, lock in the habit and pick the next one
  • if it didn’t, check you were actually applying the behaviour change. look at the longest segments from bad meetings and ask why

​
comparing like meetings

training metrics are context-dependent. don’t aggregate discovery calls with internal standups. a useful mental filter:
  • meeting type: discovery, demo, interview, 1:1, standup, presentation
  • who’s in it: customer, colleague, candidate, manager
  • your role: driver, listener, presenter, interviewer
compare like for like. if you’ve done 3 discovery calls this week and the average fillers/minute is trending down, that’s a real signal. one discovery call vs one standup tells you nothing.

​
using markdown export for external tracking

if you want to plot the metric over time in a tool you control:
  • enable auto markdown export on macos: settings → integrations → auto markdown export
  • each saved meeting lands as a markdown file with all training metrics in the front matter
  • use webhooks to post every saved meeting into a spreadsheet, notion, or linear
example: zapier scenario that extracts meeting.training.speakers[0].fillersPerMinute from the webhook payload and appends to a google sheet. see webhook recipes.

​
signals you’re actually improving

not just “the number went down”:
  • fewer edits when you play the meeting back. you used to want to re-record; now you don’t
  • people stop commenting on the thing you were working on
  • the metric holds under pressure - a stressful call doesn’t regress you to your old baseline
  • you notice in real time. you catch yourself about to say “um” and pause instead
the metric is a mirror, not the goal. when the behaviour becomes automatic, you can move on.

​
signals you’re gaming the metric

if your number improves but your meetings feel worse:
  • fillers down, pace down, clarity number up, silence up = you’re overthinking and sounding uncertain. the metric is fine; the experience isn’t. loosen up
  • talk ratio down but you stopped adding value = you swung too far. the goal wasn’t “talk less”, it was “talk less when they need to talk”
  • questions up but they’re all leading = more ?s don’t help if you’re not actually listening
the metric is a cue, not a constraint. if it conflicts with the experience, trust the experience.

​
monthly review ritual

once a month (10 minutes):
  1. open meeting history, filter by meeting type
  2. scan the training pane on each
  3. note patterns: did you regress? improve? stay flat?
  4. pick next month’s target
if you use markdown export, grep for the metric across your notes folder and sort by date.

​
troubleshooting

  • metrics look empty - the meeting had no transcript content, or all segments were marked interim. see troubleshooting
  • numbers feel wrong - check you’re in the right language. an english filler list won’t catch spanish fillers
  • a bad meeting looks good - the metric measures what was said, not how it landed. use it alongside self-reflection, not instead of it

​
see also