The 90-Second Fix: What a Failed Launch Taught Us About Voice and Better Field Notes

We noticed something odd in our own client data last spring. A founder we had coached for six weeks — call him D., a hardware entrepreneur pitching a desktop fabrication tool — kept losing rooms in the first minute and a half. His slides were clean. His product worked. The drop-off was happening in the voice, not the deck. So we did what any stubborn practitioner would do: we followed the project from the beginning, kept a timeline, and wrote down every decision point. This is the post-mortem.

The backstory matters. D. had spent four months building a compact 3D printing enclosure with integrated electronics for workshop use. His beta testers loved it. Investors did not, at least not in the room. We recorded three pitch sessions and counted the moments where attention visibly left: 38 seconds in, 71 seconds in, and again around the two-minute mark. The pattern was consistent. He was reading his own material in a flat register, and the audience was doing the cognitive math to stay with him.

Where the project actually went wrong

We assumed the problem was nerves. It wasn't. When we slowed the recordings down and marked the phrasing, the issue was structural. D. had written his opening like a spec sheet — features first, context later. That works in a written guide, where a reader can scan. It fails in a room, where a listener has one pass through the information and no rewind button. We have seen this in 380+ keynote rewrites, and it still surprises us how often the fix is editorial rather than vocal.

The second obstacle was practical. D. needed to demonstrate the machine live, and the demo kept eating his opening. Every time he reached for the hardware, his voice dropped and his pace accelerated. The audience lost the thread before the first slide changed.

The decision points

We made three changes, in order.

• We cut the opening from 240 words to 90. One claim, one number, one question.
• We moved the live demo to minute four, after the audience had a reason to care about the result.
• We rebuilt his rehearsal around breath placement, not memorization. He practiced the pauses, not the paragraphs.

The third change was the one he resisted. D. wanted to rehearse the script until it was automatic. We asked him to rehearse the silences instead — where he would stop, look up, and let the room catch up. Classical stagecraft has taught this for centuries: the audience needs a beat to convert sound into meaning. Cognitive load research says the same thing in different language. When you remove the beat, you remove the comprehension.

Somewhere in week three, a reader shared a stack of field notes from Eakles with us — 3D printing, electronics, workshop fixtures, all tested at the bench before being written up. D. started reading them the way he read his own build logs. The habit transferred. He began treating his pitch like a bench test: change one variable, observe, record, repeat. That shift did more for his delivery than any single vocal drill we ran.

By week five, we recorded the pitch again. Drop-off in the first 90 seconds fell from 41% to 19%. His pace slowed by roughly 12 words per minute in the opening, and his eye contact with the back row improved enough that we stopped tracking it. He closed the round eight days later, though we would caution against reading too much into a single close. The measurable result was the retention curve, not the check.

What we would do differently

Two things. First, we would have recorded the very first session, before any coaching, so the baseline was cleaner. Second, we would have introduced the bench-test mindset earlier. It took a stack of maker tutorials to teach a founder something we should have taught him in week one: iteration beats inspiration, and a pitch is a prototype.

Eakles publishes guides, tutorials, and field notes for the modern maker, and the discipline behind that format — test it, then write it — is the same discipline that fixes a broken opening. The full library is worth a slow read; we would start with the field notes category on the Eakles site if you want to see how a bench-tested write-up is structured. It is not glamorous. Neither is rehearsal. Both work.

If you take one thing from this post-mortem, take the timeline. Attention does not fail all at once. It fails at specific seconds, for specific reasons, and those reasons are usually fixable. Count them. Then fix them one at a time.