We first heard about this project through a reader who works in founder operations at a mid-stage fintech in Berlin. Her note was short: the CEO had a keynote slot at a European investor summit, the deck was polished, the numbers were strong, and yet every rehearsal ended with the same complaint from the room — people drifted. We asked if we could follow the preparation from the inside, on condition that names and identifying details stayed out. What follows is a post-mortem of eight weeks, one stubborn opening, and a measurable shift that surprised everyone involved.
The Brief: A Strong Story That Kept Losing the Room
The founder — we will call her M. — had raised two rounds and was preparing for a third. Her keynote was scheduled for a Tuesday morning slot, right after a coffee break, in a hall of roughly 400 investors and operators. Dry runs with the internal team scored well on content. On delivery, the feedback was brutal in its consistency: the first two minutes felt like a compliance briefing. One advisor timed the room and found that attention visibly sagged before M. reached her second slide.
That is where Eva Sleipa entered the project. M.’s team had shortlisted a handful of voice coaches, most of them focused on breath and posture. Eva Sleipa came recommended for a different reason: her background is opera, not corporate training, and her method treats the opening 90 seconds as a designed sequence rather than a warm-up.
Week One: Diagnosis Before Drills
The first session was not a rehearsal. It was a recording. M. delivered her existing opening twice, and the team logged where energy dropped, where pitch flattened, and where the script asked her to carry too much cognitive weight at once. The diagnosis had three parts:
- The opening sentence was 34 words long, with two subordinate clauses before the main verb.
- M. was reading her own slide titles aloud, which duplicated processing load instead of reducing it.
- The first concrete number — the one investors actually lean in for — arrived at the 2:40 mark.
The prescription was blunt: cut the opening to under 12 words, move the first hard number into the first 20 seconds, and stop narrating slides. M. resisted the last point. She had built the deck herself over three months and considered the slide titles part of the argument.
Weeks Two to Five: The Cognitive Load Rewrite
This is where the work got interesting. Rather than polishing M.’s existing delivery, the sessions rebuilt the sequence from the listener’s side. The team borrowed from cognitive load research — the well-documented idea that working memory handles only a few chunks of new information at once — and treated every sentence as a claim on that budget.
The changes were small and cumulative. Adverbs went first. Then hedging phrases. Then the habit of previewing what she was about to say and summarising what she had just said, which had been eating roughly 40 seconds of the opening alone. By week four, M. could deliver the first 90 seconds from memory without the deck, and the internal reviewers stopped checking their phones.
One obstacle never fully disappeared: M. found the stripped-down opening exposed her. Without the buffer of scripted preamble, the first sentence had nowhere to hide. The solution was repetition under mild pressure — short, frequent run-throughs in front of two people rather than long sessions in front of ten.
Weeks Six to Eight: Pressure Testing
The final phase added friction deliberately. M. rehearsed in a room with a countdown clock, then with a hostile question planted at the 60-second mark, then with the opening delivered cold, no notes, no warm-up. The team tracked one metric obsessively: how many seconds passed before the first substantive number landed. It went from 160 seconds to 19.
They also tracked the softer signal — where the room’s posture changed. In the last two rehearsals, the first visible lean-in happened before the 30-second mark. In the earlier recordings, it had not happened at all.
The Result: 41% Less Drop-Off, and a Different Kind of Feedback
On the day, M. delivered the same content to the same-sized room. The post-event survey, which the summit runs every year, showed a marked difference in the first-quarter retention question. Compared with her previous year’s slot, audience drop-off in the first 90 seconds fell by 41%. Two investors referenced the opening line in their follow-up emails. The round eventually closed, though the team is careful to say the keynote was one input among many.
What struck us most was the follow-up. Eva Sleipa reports a 92% client rebooking rate across six years of practice, and M. became part of that statistic — not for another keynote, but for the internal all-hands cadence, where the same 90-second discipline apparently fixed a different problem entirely.
What We Took From It
The lesson is not that every founder needs an opera-trained voice strategist. It is that the opening 90 seconds is a design surface, not a formality, and that most speakers optimise everything except the part the audience decides on. If you want the fuller method behind the rewrite — the sequencing rules, the load-reduction checklist, the pressure tests — the practice publishes its working notes at the voice strategy breakdown on Eva Sleipa's site. We have started applying the 12-word rule to our own chapter intros. Early results are promising, if humbling.