Public Speaking for Founders (90-Day Skill Build)

I gave my first conference talk in October 2019 to a room of about sixty SaaS founders in a hotel basement in Austin. The talk was bad. I know it was bad because the recording exists, and I have watched it twice since. The first time was three weeks after the event, and I cringed through the entire forty-five minutes. The second time was last year, six and a half years and seventy-three talks later, and I cringed in completely different places. The early-career version of me was nervous, over-prepared on facts, under-prepared on story, and trying to impress the room rather than serve it. The mid-career version of me is critiquing the technique. Both reactions are useful for any founder reading this who has just been asked to give a talk and has no idea where to start.

This is not a guide to public speaking in the abstract. It is a 90-day operating plan for a founder who has decided to add speaking to their distribution stack and wants to be competent (not yet great, just competent) by day 91. The 90-day window is deliberate. It is short enough to fit between two product cycles. It is long enough to get repetitions. And it matches roughly the first two waves of speaking opportunities a founder will hear about after putting “available to speak” on their LinkedIn headline.

Why most founders bomb their first talk

The single most common failure mode I see in first-time founder speakers is treating the talk as a content problem. They open Notion, draft a list of points, build a slide for each point, and rehearse the slides. The talk that comes out of that process is a verbalized blog post. The audience knows it. The energy in the room flatlines around minute six.

Talks are not blog posts. They are time-bound performances where the audience is a captive group and the speaker is competing with phones, side conversations, and the talk that came before. The currency of a good talk is attention, not information. Founders who treat the talk as content delivery lose attention by minute four because the format is wrong. Information dense talks belong in a written long-form. Stage time belongs to talks built around tension, surprise, and a single clean argument.

The second failure mode is over-rehearsal of the wrong layer. Founders rehearse the words. The words are not the layer that matters. What matters is the structure (problem, thesis, evidence, application, close), the rhythm (where the laugh lines land, where the silence lands), and the contingency plans (what to do when the projector fails, when the laugh you expected does not come, when a hostile question lands in Q&A). Most founders rehearse the words ten times and the structure once. The reverse is correct.

The 90-day plan

Day 0 to day 30 is content build. Day 31 to day 60 is delivery rehearsal. Day 61 to day 90 is live repetitions. Each phase has specific deliverables and specific traps.

Days 1 to 30: build the talk that you would pay to hear

Pick one topic, not three. Founders frequently want to give a talk about everything they have learned. The audience does not want a survey. They want depth on one specific question. Pick the question by asking yourself, “What is the contrarian thing I believe about my industry that other founders disagree with?” That is your thesis. If nothing comes to mind, the founder is not yet ready to give a talk. Spend a week journaling and come back to the question. The talk will not work without a real thesis.

Once the thesis is set, build the argument backward. Open a blank document and write the closing line first. The closing line is the sentence the audience repeats to a colleague at lunch. Then write the opening line, which is the line that earns you the next forty seconds. Then fill in the middle, which is the evidence chain that gets the audience from the opening line to the closing line. The middle has between three and five movements. Fewer than three feels thin. More than five overwhelms the working memory of a tired audience.

Each movement needs one piece of original evidence. Original evidence is a number from your own company, a story from your own customer, a screenshot of your own data, or a named example you observed firsthand. Borrowed evidence (research papers, McKinsey decks, generic stats) signals that the founder has not done their own homework. The audience reads it as filler.

In the talks I have given that landed, every movement had at least one piece of original evidence. In the talks that flopped, two or three movements were filler stitched together from secondary sources. The pattern is reliable. Original evidence is the difference between a talk that gets a standing ovation and a talk that gets polite applause and three follow-up emails.

Days 31 to 60: rehearse the structure, not the words

Rehearse the talk five times in the first week of this phase, then once a day for the next three weeks. The first three rehearsals are with slides. The next ten are without slides, talking out loud, walking around. The last week is in front of two or three people who are willing to be honest, not your team and not your investors, who will not push back hard.

The honest test audience is the highest-impact piece of this phase. Find someone who is in your industry but does not work for you. Buy them dinner. Ask them to listen to the talk in full and then tell you the three weakest moments and the one moment they would repeat to a colleague. Note that you are asking for the weak moments first. Most founders ask for what worked, get a polite list, and miss the criticism. Reverse the order.

Stamina is the second variable to train in this phase. A 25-minute talk feels much longer than a 25-minute conversation. The lungs work harder. The voice tires. Founders who have not trained stamina lose energy in minute 17, and the audience hears it. Practice at least three full-length runs without stopping, on your feet, in a room large enough that your voice has to fill it. Cardio in the same week helps more than rehearsal does.

Days 61 to 90: take the live reps

Book at least three speaking slots in this 30-day window. The format does not matter (meetup, podcast, panel, internal company all-hands at a friendly company, university guest lecture). What matters is reps in front of audiences who did not have to be there. Internal team meetings do not count. Customer dinners do not count.

The first live talk will go worse than the rehearsal. The second will go about the same. The third will start to incorporate adjustments only stage time can teach: where to stand for energy, when to step toward the audience for emphasis, how to read the room temperature and accelerate or slow down. By the third live talk, the founder is no longer a beginner. By the fifteenth, the founder is competent enough to be invited back.

The four mistakes that cost founders the audience

The first is reading slides. If a slide has more than seven words on it, the founder is going to read it. The audience reads faster than the founder talks, finishes the slide in three seconds, and spends the next forty seconds waiting for the founder to catch up. The fix is text-light slides with one image and one phrase per slide.

The second is over-quoting other people. Founder talks that lean on Charlie Munger, Naval Ravikant, or Paul Graham quotes feel borrowed. The audience came to hear the founder, not the founder’s reading list. Use one quote maximum, and only when the quote does work that the founder cannot do directly.

The third is the apology opening. “I am not sure if this is the right audience for this talk” or “I have not given many talks, so bear with me” buys the founder nothing and tells the audience the talk is going to be weak. The audience already paid in time. The opening should reward them, not warn them.

The fourth is the overcompressed Q&A. Most founders rush through the prepared talk to leave 15 minutes for Q&A and then run out of questions in three minutes. The fix is to plan four to six “seed” questions a moderator can ask if the audience is silent. Real questions almost always come. The seed list is insurance, not the plan.

The Joey Sendz operator’s framework for talk evaluation

I rate every talk I give and every talk I see on five dimensions, scored 1 to 5 each. Total possible score is 25. Anything below 18 needs work. The five dimensions are:

Thesis clarity (does a smart person leaving the room remember the one argument). Evidence specificity (does the evidence chain feel original or borrowed). Story craft (do the examples land emotionally or just intellectually). Audience adaptation (did the speaker adjust to the room’s energy in real time). Closing strength (did the close land or did the talk fade out).

The 25-point rubric is what I use after every talk to figure out what to work on next. It is also what I look for when I watch other founders speak. It is the operator’s version of what speaking coaches teach, compressed into something a founder can apply in 90 seconds after a talk while the memory is fresh.

After 90 days, a founder following this plan will have a defensible talk, three to five live reps, and a feedback loop. That is not “great speaker” territory yet. That comes around talk 30 to 50. It is competent enough to be invited back, recommended to other event organizers, and confident enough to take the next, harder slot. The compounding starts there.