The Honest Audience Read

What this is for

You know your talk too well to hear it the way a stranger does. This guide sets up three AI personas, each reading your draft the way one type of audience member actually reads it, so you get a reaction instead of a compliment. It is not a rubric or a score. It is a read.

When to use it

Use this once you have a full draft or a solid outline, before you lock the structure. It works best after guide 20, once the shape exists, and before you spend time polishing sentences that might not survive the read.

The prompt kit

I want you to read a talk draft as three different audience members, one at a time. Before I paste the draft, ask me anything you need to know about the event and the room - industry, seniority, why they're attending.

Here's my speaker context: [paste from guide 01]
Here's the draft: [paste it]

First, read this as THE SKEPTIC: someone who has sat through a dozen talks like this and assumes it's going to be fluff until proven otherwise. Tell me, in their voice: where do they tune out, where do they roll their eyes, and what would make them lean back in. Don't soften it.

Now read the same draft as THE PRACTITIONER: someone who does this work every day and knows the subject cold. Where does the talk say something they already know and dress it up as insight? Where does it say something genuinely useful they hadn't put into words themselves?

Now read it as THE MEETING PLANNER: the person who booked this speaker and has to answer for the choice afterward. Does this talk deliver on what the session description promised? Would they book this speaker again? What's the one thing they'd ask you to fix before the next event?

Now step back and give me the throughline across all three reads: what's the single change that would help with the skeptic, the practitioner, and the planner at once?

What good output looks like

Each persona should sound distinct, not like the same review with a different label pasted on top. The skeptic should flag specific moments, not vague dissatisfaction. The practitioner's read should surface at least one place where the talk is teaching something they already knew, which is the most common blind spot in expert-level talks. The planner's read should connect back to whatever the session was billed as, since that's the actual standard they're judging against.

Watch out for

Personas that just agree with each other. If all three reads land in the same place with the same praise, the AI is not actually differentiating the roles. Push back and ask it to be harder on the draft, or specify the room more precisely so the personas have something real to react to.

Treating one persona's note as universal. The skeptic and the practitioner sometimes want opposite things - the skeptic wants more story, the practitioner wants less setup and more specifics. Decide which audience you're actually optimizing for at this event before you act on a note.

Letting the AI invent details about "typical" audiences. If you didn't describe the room, tell it not to guess who's in it. A generic persona reacting to a generic room isn't useful. Give it real detail: job titles, what they're worried about right now, why they signed up for this session specifically.