Audience pressure-testing
A three-person AI-simulated panel — e.g., three GPs, three specialists, or three patients — whatever your primary audience is. Scored across five dimensions, with the divergence flagged. A sounding board before you commission live research, not a replacement for it.
What you get back
Each persona scores your draft across five dimensions, with a panel mean and a band — Landing well, On track, Mixed reception, Rethink the approach. Tells you where the draft is solid and where it’s slipping.
When the three personas disagree, the disagreement gets a callout. The part you can’t get from a single hedged AI voice — and usually the part you can’t see in your own draft.
Specific, ranked changes — what to fix, why it matters, and the exact persona quotes you can paste into a brief. PDF export and 30-day shareable link included.
Every report includes a behavioural read and a quote bank you can paste straight into a brief.
Who we built it for
Before MLR, before the client sees it.
If a client clocks AI in your draft, you’ve got a problem. Audience Lab is the opposite — a three-person AI-simulated audience panel reacting to what you’ve already written, not AI writing for you. The thinking, the structure, the polish: still yours.
Polished ≠ landing. Accurate ≠ landing. Compliant ≠ landing. That’s the gap we pressure-test.
Before the campaign launches, before you commission live research.
Live research is for the messages worth committing to. Audience Lab narrows down which ones those are — in days, not weeks. Procurement-friendly: Sydney-hosted database, no PHI, full data-handling shared on request.
Doesn’t replace your agency. Your agency builds it. We pressure-test it.
Before the advisory board, before the congress deck goes live.
Works for non-promotional comms — scientific narratives, advisory board prep, congress decks, MSL field materials. The product reads how content lands, not whether it’s promotional.
AU specialist personas live for oncology, cardiology, endocrinology. More in development.
Before the brand meeting, before the SOW commits you.
If your team’s been roped into post-hoc verification of brand-side AI drafts, this is the layer that sits earlier — pressure-test the messaging before you commit to recruiting respondents.
The report is your starting point, not your deliverable. Bring a sharper read into the brand meeting than the brand team has.
How it works
Step 1
Email your draft — PDF, DOCX, or slides — with a short brief: who it's for, the therapy area, and what you're unsure about. A single asset, or a campaign of up to 10.
Step 2
We read what you wrote first, then build a three-person AI-simulated panel of your audience — three GPs, three specialists, or three patients — varying in geography, career stage, and disposition. The depth that “ask ChatGPT how a GP would react” doesn’t reach.
Step 3
Scored across five dimensions, divergence flagged, with a quote bank and ranked improvements. Reviewed by Fei before it reaches you, then sent as a PDF and a shareable link — usually within a few business days.
Audience Lab doesn’t generate clinical claims. Your draft is the source of truth; the output is AI-simulated reaction patterns.
Common questions
You can paste your draft into ChatGPT and ask “how would a GP react”. You’ll get one hedged voice with no tension. Audience Lab gives you three reactions from an AI-simulated audience panel — scored across five dimensions, with the disagreement flagged. The disagreement is the diagnosis.
Often. When the three personas disagree about what your draft is for, that’s a brief-level signal. Run it on the brief, then on the draft — the gap between the two reads tells you whether the brief or the draft is doing the lifting.
No. Sounding board first; recruited research after, when validation matters. Audience Lab gives you a structured first-pass read in the cases where there’s no time or budget for a live focus group — and a way to narrow down which messages are worth committing to before you do.
Audience Lab doesn’t generate clinical claims — your draft is the source of truth. The output is AI-simulated reaction patterns and where the personas diverge. No reference fabrication risk.
Your first evaluation is free — send us a draft and see the output on your own work. After that, we work in per-evaluation packs, scoped to what you need. Tell us what you’re working on and we’ll come back with scope, turnaround, and price.
How we handle your content
When we run your evaluation, your draft goes to Anthropic’s Claude API. Anthropic doesn’t train on customer API data, and we don’t fine-tune on your content. Our database is in Sydney; AI inference runs in the US. We delete your asset and its extracted text within 30 days of delivery.
Pricing
Your first evaluation is free — send us a draft and see how the panel reads your own work, no commitment. After that, we work in simple per-evaluation packs, scoped to what you need. A sounding board before you commission live research, not a replacement for it.
Why this exists
I’m Feisia Dam — registered pharmacist, 20 years in pharma (AstraZeneca, Pfizer), now running Lucid MedComms in Sydney. I built Audience Lab because I kept watching my own work — and the work of writers I respect — ship polished, MLR-clean, on-brief. And then not land.
AI made us faster. It didn’t tell us whether the content was landing. This is the layer I wanted between ‘draft is done’ and ‘audience reads it’.
— Fei
Get started
Tell us what you’re working on and we’ll run your first evaluation — free. You get the full report: scored, divergent, with a quote bank, a PDF, and a shareable link.
Your first evaluation covers a single asset. Multi-asset campaigns — just ask.
Or email us at hello@lucidmedcomms.com — a real person usually replies same-day on weekdays.