Skip to article

How Digitas Pretests Ideas With Synthetic Audiences

Digitas builds synthetic audiences out of audience data, then talks to them. What CEO Amy Lanzi said, the caveat she added, and a first step for a team.

By Jacqueline V. TwillieNovember 20, 20263 min readGuide
Quick answer

Digitas pretests ideas by building synthetic audiences out of its audience data and talking to them before an idea goes to a client, said Amy Lanzi of Digitas North America at The Female Quotient's AI Summit on November 10, 2026. She added that you can do a lot with AI, but it does not mean the idea is strategic.

Digitas builds synthetic audiences out of its audience data, then talks to them. Amy Lanzi, CEO of Digitas North America, spoke about this at The Female Quotient's AI Summit in New York on November 10, 2026. Her remarks are reported first.

How Digitas uses a synthetic audience

Lanzi said the point is to test their ideas with those audiences before they go to clients. They pretest with them and scenario plan the work or the media plan, which she called a way to pressure test themselves.

Lanzi's caveat

Lanzi added that you can do a lot with AI, but it does not mean the idea is strategic.

Try it on one idea first

BNEDai's suggestion: try it on one idea first, then ask a few real customers whether the synthetic audience got it right. The comparison tells you how far to trust the synthetic one.

Where BNEDai uses made-up material

BNEDai's published practice is a different one. It rehearses an AI agent against made-up data before real data is involved.

The session page for the October 7, 2026 session with Women Applying AI carries a sample inbox, published by Jacqueline V. Twillie, in which every name is made up. She did not mention it in the session. A team can paste it into a Chief of Staff AI agent, ask for today's brief, and check the result line by line against a list of what the brief holds, with no real inbox connected. The steps are in how to test a Chief of Staff AI agent with a made-up inbox.

She also describes a demo habit for showing an AI agent to an audience: have the AI agent build a spreadsheet with demo names, fake emails and fake phone numbers, keep a separate demo AI agent, and share only that one. She does this whenever she shares her own dashboard or systems. See how to show an AI agent to an audience without real data.

What stays with a person

BNEDai's default is that an AI agent drafts and a person decides what goes out. In Jacqueline V. Twillie's words: "Every AI agent I set up starts with the same rule: draft, never send." A synthetic audience's reaction is input to that decision. Lanzi's caveat points the same way: you can do a lot with AI, but it does not mean the idea is strategic. The page on every AI agent starts with the same rule covers where the limits are set.

Frequently asked questions

How does Digitas use synthetic audiences?

Lanzi said they build them out of audience data and talk to them, to test ideas before they go to clients, pretest, and scenario plan the work or the media plan.

What was Lanzi's caveat?

That you can do a lot with AI, but it does not mean the idea is strategic.

Is BNEDai's made-up inbox a synthetic audience?

No. It is a separate practice that uses invented data to test an AI agent before real data is involved.

Source: remarks by Amy Lanzi, CEO, Digitas North America, at The Female Quotient AI Summit, New York, November 10, 2026. The made-up inbox and demo practices are Jacqueline V. Twillie's published work, and the suggestions are BNEDai's.

Reviewed on November 20, 2026.

More on Real Workflows for Corporate Executives