Dear {{ first_name | reader }},
Somebody in your organisation, your agency or your client's boardroom will show you a synthetic audience result this year. A chart, a sentiment score, a line saying Message B 'tested best' with regulators, residents or customers. Nobody in the room will have met those people. They do not exist.
I tested synthetic audiences in the early days of 2023 and immediately saw the potential, but also the possible limitations. This edition covers what to do when that chart lands on your desk, and how to say 'yes, under these conditions'.
Enjoy.
PS: if you have not done so, please add your insights to the RiskComms Global Outlook Project - and share with other communication professionals.
WAG THE DOG NEWSLETTER | ISSUE WEEK 40, 2026
KEY TAKEAWAYS
Start with the exploring jobs. Awkward questions, misreadable phrases, jargon checks and spokesperson drills save time and cost little when the model is wrong.
Validate in peacetime. A synthetic audience is most tempting mid-incident, which is when you can least check it.
The research is weakest where your work is hardest. AAPOR flags marginalised groups, culturally specific concepts and emotionally charged topics.
Six checks give you a defensible answer for a board, a vendor or a chief executive taken with a chart.
Never use simulation to choose protective action or to stand in for people who have been harmed.