Virtual focus groups: 100 consumers, hours not weeks

Booking a classic focus group is a commitment. You rent a facility, recruit eight to ten people who fit the target, feed them sandwiches, watch from behind the glass, and two or three weeks later a deck arrives. The invoice sits somewhere between €3,000 and €8,000. For a naming call, a pricing tweak, or an ad you have to ship on Thursday, that isn’t research. It’s a reason to skip research and go with your gut.

A virtual focus group — sometimes called synthetic research — is the shortcut I’ve come to trust for a lot of those decisions. Instead of ten real people in a room, you run 100 to 200 AI personas, each built with a different profile, budget and starting opinion, and you let them react to your idea. Ours is called the SimulatOR, and the word I want you to hold onto is argue.

What actually happens inside it

Most people picture a survey with robots. It isn’t that. A survey freezes one opinion per person and averages the column. Asking a single AI “what do you think of this name?” is worse: you get one confident voice that has quietly read your prompt and decided to agree with you.

The SimulatOR does something closer to a real room. Each persona gets a life. A 34-year-old parent with two kids and a tight budget. A sceptical procurement manager. A brand-loyal early adopter who’ll pay more for the logo. Then they don’t just vote — they talk to each other across 10 to 20 rounds. Someone raises an objection. Someone else counters it. A price that felt fine in round two gets torn apart in round seven, when a budget-conscious persona points out the competitor is €40 cheaper. Opinions move. That movement is the data.

By the final round you’re not reading an average. You’re reading how a consensus formed, which objections kept coming back, and which ones died quietly. That’s the part a spreadsheet of survey answers can never give you.

What it’s genuinely good at

Here’s my honest rule: the SimulatOR can test anything a human could judge by reading, watching or comparing on a screen. That covers a lot of expensive decisions.

• Naming and taglines. Run five candidates and watch which ones get misheard, mocked or ignored.
• Pricing and packaging. Put three tiers in front of 150 budgets and see where people flinch, where they upgrade, and which “good, better, best” middle option actually wins.
• Ads and landing pages. Test the hook before you spend on media. This pairs well with a video concept you’re about to shoot; far cheaper to kill a bad script here than in the edit.
• Positioning and messaging. Which promise lands, which one sounds like everyone else, which one a rival could steal word for word.

For any of those, you get a report in 24 to 72 hours for a fraction of the €3,000–8,000 a classic panel costs. When you’re choosing between three directions and don’t want to bet the quarter on a hunch, that speed changes how often you bother to check at all.

Where it falls flat, and I’ll say so

A synthetic panel has no body. So it cannot tell you a single useful thing about:

• Taste and smell. No AI knows whether your new recipe is nicer than the old one.
• Touch and texture. The feel of a fabric, the weight of a bottle, the click of a lid: invisible to it.
• Anything physical or sensory. If the decision lives in the hand or the mouth, close the tab and go find real humans.

There’s a subtler limit too. These personas are built from patterns in data, so they’re sharp on the typical and weak on the genuinely weird. A truly novel idea that no market has reacted to yet is exactly the case where I’d trust them least. They’re a brilliant map of what people usually do, not a crystal ball for what one specific person will do next Tuesday.

It sharpens the bet, it doesn’t remove it

I want to be careful here, because it would be easy to oversell. A virtual focus group does not replace talking to your actual customers. It makes that conversation cheaper and better aimed.

Think of it as a filter. You walk in with eight ideas and no way to prioritise. The SimulatOR kills the four that fall apart under argument, flags the two objections you hadn’t thought of, and hands you the two directions worth putting real money and real interviews behind. You still talk to real people. You just stop wasting their time, and yours, on the options a room full of synthetic sceptics already dismantled.

That’s also why it sits so naturally next to strategy work. The simulator is great at which of these, and honest strategy is what turns the winner into a plan. One narrows the field fast; the other decides what the field was for.

Where I’d start

If you’ve got a naming shortlist, a pricing page you’re nervous about, or an ad you’re not sure survives contact with a sceptic, that’s the sweet spot. Run it through 150 personas, let them argue for twenty rounds, and read the fight.

You’ll get a clear recommendation, the objections that kept surfacing, and an honest confidence level. In days, not weeks, for a fraction of a classic panel. See how the SimulatOR works and what a run looks like before you commit anything.