Loading...

Put your ads in front of your brand’s own audiences. Learn what to run, what to fix and what to keep.
Focus Groups is the AI synthetic focus group built into Ad Legends, with a panel drawn from the audiences saved to your brand. Use Ad Test for fast AI ad testing: which of up to three ads leads. Run a full study for the why: where audiences split, what to fix first and what to test next.
The label “AI focus group” covers three different things (synthetic panels of AI-generated participants, AI-moderated interviews with real people, and AI analysis of recorded human groups), and Ad Legends Focus Groups is the first: a synthetic panel built from your brand’s own saved audiences.
Already signed in? Open Focus Groups
The creative does much of the selling. Testing it the traditional way means recruiting, scheduling, a facility and a moderator, so a lot of work goes out with no read at all.
Published figures, from the sources named. They are not Ad Legends data.
Nearly half
Across nearly 450 consumer packaged goods campaigns, the creative drove nearly half of the short-term incremental sales: the largest single factor.
40%
In Kantar’s Media Reactions survey of more than 1,000 senior marketers, 40% said they test little or none of their advertising, while 80% agreed ads should be tested early.
Source: Kantar, 2025
$8,000 to $12,000
Two in-person focus groups in a major metro can reach $8,000 to $12,000 per project before travel or rush fees. Facility rental often runs $1,500 to $2,500 a session, participant incentives $100 to $150 a person on average, and one online group may run $4,000 to $7,000.
Source: Greenbook, 2025
$7,000 to $20,000+
Per focus group, in one full-service US research agency’s published price guide.
Source: Drive Research, 2026
Several weeks
Typically, from kickoff to report, in published practitioner estimates for a focus group or qualitative project.
Sources: SIVO Insights, 2025 · CleverX, 2026
6 to 9 people
In one group, for one to two hours. Strong personalities and groupthink can shape what gets said.
Source: Nielsen Norman Group
Ad Test reads up to three ads in a few seconds, at 1 credit per ad. A full study comes back in minutes, longer with a bigger panel or more pieces, on almost any kind of work you made in Ad Legends, and its credits show before you launch. No recruiting, no scheduling, no facility. It runs when the work is ready.
Test far more of your work, far earlier, and save real research for the high-stakes calls.
Same brand, same audiences, two depths. Pick the one that fits the decision on the table.
Which ad leads?
Pick up to three image ads you made in Ad Legends, say what you want people to do, and get one read back in a few seconds.
Why, for whom, and what next?
The Research Director helps you set up a study on up to three pieces of saved work, runs it with a panel built from your audiences, and hands back an executive brief.
Start with Ad Test. When the read is too close to call, or you need to know why, open the full study straight from the result.
Two examples of the shape of each answer. The ads, the audiences and every number in them are made up.
Ad Test · illustrative
What you want people to do: start a free trial
Creative B leads (7.4 to 6.2). Strongest: the offer or promise. In the way: there is too much to take in.
How sure: mediumRead by 8 saved participants
Executive brief · illustrative
Which route should we develop next? · Two static ads · Same placement · Two audiences
Confidence: medium
Four steps, with the Research Director beside you. It can suggest answers for your setup. You decide which ones go in.
01
Pick one to three pieces of saved work. Compare like with like, or study one piece on its own.
02
Choose the audiences, set how much each one counts, and see the panel of 5 to 24 AI-generated participants.
03
Name the decision and what matters most: attention, clarity, credibility, brand recognition, action and more.
04
Check the work, the panel and the credits, then start. Close the tab if you like: the study keeps running.
The call, how confident it is, the audience split, what to fix first, what to protect, the risk still open and the next test.
Download the brief as a PDF for the meeting where the work has to move.
Ask the finished study a question. The answer comes from what that study found.
Every study stays with the brand, live and past, tied to the version of the work you tested.
A full study reads one to three pieces of saved work from across Ad Legends. Ad Test reads image ads. What each kind of work supports shows during setup.
Fast Ads, ad assets, bulk ads, generated ads, billboards and out-of-home.
Animated ads, sizzle videos, finished Cutroom cuts, Video Studio clips, video scripts and audio spots.
Landing pages, pitch pages, pitch decks and presentations.
Ideas, briefs, strategies, Ad Strategist routes and manifestos.
Social posts, blog posts, PR and outreach emails.
Ad campaigns and the campaign work inside them.
Search by name. No name to go on? Describe what the work shows and the picker finds it by what is in the picture.
In Ad Test, search your ads by headline or mood.
Every participant starts from an audience saved to your brand. They are AI-generated, not real people, and you see who is in the room before anything runs.
Brand Memory
Audiences live in Brand Memory with the rest of your brand, so every test starts from the same definitions.
See Brand MemoryAudience Atlas
Audience Atlas researches and ranks the audiences worth reaching and saves them to your brand, ready for a panel.
See Audience AtlasSide by side
A full study can draw on up to 12 audiences and weigh each one, so you see where they agree and where they part ways.
Three kinds of research get called a focus group now. Here is what each one is, and what it is for.
| Traditional focus group | AI-moderated interviews | Synthetic focus group Ad Legends | |
|---|---|---|---|
| Who takes part | 6 to 9 recruited people per group, in a room or online. Source: Nielsen Norman Group | Real, recruited people, each interviewed by an AI moderator. | AI-generated participants built from your brand’s saved audiences. A full study seats 5 to 24 of them. |
| How long to an answer | Typically several weeks from kickoff to report, in published practitioner estimates. Each session runs one to two hours. Sources: SIVO Insights, 2025 · CleverX, 2026 · Nielsen Norman Group | Faster than a live group, but people still have to be recruited and take part. | A few seconds for an Ad Test read. A full study starts the moment you launch it: no recruiting, no scheduling. |
| Typical spend | $8,000 to $12,000 for two in-person groups in a major metro, or $4,000 to $7,000 for one online group. One agency lists $7,000 to $20,000+ per group. Sources: Greenbook, 2025 · Drive Research, 2026 | Recruiting and participant incentives, plus the platform. | Legend Credits from your plan: 1 credit per ad in Ad Test. A full study shows its credits before you launch. |
| What you learn | What people say in the room, in their own words, with follow-up questions on the spot. | Real people’s answers in depth, gathered one at a time and summarized for you. | Which ad leads, or too close to call. In a full study: where audiences split, what to fix first, what to protect and the next test. |
| Best for | High-stakes launches and investments, where you need to hear from real customers. | Hearing from more real customers than one live group can hold. | Testing far more of your work, far earlier: picking between ads, fixing a draft, choosing the next test. |
| Limits | Slow to arrange and costly per group, and strong personalities and groupthink can shape what gets said. Source: Nielsen Norman Group | Still needs recruiting and real people’s time, and the read is only as wide as who took part. | Participants are AI-generated, not real people. Directional: not a representative sample and not a forecast of results. |
Participants are AI-generated, not real people. Findings are directional and do not predict business outcomes. When the stakes are high, check the call with real customers or market evidence.

Your brand’s audiences, your actual work, and an answer for the next decision.
Already signed in? Open Focus Groups
Last updated