CatalyseAI · AI survey consultant
Plan a study in a short chat
Tell the consultant what you want to test. It asks a few short questions, suggests a default for each one, and shows the whole study on one plan card before anything is built.
Why it matters
Most bad research is decided at the design stage
The wrong method, a leading question or too few people per idea can't be fixed once the answers are in. Setting a study up well takes method knowledge that not every team has on hand, and a blank questionnaire gives no warning when something is off. The consultant brings that knowledge into CatalyseAI as a conversation. You answer in plain words, and you keep the final say on every setting.
The conversation
How the conversation goes
- 01
Say what you want to learn
Type it in your own words, for example “we have three new flavors and need to pick one”. The consultant matches it to a study method from our catalog.
- 02
Answer a few short questions
It asks no more than two at a time: what you are testing, how many ideas, how people will see them, whether to show a price, the category and brand, and who should answer. Each question comes with a suggested answer you can accept.
- 03
Check the plan card
Every setting sits on one card. Change any of it by typing, for example “make it 400 people per idea” or “drop the price question”.
- 04
Confirm, then keep editing
Nothing is added to your questionnaire until you confirm. Then the method's questions load into the builder, and you can add your own questions after them.
We have three new toppings and want to launch one.
A concept test fits: one topping per person, so each is judged on its own. How should people see them? I suggest a picture with a short description.
Picture with a description.
Should we show a price? If yes, one price per topping, with the pack size.
Yes, $5.99 for a 12-inch pizza.
What is the category and the brand name people will see?
Frozen pizza. The brand is Oven Club.
Who should answer? I suggest US adults who bought frozen pizza recently, with 300 people per topping.
Only parents, please.
Done. The plan card now says US parents who bought frozen pizza recently. Video interview stays off, and the toppings are judged against each other. Confirm to load it.
Plan confirmed. The concept test loads into the builder.
The plan card
What it sets up for a concept test
Every setting on one card. Change any of it by typing, and nothing is built until you confirm.
Concept test settings
- What is testedA product, a service, a menu item, an offer, a feature, or a way to position a product.
- IdeasOne to five, with the names people will see.
- FormatText, a picture, a picture with a description, or a video people must watch.
- PriceOptional. One price per idea, with the pack size or period.
- AudienceUS adults who buy the category, unless you choose otherwise.
- People per idea300 suggested, 200 at the least.
- Video interviewOff unless you turn it on. When on: one video question with two AI follow-ups.
- VerdictIdeas judged against each other, against targets you set, or against your current product.
What it looks after
What the consultant looks after
Five things the consultant checks so a first study still reads well.
- 01
A method that fits the decision
It suggests a method and explains why. Bring more ideas than a concept test can take, and it suggests screening them first or splitting them into two studies.
We have three flavors and need to pick one.
Suggested methodConcept test- One flavor per person, so each is judged on its own
- Scores from 0 to 100 for each flavor
- A plain verdict for each, with reasons
What if we have eight?
A concept test takes up to five. Screen the eight first, or split them into two studies.
- 02
Defaults that hold up
A suggested number of people per idea, fully labeled five-point scales and a sensible starting audience, so a first study still reads well.
- People per idea300suggested
- Scales5 points, every point labeledsuggested
- AudienceUS category buyerssuggested
- 03
The same core questions every time
Each method's core questions are worded the same way in every study, so you can compare scores across your own studies.
Study 1, MarchHow likely would you be to buy this?
Study 2, JuneHow likely would you be to buy this?
Study 3, SeptemberHow likely would you be to buy this?
- 04
One set of settings
The chat, the setup form and the plan card all read the same settings. Change something in one and it shows in the other two.
ChatSetup formPlan card - 05
A warning before a score breaks
Reword any question freely. If you change the answers or the scale of a question that feeds a score, it warns you first, because that score can no longer be calculated.
Q3 · scaleChanging this scale stops the Trial score from being calculated.
Methods it sets up
The concept test first, then one method at a time
Each method adds its own questions to the chat. The order below is the order we plan to build them in.
FAQ
Questions, answered
Which studies will it set up?
The concept test first. Pricing studies, video ad tests, brand funnel and tracking, pack tests and other methods follow one at a time, each with its own set of questions in the chat.
Can I change the study after it is built?
Yes, in the chat or in the builder. Wording can change freely. Changing the answers or scale of a core question turns off the score that question feeds, and the builder tells you before you do it.
Does it work for audiences outside the US?
The first method is set up for US audiences only. If you ask for another country, the consultant tells you so.
Related
Goes well with
Have a study in mind?
Tell us what you want to test, and we will show you how the consultant sets it up in CatalyseAI.
Nearest office: Orlando, FL (corporate office)