Building a Survey With the AI Builder

Three ways to create a survey, the question types available, and how branching logic works.

There are three ways to create a survey in FeedbackRobot. Most people should use the first one.

Describe what you want to know

Type a plain-language description of what you’re trying to find out — “check whether diners were happy with the speed of service and whether the server explained the specials” — and the AI builds a working survey from it.

It picks question types, orders them sensibly, and words them neutrally. You get a draft in seconds that you then edit, rather than a blank page.

Be specific about the decision you’re trying to make. “Customer satisfaction survey” produces something generic. “Find out whether our new online booking flow is confusing for first-time customers” produces something you can act on.

Start from a template

If you want a standard, comparable measure — NPS, CSAT, CES — a template gets you there faster and keeps the wording consistent with how those metrics are normally asked.

→ Using a survey template

Start from blank

Full manual control. Slowest route, occasionally the right one when you have an exact set of questions already agreed.

Question types

Type	Use it for
Rating scale	Measuring something you’ll track over time
Multiple choice	Known, countable options
Open text	The why behind a score — where your best insight lives
Yes/no	A clean branch point
Ranking	Forcing priority between competing options

A good short survey usually mixes one rating question with one open question. The rating gives you a trend line; the open answer tells you what to do about it.

Branching logic

Branching shows different follow-ups depending on how someone answered. The classic use: if a customer rates you poorly, ask what went wrong; if they rate you well, ask whether they’d leave a public review.

This is worth setting up even on a short survey. It means unhappy customers get asked for detail you can act on, and happy customers get pointed toward a review — without either group seeing questions meant for the other.

Preview before sending

Preview the survey as a respondent will see it, on both desktop and mobile. Most feedback surveys are opened on a phone, and a question that reads fine on a laptop can be unwieldy on a small screen.

Check for: questions that scroll off, rating scales that wrap awkwardly, and anything that reads like an interrogation.

Editing after it’s live

You can edit a live survey, but be careful with questions that already have responses — changing wording mid-flight means you’re comparing answers to two different questions. For anything you’re tracking over time, create a new survey instead of editing the old one.

Building a survey with the AI builder