Do I still need a separate survey tool?

Not necessarily: questionnaire, fieldwork and analysis sit together in deepsight cloud. If you want to keep your current tool, upload its exports as Excel or CSV files and analyze them the same way.

Which question types are available?

Single and multiple choice, scales – for example 0 to 10 for likelihood to recommend – open-ended questions and matrix questions.

What happens when the allowance for AI follow-ups is used up?

The survey keeps running; open-ended questions are then asked without a follow-up. The field monitor shows how much of your allowance has been used.

Are the responses anonymous?

That depends on your questions and on how you distribute the survey. The AI follow-up during fieldwork and analyses started directly from the survey run without anonymization. To mask names, email addresses or phone numbers in open-ended responses, run a separate analysis of the survey dataset with anonymization enabled; this run counts toward your text allowance again. If translation is switched on, texts are translated before masking. The anonymization does not detect every piece of information – check results before sharing them.

Where is the data processed?

deepsight cloud runs on Microsoft Azure in the EU. For AI features such as the draft and the follow-up, we use Azure OpenAI; Microsoft does not use your data to train AI models. Processing is governed by a data processing agreement under Art. 28 GDPR, and a list of the service providers we use is available on request. More under Security.

Create and analyze surveys with AI

Common hurdles

Asking the questions is quick. Everything in between takes the work.

Retyping the questionnaire, keeping an eye on fieldwork, exporting data, coding open-ended responses by hand. Survey after survey.

The challenge: The questionnaire gets written twice
The questionnaire from the last study sits in a Word file or PDF. Someone then retypes every question and every answer option in the survey tool.
With deepsight: A draft from your document
deepsight reads in the questionnaire and turns it into a draft with questions, question types and answer options. Methodological notes appear next to each question.
See the draft
The challenge: Open-ended answers that stop at one word
Online, nobody probes. “What should we improve?” gets “Support.” – and the why stays unclear.
With deepsight: Following up while the person is still answering
The AI asks a short, neutral follow-up, such as “What exactly about support should we improve?” If an answer already says enough, there is no follow-up.
See the AI follow-up
The challenge: Does the follow-up cost you participants?
Every follow-up makes the survey longer. If participants drop out because of it, you lose exactly the answers you were after.
With deepsight: Spotting drop-outs while fieldwork is running
The field monitor shows participations, how many of them are still in progress or were abandoned at each question, and the status of your quota. If many participants drop out at an open-ended question, you lower its follow-up depth – effective from the next participant.
See follow-up depth
The challenge: An export between fieldwork and analysis
Export the data, load it into the analysis software, map the columns, code open-ended responses in a separate file. Export after export.
With deepsight: Analysis without a detour
In the analysis dialog, you choose banner tables for the closed-ended questions and a topic analysis of the open-ended ones – with no export in between.
See topic analysis

1 · Draft

Import a brief or document

deepsight creates a draft from your brief or an existing questionnaire. You decide on every suggestion yourself.

From a brief to a questionnaire — Illustrative example · Read the content
An illustrative brand study: a brief becomes open and rating-scale questions. An AI flag highlights a leading question. The researcher reviews and applies a neutral suggestion, then publishes the questionnaire.

An optional AI follow-up

For open-ended questions, deepsight can ask a follow-up when an answer is brief. Participants can skip it.

How the follow-up works

For brief answers: the AI follows up when an open-ended answer says little about the question.
In context: the follow-up refers to the answer and to the goal of the question.
Optional: participants can skip the follow-up.
Can be enabled: you switch the follow-up on; it uses a separate allowance.

AI asks a focused follow-up — Illustrative example · Read the content
“The design.” prompts the AI follow-up “What do you like about it?”. The participant types “The packaging feels premium. But the price is too high for me.” Topics and sentiment preserve both perspectives. Follow-ups can be enabled, and responding remains optional.

Reaching participants: You share the participation link through your own channels, such as your email list, intranet or panel provider; redirecting participants back to the panel is supported. Participants do not need an account; participation is currently available in German only. Questionnaires can include matrix questions, skip and filter logic, and a quota question with screen-out.

3 · Analysis

Open-ended and closed-ended responses together

Analyze open-ended and closed-ended questions together — Illustrative example · Read the content
100 illustrative responses, two perspectives: the recommendation question produces an NPS of +31 (49% promoters, 33% passives, 18% detractors). Open responses reveal design, quality and price as topics. An excerpt labeled as an original response is visibly typed. Both analyses belong to the same study.

Responses go straight into the analysis. Open-ended answers are organized by topic and can be compared with the closed-ended questions. You build tables and analyses from these results and check them before you share them.

Introductory offer & billing

What applies, and what counts

Introductory offer

Introductory customers receive the surveys package at no extra charge with a paid cloud plan.

The duration and usage scope are specified in your offer before you start, and we activate surveys for your organization as part of the introduction. You receive a transparent follow-up offer before it ends; there is no automatic paid extension.

Counts toward your allowance: Analyzed open-ended responses count toward your text allowance.
Repeat analysis: A new completed analysis run can count the same texts against your allowance again.
Data quality checked: Once you have completed the data quality check, filtered-out responses do not count. Without this check, every row in the dataset counts.
Does not count: Closed-ended responses and quantitative tables do not use the allowance. There is no additional per-participant survey fee during the agreed introductory period.
Separate allowance: AI follow-ups use a separate agreed monthly allowance.
Agreed in advance: Larger surveys and additional usage are agreed in advance.

All plan details are on the pricing page.

Use cases

What you can find out with a survey

How satisfied are customers – and how do they explain their rating?
What would employees change about processes, collaboration or equipment?
What do users expect from a new offering before it launches?
How do participants rate an event, a program or an initiative?
Which topics do participants raise that nobody thought of when writing the questionnaire?
What has changed since the last survey?

Good to know: You’ll find examples for each field under customer surveys, HR and social research.

Starting point

What you start with

A briefGoal, target group and topics in a few sentences. This turns into questions with a question type, answer options and a research goal for each question – as a suggestion.

An existing questionnaireFinished or half-finished, as a Word, text or Markdown file or as a PDF containing text. In the editor, you can still see which part of the document each question comes from.

Your brandingYour organization’s logo and accent color, plus your own texts for the introduction, the end page and screen-outs. Font sizes are fixed so the survey stays easy to read, on a phone as well.

Process

From draft to checked result

Refine
You work on the draft in the editor. On request, the AI suggests changes, such as a more neutral wording, and lists them so you can accept them one by one.
Test
In a test run, you go through the paths in your questionnaire and see where the AI would follow up, and why. Test responses don’t count and never end up in a dataset.
Launch fieldwork
You launch fieldwork, with a fixed end date if you like. You can adjust your quota targets during fieldwork; any reduction is documented.
Analyze
Several open-ended questions are analyzed in one run. Beforehand, a preview shows how many texts count toward your allowance and how much of it is left this month. Each participation counts once per analyzed open-ended question, even if the answer is empty; an interim and a final analysis each count in full.
Topic analysis
Check and compare
You check the topics against the original responses and adjust them. Then you compare groups, with significance indicators if you like.
Group comparison

Results

What you have at the end

Banner tablesThe closed-ended questions as cross-tabs with column percentages or means, right in the cloud. deepsight builds the banner automatically from the quota question and other single-choice questions.

Topics with evidenceOpen-ended responses, including follow-ups, organized by topic, with the verbatim responses for each topic.

Report and PowerPointShare results as a report via link, export them to PowerPoint or put them together in a dashboard. You can export the data to Excel or CSV.Go to reporting

The next waveYou duplicate a survey as a follow-up survey. If a question matches an existing time series, you add the results as a new wave.

Good to know: Wave comparison is included in every paid plan – see the plan comparison.

Methodology

What matters in AI-assisted surveys

Suggestions stay suggestions

Accepted AI suggestions are marked in the editor. Methodological notes you decline remain visible next to the question, so it stays clear what was decided.

Follow-ups in moderation

For each open-ended question, you set how many follow-ups the AI may ask at most (follow-up depth), and in which tone. Empty or skipped answers never trigger a follow-up.

A fixed questionnaire in the field

Once fieldwork starts, the order, question types and answer options stay as they are. You can only add new questions after confirming, because earlier participants never saw them. In follow-up surveys this is locked so waves remain comparable.

Provisional stays provisional

Analyses during fieldwork are marked as provisional. As soon as you close fieldwork, the banner tables are updated to the final responses; a fixed end date only closes the participation link.

FAQ

Questions about surveys in deepsight cloud

Related to surveys

MethodologyHow to review AI-based analysis with expert judgment, correct it and keep a record of each result.Go to methodology Features at a glanceAll analysis features of deepsight cloud, from the data quality check to the assistant.Go to features Anonymizing text dataMethods and limits of automated masking in open-ended responses, with an illustrative example.Read article

Plan your next survey

We will show you the draft, the follow-up and the analysis using your brief.

Ask about the introductory offer Request a demo

deepsight cloud · Surveys Your brief becomes a questionnaire draft.