Method

Traceable from conversation to statement.

Here is a closer look at how Rezendo works: how an interview is built, how we analyse and how you can tell that the results hold up. You do not need this knowledge to use Rezendo. But you can check it all the same.

Book an intro call
91%
of core patterns supported several times over after 50 conversations
How we work

What stands behind a report.

Seven topics that experts check before they trust a new method. Here are the short answers; the detailed ones are in the questions and answers further down.

Conversational research · Qual at scale

Between qualitative and quantitative research

Qualitative research uncovers the why and the what, usually in a few in-depth conversations. Quantitative research measures how much and how often. The choice between them was never about content but about economics: conversations were expensive. Conversational research combines both, because the AI holds hundreds of conversations and follows up where it gets interesting.

You get:
  • The reasons behind a number
  • How many people share a reason
  • Both from the same conversations
Goal · Guide · Test run

How an interview is built

You name your goal, taboos and the questions that should stay comparable. From these, the AI derives the conversation. Before it is sent, you experience it yourself in a test run and approve it.

At a glance:
  • Benchmark questions such as NPS stay word for word the same
  • Open questions with follow-ups where it gets interesting
  • Form of address and language suited to your participants
  • At the start, everyone learns that they are talking to an AI
Analysis · Evidence

How the analysis works

Figures are counted from the answers, not estimated by the language model. Every statement in the report states how many people support it and leads to the original voices.

At a glance:
  • Strength of evidence shown openly for every statement
  • Analysis per segment
  • Where answers are ambiguous, the report says so
Objectivity · Reliability · Validity

How robust the results are

Every conversation follows the same goal and the same analysis, with no interviewer effect. In our surveys, the patterns are settled after about 50 conversations and do not change afterwards; this is consistent with saturation research (Guest et al., 2006; Hennink & Kaiser, 2022). How closely humans and AI agree in their interpretation is something we are measuring in a research project with OST.

At a glance:
  • Core themes from 30 interviews; from 50, 91% supported several times over
  • Every statement can be checked down to the original voice
  • Who responded is shown in the report; representativeness depends on your invitation list
Do the maths yourself in the Insight Check
Full service · Self-service

What you control and see

With full service, Rezendo takes care of preparation, interviews, analysis and presentation. Once set up, your team can also run surveys itself. Either way, you can read every conversation word for word.

At a glance:
  • Live status of participation
  • Every conversation word for word, downloadable as a PDF
  • Further analyses, for example for another committee
NPS programme · API

How it fits in

Your existing satisfaction measurement keeps running; Rezendo supplies the reasons behind it. Through an API, interviews start automatically after an event in your system, for example after a service case.

At a glance:
  • No new tool for your participants: a link is all it takes
  • Invitations sent from your own domain
  • Interviews in your branding
Switzerland · FADP · GDPR · EU AI Act

Data protection and data location

Rezendo is hosted in Switzerland. The AI works with GPT models via Microsoft Azure, within the EU Data Boundary including Switzerland; no AI is trained on the answers. Before we start, we sign a data processing agreement with you.

At a glance:
  • Confidential and pseudonymised
  • Analysed at segment level, groups only above a minimum size
  • Compliant with the EU AI Act: not a high-risk application, and participants know from the start that they are talking to an AI

The Rezendo Insight-Check

Check how precise your number is and whether you already know the reasons behind it.

For example all active customers or all employees.

Completed questionnaires or finished interviews.

The number (NPS, satisfaction)

NPS margin of error±18.6 points

Shares such as «satisfied»: ±9.3 %

High uncertaintyThe measured NPS can differ from the true value by up to 18.6 points. Use it as a rough indication, not as a KPI.

95 % confidence, worst case, randomly selected participants.

The reason (AI-moderated interviews)

Core themes mentioned100 %

Confirmed several times (at least 3×): 99 %

Patterns confirmedPractically all core themes have been mentioned, most of them confirmed several times. Further interviews sharpen the ranking and allow analyses per group.

Measured in three Rezendo surveys with 211 to 529 interviews, averaged over 1,000 random orders. Data up to 100 interviews. Across all seven surveys with at least 50 interviews, on average 91 % of the core themes were confirmed several times after 50 interviews.

Our assessment

Your number is still blurred. With this number of interviews the reasons behind it are on the table. In an intro call we show you what your data says and where interviews close the gap.

Book a free consultation

Method &
quality

AI analysis

Participants &
acceptance

Data protection &
security

Access &
effort

Ready for the reasons behind your figures?

Book an intro call