---
title: "We don't guess. We ask."
description: "As a UX research agency for chatbots and voicebots, we talk to your users and test every dialogue before it goes into development."
url: https://www.lautmaler.de/en/services/research
lang: "en"
alternate_de: "https://www.lautmaler.de/services/research"
---
[Services](https://www.lautmaler.de/en/services) Research

02 / Research

# We don't guess. We ask.

As a UX research agency for chatbots and voicebots, we talk to your users and test every dialogue before it goes into development.

[Discuss a project](https://www.lautmaler.de/en/contact)[See all services](https://www.lautmaler.de/en/services)

![Illustration: Nutzerinterview – zwei Frauen im Gespräch, dahinter Notizen und Auswertungen](https://www.lautmaler.de/uploads/Oc9bNmGb1iQQHAEwvVKPz-svc-research-scene.png)

Case study

## In practice

![Hochhausfassade mit dem Logo von managbl.ai](https://www.lautmaler.de/uploads/NBPsoCmUQG2gvu9ufzNtc-Teaser_Projekte_Managabl_AI.png)

[Discuss a project](https://www.lautmaler.de/en/contact)[View the MANAGBL.AI project](https://www.lautmaler.de/en/projects/managbl-ai)

MANAGBL.AI

/ AI phone assistant for property managers

Some errors you can't hear. You only see them in the ticket.

MANAGBL.AI rebuilt its phone assistant for property managers and brought us in for an independent review. We tested it the way callers experience it: 169 test calls, from everyday requests to noise, spelling and targeted attacks. We matched every call against the ticket that reaches the team. The result: 19 error classes, sorted by urgency. MANAGBL.AI is now optimising its system on that basis.

70%

prefer a human when in doubt. That's why we build bots that know when to hand over.

100x

is how much more an error costs after launch compared with one the test finds.

80%

of all features are rarely or never used. Without research you build past the need.

/ What's included

## What our chatbot UX research includes

We talk to people, analyze data, build personas and test. Before launch and after, because only in operation do you see how people talk to the bot.

### Qualitative & quantitative research

We talk to your users and observe them in their real context. Unlike a classic user research agency, we also analyze your bot's conversation logs and usage data. That shows us what people really need, not just what they say.

### Personas & user journeys

From the findings we build personas and user journey maps. They show who talks to the bot, about what, and where things get stuck. A shared reference point for the whole team.

### Usability & A/B testing

Usability test or A/B test, we test prototypes and dialogue variants with real users, moderated by us, before expensive wrong decisions reach development. We decide based on data, not gut feeling.

### Analysis & recommendations

We analyze the test results and tell you what needs to improve first, and why. You won't get raw data nobody can use from us.

### Research after launch

After launch we keep reading conversation logs and feedback. That way we spot patterns and new weak points as soon as they appear. We don't wait for the next report.

/ FAQ

## Frequently Asked Questions

Here you'll find the questions we're asked most often. Can't find what you're looking for? **Then just chat with our bot!**

[ChatLord](https://www.lautmaler.de/en/contact)

**At what point in a project do you start research?**

Right away, not just at the end. First findings already feed into the strategy phase, and from then on research accompanies every phase. For us, decisions are based on real insights, not assumptions. We've worked this way from Berlin since 2007, for companies such as Telekom, BVG and ING.

**What if we don't have any user data yet?**

No problem. That's what qualitative UX research methods are for. With interviews and contextual inquiry, meaning observation in everyday work, we understand your users' needs even without a data basis. The numbers come later, once the bot is running.

**How do you get from research to concrete design decisions?**

We condense the findings into personas and user journey maps. That's where it gets concrete: how a conversation starts, where it needs detours, when the bot is better off handing over to a person. We design for the people who actually use the bot.

**Does research end after launch?**

No. That's when the real data starts. We continuously analyze conversation logs and feedback and derive improvements from them. A bot is never “finished”. It gets better with every conversation.

**How do you measure whether our bot really works?**

Not by a single number, but by what fits the goal: containment rate, task completion, satisfaction, drop-offs. A bot that “handles” 90% but helps nobody is not a success. We measure real benefit, not vanity metrics.

**What happens to the results after a usability test?**

They don't end up in a report nobody reads. We turn them into clear, prioritized recommendations. What needs improving first, and why. These feed straight back into design and development.

## Do you know what your users really want?

![Conversational UX Designer & AI Architect - Dialogdesign, NLU Development, NLP, Implementierung, Computational Linguistics, Data Science, UX Design, UX Research, Social Media Marketing](https://www.lautmaler.de/uploads/6rqUDUPct18f01Rzj0Jzx-team-strip-3.webp)

Your contact

Katrin Jansen

/ Conversational UX Designer & AI Architect

[Contact](https://www.lautmaler.de/en/contact)
