OpenDialog Docs
OpenDialog 2023.1

User needs


A critical input to informed design is an understanding of the typical user. Who are they? What are they wants, their needs, their expectations? How and where do they expect to use the product? What's important to them? What's useful to them? What do they absolutely not want? The answers to these questions yield insights that the conversation designer can then apply in their design.
We can get insights into these questions through primary and secondary research.
Primary research refers to the process of getting input directly from users. This can be in the form of interviews, surveys, or observations to name a few. These methods allow us to validate the use case with the users, just like we validate the use case with the business owners. This type of research is conducted to answer specific research questions and is often used to collect data that is not available through secondary sources, or to validate or supplement existing data. Primary research is typically conducted by a User Experience (UX) Research professional, or someone who masters the skillset needed to conduct user (UX) research.
Secondary research (also called desk research) refers to the process of collecting and analyzing existing data such as customer service call logs, IVR logs, internet searches, market research reports, etc. The effectiveness of secondary research to get the sufficient insights varies greatly; in cases where a lot of pre-existing user-focused research is available, it can provide a very solid starting point. At other times, it merely manages to give a broad understanding of the topic and trends. In that case, it is an excellent basis to start primary research from.


Learning about the user must be done before the design phase starts, as it is an critical input to design.

Consider LLMs

LLMs can be used to generate secondary research insights, specifically trends. This information can be useful, however, at the time of writing, verifiable resources are not available in e.g. ChatGPT. An internet search may be a better alternative for in-depth research as it will yield a list of verifiable resources.

Don't use LLMs

Since the definition of primary research is to ask specific research questions, it should be clear that LLMs are not an option to learn what users want and need. If an attempt is made to use LLMs regardless, the LLM output will, at best, be vague and not applied to the specific context, brand and user.