Creating a UX Data Model to Guide Product Design

The increasing number of data sources that designers and researchers have at their fingertips has made the creation of a UX Data Model a key early step in the product design process. For maximum impact, the UX Data Model should be created before any design concepts, prototypes, or testing takes place. The UX STRAT conference has had many presentations on this topic, so a substantial portion of my UX Strategy Workshop (https://bit.ly/3IeHiPh) is dedicated to the process of creating a UX Data Model. The following is a summary of the process I teach in the workshop.

1. Meet with Stakeholders to Understand the Scope

Obvious, yes, but there are some facts you should gather during these initial meetings that may be less intuitive. After you understand the brief and expected direction of the product, ask about the anticipated impact of the product on specific business KPI's. This will allow you later to create design KPI's that align with the anticipated business outcomes. Another very important step not to miss: Ask for a rough estimate of the overall potential financial reward that would result if the product design is very successful, and also the risks that could arise from a less successful product. I'll explain why the rewards and risks are so important later in the article.

2. Determine the Questions that Research Should Answer

Many researchers apply the methods they know to every design problem they encounter. This reminds me of my neighbors Raymond and Robert when I was growing up. They fixed cars in their front yard, and it seemed that most of the repairs their cars needed involved a hammer, because they didn't have a very big tool kit. They were banging away all day, hoping for the best.

Instead, the methods selected should be based on the questions that need to be answered in order to create a strong product design strategy, and the confidence level required before a strategy and design direction can be determined. Some initial questions may include: What are the main dis-satisfiers in the current experience? What kinds of capabilities would drive adoption of the product by a particular behavioral segment? What product capabilities will move the needle in terms of the business KPI's?

3. Establish the Scope of Research and Data Analysis

Bryan's Law: The Rigor (R', aka time, resources, expertise) that should be applied to any given design problem should be proportional to the potential Rewards (R'') and Risks (R''') expected from the resulting product design.

For low risk/low reward design problems (measured in thousands), don't expend a lot of resources. For high risk/high reward design problems (measured in millions), create a very robust set of methods and UX Data Model, with substantial resource allocation to getting the UX right.

4. Determine Research and Analysis Methods

As I mentioned, the methods used should be based on the questions you are trying to answer. And the scope should be based on the Rewards and Risks of designing a successful product. Quantitative methods should answer questions that depend on relative size, prevalence, preferences, impact, etc. Qualitative methods should focus on discovery and quick evaluative studies, i.e. things you don't understand or don't know enough about to formulate a quantitative approach. Machine learning is quantitative, but different enough from traditional statistically significant research methods to merit separate consideration. Machine learning is helpful for finding patterns much more reliably than other methods in specific situations, especially suited for large scale products when there is no reliable initial hypothesis to base other methods on.

5. Create the UX Data Model

The Data Model should include the following:

a. Data sources (Quant, Qual, ML, etc.)

b. Plan for analyzing the different data streams

c. Approach for synthesizing results, combining them in ways that answer the research questions

d. Relative weighting for each data source

e. How the UX Data Model will be used to guide product design strategy (steps, roles and responsibilities, ideation, prioritization, etc.)

6. Create Design KPI's

Create a list of design KPI's that align with the business KPI's identified earlier. This list will be the basis of a scorecard for the product design strategy. For example, if a business KPI is the number of new loyalty program members, a design KPI might be drop-off rate for the loyalty sign up path.

7. Connect the Product Design Strategy to the Data Model

Item 5e above is about planning how the UX Data Model will impact the product design strategy. However, executing the plan is different, particularly as people with a stake oppose the results because they don't agree with the direction. There should be a clear connection between the data and the design strategy, not an indirect fuzzy one. This sometimes takes quite a bit of effort after the Data Model is complete. Obviously the final product design will be iterative throughout the design and development process, and the UX Data Model will probably evolve as well. But the more solid the connection between the Data Model and business KPI's is, the more value it will have into execution, release, and beyond.