Previously, we talked about data that a company can collect through web or financial transactions. Now we'll cover data you can obtain by asking your customers for their opinions, which we'll call solicited data.
Why do we solicit data?
- Solicited data can be used to create marketing collateral, such as a post about what percentage of your customers are satisfied with your products.
- It can also be used to de-risk decision making process, such as when we survey users to gauge interest in a new product.
- Finally, it can be used to monitor quality, such as when BizTech asks you guys to rate this session.
Types of Solicited data
Common types of solicited data includes surveys, customer reviews, in-app questionnaires, and focus group discussions.
This image shows a very common type of solicited feedback: Net Promoter Score, or NPS, which asks how likely a user is to recommend a product to a friend or colleague.
Solicited data can be qualitative, such as conversations and open-ended questions, or quantitative, such as multiple choice questions or rating scales.
Qualitative data is very subjective and requires a lot of analysis. Quantitative data can be easily summarised in a graph or chart.
In general, collection of small-scale qualitative data is good for generating hypotheses. For example, a focus group might provide some ideas about what features we might want to build.
Larger scale quantitative collections is needed to validate these hypotheses. For example, we can ask users to rank a list of features from most desirable to least desirable.
Revealed and stated preferences
It's important to remember that solicited data generally tells us our user's stated preferences. A stated preference is what someone tells us they want to believe, and is somewhat hypothetical.
When a user actually takes action, such as purchasing a product, we learn their revealed preference. We hope that our users' stated preferences are good indicators of their revealed preferences, but that isn't always the case.
Many people have a stated preference to go to the gym frequently. However, many of the same people's revealed preference is only to go occasionally. Some gyms have an entire business model based off of the expected difference between people's stated and revealed preference for exercise.
Now that we know some types of solicited data, and are aware of the pitfalls of revealed and stated preferences, let's go over some best practices.
- First, we should try to be as specific as possible when asking questions. This specificity should apply to both the wording of the question and the potential answer choices that we give.
- Next, we should avoid loaded language, especially if it might bias respondents toward a particular choice. For example, avoid using adjectives to describe possible choices and try to be objective.
- Whenever possible, calibrate your survey by comparing to known quantities. For example, rather than asking a respondent if they are interested in a new product, ask them to compare their interest in that product to their interest in one of your current offerings or the offerings of a known competitor.
- Finally, it can be tempting to ask as many questions as possible. Fight that instinct and instead ensure that every question you ask will help you take a decisive action. For example, if customers indicated a preference for Feature A over Feature B, you should commit to building feature A first.