Wolfram Alpha



This was a 36-hour project I worked on before I started my official work with Wolfram Research. Wolfram Alpha, one of the many products developed by Wolfram, aims to provide knowledge by collecting computable data and helping people make sense of it. According to the Wolfram, "Wolfram Alpha aims to bring expert-level knowledge and capabilities to the broadest possible range of people—spanning all professions and education levels". This goal of making data accessible is not an easy task by itself; neither is making it [data] accessible across diverse target audiences. Nonetheless, the team at Wolfram Research has done a phenomenal job at designing the Wolfram Alpha system. By interacting with the system, it is evident that there is a well-thought-out interaction experience. As I dissected through the Wolfram Alpha system in search for "issues" that could be addressed, I ended up appreciating the system more actually. However, with all designs, there are improvements that can always be made.

My goal was to gather data from people using the system and present insights for redesign. In this report:

  1. I start with a few predispositions/assumptions.
  2. I present the results from both the primary and secondary research.
  3. I present a few of the insights I extracted from the research.
  4. I present design suggestions based on the insights and usability data


We are continually consuming information. With the pervasive use of the Internet, mobile, and social networking services, our consumption of information is continuous, so is the production of this information. As more information is produced, the presentation of this information becomes valuable for any meaning to be extracted out of it. This leads to the second predisposition--too much information at once and information not laid out correctly is more difficult to process. If data is not well organized, then creating meaning from the data becomes more difficult. The last predisposition is that navigation systems help us interact with complex data. From physical to digital environments, navigation systems facilitate our access information.


I conducted primary research to gather data on how users actually interact with the Wolfram Alpha system. Over a period of 36 hours, I observed and collected data on 13 users. Each session with each user lasted about 15 minutes. In each session, I prompted the user to use Wolfram Alpha and to think out loud as they interacted with different parts of the system. The users ranged from computer science graduate students to middle school students and everything in between. 2 out of the 13 users had actually interacted with Wolfram Alpha before (had prior knowledge of the system).

The following is a summary of the most common observations derived from the primary research:

  1. All the users, except for the ones that had previous exposure to Wolfram Alpha, were unsure of the system’s purpose at first sight. Is it a search engine or what? One user left the site and went on Google to search for “Wolfram” in order to know what to do with the Wolfram search box. Also, users were not sure of the querying mechanism. Users mentioned feeling boxed in into a way of asking a question or searching in order to get relevant and/or accurate results.
  2. Some users mentioned that the system seems tailored for “really smart people”—scientists; math-lovers, and etc. One user mentioned, “I would think it is a medical system or something, because when I read computation knowledge engine, I don’t know. I would lose interest fast because it seems very technical.”
  3. No user read through the search results. Most of the users would read the first few lines of the input interpretation data and would scroll down to the bottom scanning through the numbers, charts, etc., and paying little to no attention to the rich data in the query results. Most users mentioned that at times there was so much data and numbers and it would be too overwhelming to read through. One user mentioned, “there are so many numbers, I don’t care.”
  4. Only one user was able to find the settings button in less than 8 seconds. And this was because this particular user’s screen was at a lower resolution than all the rest.
  5. All users wanted an option choose a custom background color as part of the settings menu.
  6. Two users disliked the fact that the system displays results regardless of the accuracy of the input interpretation. As one user mentioned, “it should ask me if I meant this, or that instead of giving me results I don’t want.”
  7. Two users attempted to interact with the timeline pieces, which are static as of now.
  8. Some users wanted more visuals (more photographs, video, etc.) in addition to charts and graphs.

In addition to primary research, I also conducted secondary research by examining different web services and their approach on designing for search, retrieval, and interaction with information--the most popular one being Google Search. I also looked at Ask.com, and different organizations' knowledge base sites.


To improve the experience of Wolfram Alpha, I combined and picked a few of the issues presented above and contracted down to the following insights:

  1. Introduce users to Wolfram Alpha. Visitors to the site, especially first time visitors, should be able to know what Wolfram Alpha is and how to use it by simply being on the home page. If a user wants to know more, this information should be visible and not more than a click away. Currently, more information is about 2 pages deep from the home page.
  2. Facilitate searching. Offering auto complete not only makes searching faster, but also makes learning the system easier. This is especially true for a system with a limited amount of very specific data as Wolfram Alpha.
  3. Facilitate accessing search results. Provide an easier mechanism for navigating through the data. Display a manageable amount of data (overview) with the details on demand.


The first part of the design suggestion applies to the main search page. The second part applies to the search results page.

Part 1

As mentioned in the insights, users should know what is Wolfram Alpha and how to use it by simply being on the home page. The system should be intuitive enough at first sight that a user knows what to query and what to expect on the results page. Introducing users to the “what” and “how” of Wolfram Alpha might be more of a marketing issue; however, interaction design has a role to play in this as well.
sketch 1

Currently, there is a tour on Wolfram Alpha that does a great job at introducing new users to the system; however, the tour is hidden under the “about “ section. Bringing this out to the home page will increase its visibility and use.
sketch 2
wireframe 1

The current design does a good job at showing a query example. However, this could be supplemented by offering autocomplete suggestions as the user is typing based on popular searches or on data stored in the system. Not only does this make searching easier and faster, but it also provides a user with further context and an idea of the format for data input and output. See image above.

Part 2

On the search results page, I suggest a tree-style/vertical navigation system for the different query use cases. For example, the query “alpha” can be a character, company, Internet domain, unit, city, and etc. This information takes a user longer to parse when presented in a string/sentence that a user has to read. Instead it could be put in a vertical (or even horizontal) navigation system that the user can simply interact with to switch between different query use cases. See figure 3.

Minimizing the amount of data that is shown to the user is another design concept I suggest. As presented in the primary research, most users were overwhelmed with the data and did not read through it. I suggest collapsing all but a few (3-4) boxes and allowing the user to expand/maximize only the data he/she chooses to read. See figure below.
sketch 3
wireframe 2

The image below show some of the recent implementations of the Wolfram Alpha home page. Although I don't believe these were implemented as a result of my recommendations, they do validate some.

implemented version 2