Goals

Megaputer Intelligence needs to provide a lightweight text mining experience to general public, simplifying the features of company's key product PolyAnalyst​™.

But Why?

There IS a free demo for PolyAnalyst​™. Still, it requires a bit of background knowledge in data analytics to explore the full capability of the product. Thus, the company needed an easier, beginner-friendly interface to target broader range of clients. See the product concept below for more details.

Requirements

A set of desktop and mobile interface which users with no or minimal experience in text analytics can easily guide through.

Minimum viable feature includes 6 pages including:

1. Landing page

2. Analysis overview

3. Entities extracted from the user-provided text

4. Facts extracted from the text

5. Sentiments extracted from the text

6. Linguistic analysis of the text

Users

There are three main user groups for this product:​

1. General public with interest in data analysis but less background knowledge

2. Internal sales department responsible for publicizing the parent products online, or at offline events such as conferences

3. Potential clients (business representatives) quickly browsing through data analytics softwares

What did we learn from them?

1. While the existing parent products followed the dynamics of B2B, users of Sapremo™ expect B2C experience. The parent product interfaces are rather text-heavy and achromatic to target a specific group of users; data analysts. Thus, to make the experience more memorable and universal, expanded usage of image and color coding would be necessary.

2. Sapremo™ users are independent, meaning that they will explore the text analysis features without a training or assistance from the company’s internal data analysts. Therefore, the analytics features should be trimmed down and displayed comprehensible enough for college level audiences.

3. Sapremo™ users come from various domains, so the text they wish to analyze will also be diverse. The interface should cater to such wide range of the text, from technical writings to product reviews.

4. Not all users will have their own text to analyze. Some just want to test the capability of Sapremo™ with the lowest possible effort (minimal clicks). Therefore, we need to provide them a set of default texts to reduce their labor.

Wireframe

A sketch of general layout is a good place to start in order to apply and visualize the four key findings from user research. Below are the two sample wireframes developed during this phase.

The landing page addresses the effort of translating PolyAnalyst​™’s B2B dynamics into B2C experience. A no-scroll page that consists of three simple elements, header, body, and footer is more universal approach to a general users, compared to content heavy tabs, panels, and context menus which many text analysis software choose to employ. Such simplicity in the interface also recognizes how most Sapremo™ users may guide themselves through the product on their own.

The analysis overview page, proves our effort of considering users from various domains, and how they might want to see the capability of Sapremo™ in a glance. The right column of the body content, where the alternating row is, indicates a brief summary of analysis. Each row represents a result from different tabs (a.k.a. different analysis features for various domains of texts) so that the users can view the essence of the product without additional clicks.

Development

The next phase is to create a set of low-fidelity prototype with sample texts and images. In this stage, I was given a set of sample text from in-house linguists to adjust the arrangement of the interface and fine-tune the appearance of individual elements based on the amount of the text given. 

Text domain detection as a default feature was discussed at this stage for optimized analysis results, considering that the accuracy of sentiment analysis results is almost always lower, uncovered via quantitative research, when users do not specify the domain (context) of the original text. Note that the detection happens automatically, but users can also adjust the domain if needed.

Also, depending on whether the text is fact-based or sentiment-based, I added a small UI adjustment to mark which page includes the most meaningful analysis result for a user.

Final product

After discussing the low-fidelity prototype with the stakeholders and creating several drafts to reflect the feedback, I proceeded to develop high-fidelity prototype. Desirability and aesthetics (based on the company branding) has been the main concern in this stage; to apply the key findings from user research, use of image and color coding has been highly emphasized, differentiating Sapremo™ from the parent product.

Note that the previous images only capture part of the final product. The product consists of 42 frames of desktop interface and 40 frames of mobile interface.

The prototype also includes measurements and proportions of the interface—following the 4px grid system—based on the standard screen size. It serves two purposes, helping the stakeholders to understand general sizing of UI elements and allowing the front-end developers to take reference during the implementation.

Mobile version

Below are the samples of mobile interface, transplanted from the high-fidelity prototype of the desktop interface. Mobile interface measurements and proportions were adjusted from the desktop environment, considering the narrow and vertical screen proportion. Therefore, several text-heavy elements were revisited and altered to improve readability.

Adjustments during the implementation and testing

Although the final prototype got approved and implementation has started, there were several instances where stakeholders asked for fine-tuning due to implementation difficulties, usability issues, or unexpected use cases found during the testing. Below are the samples of such adjusted prototypes, delivered upon careful discussions between the developers and I.