Where data & decision-making intersect.
The design challenge!
Redesign and integrate Periscope Data’s B.I platform into the Sisense product. Design a new tool that allows data analysts to create complex multi cell analysis using markup, text, and code (SQL, Python and R) in a single document.
Sisense is a business intelligence (BI) solution that provides advanced tools to manage and support business data with analytics, visuals and reporting. The solution allows businesses to analyze big datasets and generate relevant business trends for them. Sisense allows businesses to combine data from many sources and join them into a single database. Sisense’s customers include General Electric, Nasdaq, Philips, Spotify, & Wix.
Sisense Acquires Periscope Data.
Periscope Data joined forces with Sisense in a merger that makes it possible to deliver an end-to-end BI and analytics. The merger now gives data science teams access to a cloud platform with capabilities like cloud-native data integration, advanced analysis with SQL, R, and Python, and integration with production cloud machine learning systems.
Periscope Data is recognized as a “game-changer” by Next Wave Business Intelligence and the 4th fastest-growing company in North America by Deloitte. The merger gives Sisense over $100M in annual revenue with a global team of 700 data and analytics experts
See the Techcrunch article
The Market Size
By 2020, 90% of business professionals and enterprise analytics say data and analytics are key to their organizations digital transformation initiatives. According to a recent research study, approximately, 58% of organizations worldwide plan to adopt big data technology in 2018. The organizations will adopt hybrid IT infrastructure management capabilities. The growing adoption of big data and AI in industries including IT & Telecom, BFSI, and Healthcare among others is further fueling the demand of the big data analytics market
As a lead designer at Sisense, I led the design and product initiatives to integrate periscope data into the Sisense platform. I defined the user experience, interactions, information architecture, wireframes, visual designs and user research to validate the product direction and user experience.
Gathering business goals
The biggest differentiator about the Sisense for Cloud Data Teams product is the ability for businesses to be able to turnaround core reporting and answer complex questions in their first hours of using the product. The ability to write SQL, Python and R against modeled and unmodeled data allows Data Analysts and Data Scientists to rapidly explore, transform and visualize data with the flexibility of code, and leverage advanced statistical techniques to create analyses of a higher caliber and extend from descriptive to predictive insights.
- Merge Sisense & Periscope Data.
- Offer different packages to meet demand.
- Unify the product experience.
- Strengthen Sisense as a SaaS cloud company.
- Increased adoption of cloud data warehouses.
- Increased hiring demand for data analysts & data scientists.
- Increased investment in analytics.
Knowing Our Users
Sisense for Cloud Data Teams (previously Periscope Data) serves 3 major customers: the data engineer, data analyst and the business user. Each play an important role within a company.
Builds data pipelines that transform raw, unstructured data into formats data scientists can use for analysis. They are responsible for creating and maintaining the analytics infrastructure that enables almost every other data function.
Is responsible for bridging the gap between IT and the business, determining requirements and deliver data-driven recommendations and reports to executives and stakeholders.
Is someone who examines information using analysis tools and uses raw data to help their employers or clients make important decisions by identifying various facts and trends. They transform business questions into reports through the use of code to drive visualizations that uncovers ‘why’ something is happening.
Who's Our Target Persona?
The Data Analyst
Why is the Data analysts important for us to target?
- Data analysts hold the buying power over the tools they use.
- Data analysts create all the reporting within a company.
- Data analysts transform large complex business questions into a tangible result.
The Data Analyst’s Journey
After interviewing several data analysts and discovering the many different tasks involved in the analysis process to create dashboards & live reports; I found that reporting is about 60% of their overall duty. A major task that takes 40% of an analysts time is answering 1 off questions related to a report delivered to a business user. This particular task is called ‘Ad hoc analysis’ which is any kind of analysis digging deeper into 1 specific area of a live report or dashboard, to obtain more details about it for a single use. Many times, ad hoc analysis is done in response to an event, such as a sudden dip in production or loss of customers where a business user needs to understand why this sudden dip in production is happening. An analysts may create a report that does not already exist or drill deeper into a static report to get details about accounts, transactions or records on the spot. Prior to Sisense acquiring Periscope Data, this user need or problem was not solved until now.
Along the user journey, data analyst encounters many pain points making reporting and ad hoc analysis difficult. Below are a few examples of what a data analyst goes through.
- Finding the right data.
- Fishing through many databases.
- Cleaning the data.
- Translating a business question into quantitative results.
- Building a hypothesis with code
- Small area to write code
- Losing track of what was tried
- The trial & error process
The Problem Statement
After going through the user journey and validating it with serval users, I developed a problem statement to focus both design and product towards one goal.
Analysts are forced to develop complex analysis in confined areas, making the iteration process of writing and testing code to drive business decisions slow and frustrating.
The design phase…where everything comes together!
In order to develop this new product that touches upon user & business needs, the design process uses heavy amounts of design iterations, user feedback and testing to validate our direction, because of this I used the double diamond approach to both discover needs and define the design.
Design Strategy & Research Plan
To kick start things, I developed a 12 week plan to start design and research. The 12 weeks will uncover user needs, desirability & market fit through user research and design iterations. User testing sessions involve a series of moderated testing, 1-on-1 user interviews and surveys to assess the customer’s needs, understand their pain points and areas of frictions. The focus for Design is to discover areas of improvements in the overall experience and to develop alternative design iterations, generate new ideas, and to validate assumptions with high & low fidelity UI designs.
Research methods used
- Moderated remote testing
- 1-on-1 user interviews
- A/B testing
- Concept sketching
- Low-fidelity design mocks
- High-fidelity design mocks
Developing a How Might We Statement
I created a How Might We statement to focus the design efforts into a question to be solved, this turns the problems into opportunities for generative thinking on how to best solve the problem discovered during my user research.
How Might We…
Empower analysts to develop and iterate upon many ideas using code in 1 environment to answer complex questions and tell compelling data stories through visualizations to drive business decisions?
Designing Around Our Areas Of Opportunities & H.M.W
After reviewing the analyst’s journey, I identified a few areas in which we can enhance their workflow and create a better experience. These areas of opportunities were then validated with end-users to further enhance their process. Going forward, I used our ‘areas of opportunities’ and the ‘how might we statement’ as a bases for developing designs. The wireframes below is the first phase in my design process and used to understand user needs.
Enhance the iteration process & help develop faster analysis.
Make complex code based easier to understand.
Break analysis into smaller bits & reference different areas of code.
Tell rich data stories with charts, images, videos & text.
Designing The Wireframes
The first set of wireframes are based on opportunity #1 (Enhance the iteration process & help develop faster analysis) using the Periscope look & feel. At this point in time, both Sisense and Periscope Data did not have a clear direction for how the product should be merged. Because of this ambiguity, I used the Periscope look & feel to jumpstart design and the testing phase. The wireframes illustrate the concept of adding multiple charting blocks allowing data analysts to see multiple views of their data. As a product group, we assumed this ability would aid analysts in developing faster analysis but our assumption proved to be wrong.
After a few rounds of customer feedback, the wireframes prove to be a step forward in the right direction, analysts gave positive feedback and wanted to have multiple kinds of visualizations to help expand their view of the data, but our customers wanted more. In order to answer a business questions through data, our customers need the ability to make many iterations of a data set, to write and run code and have the ability to rewrite and test their dataset in 1 environment. The iteration of a codebase or dataset is the highest and most valuable area to pursue going forward.
2nd Round of Wireframes
Taking the feedback from the last set of wireframes, the main focus here is the ability to add many blocks to enhance the code iteration process. These blocks give analysts a way to brake their code into smaller parts and to test their results in 1 area. The wireframes illustrate the ability to 1) create code blocks, 2) write code within a block and 3) run the code block to see the results. Each code block can be ran independently from one another, giving the analyst the power to see and understand how their dataset progressed into a finalized outcome. In addition, blocks can be used for text, images and different forms of visualizations to create a narrative around a dataset. Our customers responded very positively to this new direction of adding content blocks, and proved to be an important key in laying out the foundation of this new product. This new ability of adding code blocks to test and iterate upon a dataset solves our user’s frustration of writing and testing large datasets found in traditional code editors and makes the iteration process easier and intuitive. By the end of this testing phase it was determined that this product should fit within the Sisense look and feel in order to promote the merger of Periscope Data into Sisense.
Starting Designs Based On Our System
After wrapping up testing wireframes with users, it was determined that Sisense’s look & feel will drive all new products to help promote a unified company. The next phase for design is to incorporate the Sisense look & feel while incorporating user feedback to further enhance the iteration process. Since Sisense focuses on nontechnical users to create live reports and dashboards, the look-&-feel promotes the idea of ‘friendliness’ and ‘simplicity’, it’s important to carry these same ideas into our new platform that focuses on technical users.
High Fidelity Design Iterations
Throughout this journey of understanding the needs of data analysts, the biggest value is the ability of adding multiple blocks to write and test analysis with code. In these designs, I tried a few different ways of adding blocks to discover a solution that promotes the code iteration process. From the wireframes, analysts expressed an interest in knowing where new blocks are added. This design uses a placeholder as an indicator of where new content will be added to the page. By clicking the ‘+ Code’ or ‘+ Markdown’ a new block appears in the placeholder area, giving a visual indication of where new content is added.
Data analyst, like many developers love maximizing space to code, and this design created less space to work with. Analyst did however preferred the bottom placement of the buttons; the placement gave analysts a visual cue that blocks could be added underneath other blocks. I used this bottom placement as a backbone for future design iterations while also maximizing the space in the editor for analysis.
Button Placement. Let’s Optimize It!
In this design, I focused on maximizing space for the purpose of writing more code while reducing the size of the buttons. Analyst expressed the importance of having ample space to code, the previous design took too much vertical space leaving little screen real estate for code. To solve the issue of space, I designed a series of floating buttons to add additional content, and reduced its size to give analysts more space for analysis. Analysts want and need the ability to add lots of content while having plenty of room dedicated for code.
After reviewing the designs with our customers, the smaller sized buttons help to maximize space of the editor but 2 out of 5 analysts struggled to locate them. During my interview sessions, I uncovered that the empty space around the buttons caused users to overlook them. In my next iteration, I focused my design efforts on better placement of the buttons and improved visibility.
Finding The Best Solution
In this design iteration, I placed the add more content buttons on the actual content blocks. After user-testing this method 5 out of 5 analysts were able to see the buttons and understood that new content is added to the bottom of an already existing block. Because of the linear progression of writing code from top down, the bottom placement on the content area allows analysts to see and create another block with ease. This new design is able to maximize the area for code while allowing analysts to see and create many blocks for an iterative approach to answering complex business questions. Analysts using this method can now brake large datasets into smaller pieces, run individual parts of a dataset to see if their hypothesis is correct, write new code and retest until they arrive to a final result. By giving the user the power to add multiple blocks became the key factor in solving our user’s need to have a tool that supports an iterative process of writing, running and retesting datasets.
The All New Sisense For Cloud Data Teams
Sisense for Cloud Data Teams allows businesses to connect to any cloud data source to explore their data and build powerful analysis quickly.
Add Multiple Content Blocks
Sisense For Cloud Data Teams brings a new kind of interactiveness and flexibility for a better code editing experience & data analysis process. Data analysts can now tackle large complex data by writing live code in individual blocks that can be ran independently, making it easier to experiment, explore and develop faster analysis.
Iterate On Ideas
Sisense For Cloud Data Teams allows analysts to experiment with their codebase to arrive at answers quickly. Analysts no longer have to write large datasets in one confined area. Instead, analysts can break their codebase into smaller areas and run each section of the code individually, change things and see the results of that change instantly. This process of experimentation allows analysts to build upon many ideas and arrive at a final hypothesis through a process of experimentation and iteration.
Want to learn more about this case study or others projects?
For details on my design process or interested in collaborating together, feel free to reach out.