top of page
Artboard 1_3x.png

Designing for Data Analytics Teams

When DemystData started in 2010, it was a data consulting business built upon a simple premise - the universe of data is expanding and every business needs to access the right data quickly and safely. As the business grew and expanded, so did the ecosystem and technology available to data scientists as part of their everyday toolkit.

Marketplace Preview.png


 When I joined the team 2018, the mandate was to take a consulting business and build it into a self-serve SaaS platform. In order to keep scaling the business, the Demyst team knew that it was important to give customers more self serve options to explore and access data while the consulting team focused on helping clients tackle really tough business challenges.


A Radical Redesign

In just a few years since its inception in 2010, DemystData transformed from a data consultancy building models for banks and insurance companies, to a global SaaS platform. By 2018, Demyst had on-boarded over 100 data products and was delivering data solutions to some of the largest international banks, insurance companies and consulting firms.

The Demyst Labs platform — designed in 2017, was struggling to scale alongside the rapid growth of the company and the enterprise demands of its customers. User growth was stagnant. Divergent features competed for focus. Basic usability was challenged.

The Labs platform had become a tool optimized for our team, not our users.

The Evolution of the Demyst platform from 2017 to 2019

The Challenge

Designing for Scale

Our goal for the project was to redesign our Labs platform to help teams intuitively discover, access and manage external data.

The original premise seemed simple: one click to access data. However, we weren’t trying to simply become another third-party data vendor. Our ambitions were to create a strong foundation for a product that would grow with our rapidly evolving enterprise needs, while expanding our marketplace offerings.

Our high level goals were to:

  1. Make data access fast and easy for data scientists everywhere.

  2. Give data officers more control over their information security and data spend.

  3. Create a platform for exploration and deep levels of engagement.

My Role

I led the product and content strategy, market research, and usability testing for the Demyst Labs platform from June 2018 to August 2019 and collaborated with two other designers to create the Admin page, Catalog Search and API Manager features.

In addition, I led the end to end UI/UX design of the Attribute Search Experience, collaborating with a data strategist, marketing manager, and VP of product. 

I stopped working on the project after the closed beta launch of the Attribute Search Experience in August 2019.


Retracing Our Steps

When I joined Demyst as a product manager, it was important for me to understand the legacy of the product, including past user research, established personas, the history behind existing design decisions, and general product insight passed from the early days of Labs. 


The MVP of the Labs platform was still being used primarily by our internal team to help automate some of their work, but had never actually been exposed to many clients.

Considering this problem and acknowledging our users’ current “hack” solutions, which included lots of offline data manipulation, we wanted to design a more robust solution that would allow data scientists and business analysts to easily collaborate on files, share favorite products, and manage data security. 


Early Insights

We tested the existing Labs platform with 13 data scientists and business analysts from our client base. Our goals were to understand the challenges data owners and buyers faced and the workarounds they employed.

We found that status quo for most data scientists is that it takes 7-16 months and up to $200k per source to discover, onboard, and test a single data product.

We boiled down what we knew so far into 3 main questions:​

  • Discovery: how do I stay on top of the emerging ecosystem?

  • Access: how do I justify the internal friction before the value is known? 

  • Risk: how do I manage governance across 100s of products?

The Discovery

You have to keep innovating to do what you did a year ago.

Demyst User Insights.jpg

Research synthesis exercise we conducted to group our research into common themes

When we dug into our user research, I was surprised by the depth of insights we found. I expected to uncover platform related annoyances that would give us clear insights on what to build next. 


But after synthesizing all of our research, it became apparent that our users had very fragmented and often competing needs that simply couldn’t be solved with the product we had at the time. Data scientists were more frustrated with the experiences they had offline. As external data became more integral to their business, their expectations evolved. Not only were they leaving our platform to analyze their data, but it was not uncommon for people on the same team to run the same analysis more than once.

“The biggest competition is a client’s internal capabilities so you have to outdo them to provide some distinctive value.”


If power users with a deep understanding of data and tech literacy were having trouble finding value in our platform, how confusing was the user experience for non-technical data buyers or new to market startups with challenging technological and financial constraints? 


It became clear that we needed to understand more about our user’s motivations beyond simply accessing data. 

Deeper Insights

Before we started designing, it was important to define success and understand the data access funnel at scale.


Prior to the redesign of the Labs platform, the number of API transactions made for a single data source was the only proxy we had to measure user activity.


As a team, we sat down and unraveled the concept of the data access. We decided to track activity on the dimensions of discovery, access and analytics.


To investigate the health of our data access funnel, I set up an analytics tool that would allow us to track user events on the platform. This was a critical step to allowing us to capture user intent and the areas where they were getting stuck. We were also able to use the event tracking framework to rapidly test new features and designs. 


I also worked with a developer to set up dynamic URLs that would allow us to capture search and filter activity on our catalog.

Search Taxonomy.png

There is no one size fits all solution for external data.

Digging into the data revealed some big insights into the user experience. We saw a lot of activity at the top of the funnel during the search experience, and then a significant dropoff in data access. After doing some more research, we found that almost all data appends involved some extra coordination effort between the end user and someone on our managed service team. This data showed that the experience was hardly the ‘one-click access’ we were striving towards.

The time and energy spent on providing additional analysis and statistics to users was having an impact on the business bottom line. The time spent on each adhoc request had downstream impacts on higher revenue projects and on platform adoption. 


Users are not getting enough value upfront.

We were requiring too much commitment from our users by asking them to sign up and upload their own data before showing them the full capabilities of our product.


There are feature gaps that users are fulfilling outside the platform.

We needed a way to keep users more engaged during the data exploration phase because there were too many steps that have to occur offline (i.e. downloading data).


Valuable features are fragmented and hard to find.

Users need tools to facilitate analysis and review of data in the tool they are most comfortable with. The Demyst experience should feel comprehensive end to end, rather than fragmented solutions.


Need a scaled down version of our product for non-enterprise users.

We had many tools that were valuable for Enterprise organizations but require too many steps to access data for a regular Marketplace user.

Reframing the Problem

Fragmented features and touchpoints prevent users from finding the information they need quickly and efficiently

The Labs platform exacerbated the frustrations many users were having with data access. Problematic user experiences were driven by a lack of a clearly stated value proposition, disjointed feature sets, ambiguous product information and inefficient mechanisms of actually accessing data, which causes confusion. Ancillary communication and additional effort is required from our customer success team to help users, which leads to frustration and wasted time.

“ might we expose more information about data products up front?”


This begged the question, how might we productize the deep knowledge our customer success team leveraged every day when helping customers? Our proposal was the Demyst Marketplace, a suite of products that would each focus on one of our key offerings. We would also build a public website that would be a single point of entry for our product offerings and would help educate users to help them pick the best solution for their needs.

The Labs Redesign

Introducing the Demyst Marketplace

DemystData is a one stop shop for data discovery, access and governance. A single, united product vision to help deliver a seamless user experience. With Labs, we built a searchable catalogue with a real-time interface to samples, dictionaries and fill rates, accessible immediately under Demyst infosec and contracts within 4 weeks. 

We also expanded our offerings to Python, allowing data scientists to access and analyze data in their working environment, all without having to leave the Demyst platform. 

Lastly, we built a stronger than ever enterprise platform optimized for data governance and security. 

All this was united in our Public Marketplace, turning our static website into a powerful marketing tool. 

Demyst Marketplace.png

Final screens from the Demyst Labs redesign

The Journey

One Click to Access Data

Early on, it was important to understand the different factors that may influence the user’s end to end experience.


We set the following goals to help guide our product vision:

  • Wide Distribution: Available to community of 100K citizen data scientists across integrated platforms 

  • Deep, Unique Data: 200 full data sources with rich metadata and performance statistics

  • Data Content Newsletter: 1 post per week, to a distribution of 50K people

  • Growth: 5,000 MAUs

  • Usability: < 2 minutes from landing page to data

Multi-channel Product Map (1).jpg

This information architecture diagram helped us understand the number of layers a user had to go through to access valuable information and plan where new features should live.

A Universal Design

The existing Labs platform was poorly designed for users who weren’t large enterprises with a dedicated account manager to help them through the intricacies of data access. To move beyond the existing framework, we decided to separate our product into 3 distinct offerings that would each serve a specific purpose.

Slow and Steady

It was important to design in a way that would help the product grow while leaving current users’ workflows uninterrupted for the most part. We decided to design incrementally to avoid putting off existing users, and tried to introduce our design solution seamlessly into the current enterprise platform.

As we began ideation, four key design challenges emerged:

  • How might we help our users gain a competitive advantage through shared insights?

  • How might we allow teams to test data and collaborate in a secure environment?

  • How might we expose valuable information about datasets early in the user journey? 

  • How might we create more self serve features that would alleviate the burden on our managed service team?

Understanding the User Journey

From Disparate to United

We created an experience map as a tool to understand the gaps and opportunities in our end-to-end solution. This exercise helped us ensure that the features we are building out in each channel are cohesive and serve a specific purpose. 

The public catalog is for powerful searching and information gathering. The Enterprise web app and Python package are for data enrichment.

Based on these insights, we decided to focus on these key ideas:

  • Centralize and link all of the available channels to access data in the public catalog

  • Optimize our web app for data enrichment purposes over search purposes

  • Prioritize tools that allow users to access data they need quickly, without uploading their own data

Multi-channel Product Map.jpg

The user experience map helped us align on different channels users could interface with during their end to end journey. 

To educate the team on our vision, I created a feature comparison map to help lock down the purpose and functionality of each of our product offerings.

Multi-channel Product Map (2).jpg

Strengthening the Enterprise Toolkit

Quick Search

Some of the most frequent questions resolved by our customer success team revolved around choosing the best data source for a given use case. 

With our updated search capabilities, searching through all the external data products through our web interface became easier than ever. Whether you know the data product, specific attribute, or just general category you’re looking for, just start typing a key word or phrase and we’ll do the rest. The catalog search instantly shows the most relevant results per category, and refreshes them whenever you change your search term. Now you can focus on the most relevant results by product, attribute, and tag and expand any category where you want to see more results.

Some of the key design decisions included:

  • Grouping search results by categories

  • Adding icons that would help users quickly identify the category

  • Displaying the top 3 results per category to help users filter for the most relevant result (9).gif

A demo of the Quick Search functionality

Admin Tools

An interesting challenge that comes with selling a SaaS product is the dichotomy between the buyer and user of our app. For enterprise companies, compliance and visibility is a key goal for data buyers. 


We updated our admin page to give admins an easier and more secure way to manage users and APIs, configure organization preferences, and monitor user activity. The redesign gave admins more control over their organization and how they want to manage their team’s activity end to end.


To come up with the dashboard design, I did extensive interviews with our admin users and managed service team, and pored over our inbound requests. There were several themes that popped up over and over again:

  • How can we track which users have been active on the platform?

  • How much money are we spending on data?

  • How many API transactions are we running?

  • How can we audit user events on the platform?


To address these questions, I worked together with our backend engineering team to understand what elements we could incorporate from our existing platform and what data we could pull from the external tracking tools we had set up. We also had to ensure that any data visualizations worked with the MDBootstrap component library we were using at the time.

Admin Dashboard-page-001.jpg

Final concept for the Admin dashboard

Python Packages

As product manager of the Labs platform, I worked closely with our API product manager to build a tool that would closely reflect our existing capabilities in the web app. 

Our Python package became an invaluable tool to help us iterate and test new capabilities that would require more effort to build into the web app. Behind the scenes, as we worked on expanding the catalog we were also working on improving the metadata we were collecting about our products. 


Collaborating closely with our head of data and senior engineers, I came up with a set of files that would could use to run panel tests against our integrated products. We also used our user research to determine what statistics to calculate off the back of these tests. 


We found that once users had enriched their file with external data, they wanted to see the stats around them. They could do this easily through a single method, report(), from our Python API. (4).gif (5).gif
Screen Shot 2019-10-19 at 2.44.08 PM.png

The Public Catalog

Getting access to data should be easy. The Demyst Marketplace allows you to explore hundreds of data sets at no cost, and now you can access sample data for each external data product, or request sample data with the click of a button.


We know that determining the most relevant data for your enrichment is a challenging task, and we want to make sure you are armed with all the information you need to make the right decisions. It’s important to get real data into your hands, so the sample data we provide showcases real attributes returned against DemystData’s representative population sample. This way, you can compare data products according to your business needs and best practices. And the best part is, you can access all of these insights with no cost, contracts, or uploading your first party data.


The biggest piece of the puzzle was building the public data catalog. From a technical standpoint, we were building an entirely new platform from scratch -- one that was lean and fast, decoupled from our legacy code.


From a UX perspective, we were opening the floodgates to a whole new set of users. The opportunity here was tremendous. The public marketplace was where we could educate our users and showcase the value of our product. It was also where we could optimize for SEO to drive new traffic to our platform.


Early on, we went through several design iterations as we tried to pin down the best way to the information we displayed to users. Some of our key design decisions included:

  • Ensuring the catalog was responsive and mobile friendly

  • Replacing file downloads with data viewers

  • Focusing on targeted CTAs to capture user intent

Marketplace Group.png

Final screens from the Demyst Public Catalog (Product Detail Page on the Left, Sample Data Viewer on the Right)

Positioning the Product

With our new product suite coming together, the final step was united under one brand and product vision. I worked with our CEO and sales team to craft a new homepage that would clearly define who we are and what we do. 

I broke down each product to a feature level and tied that to our pricing model, to help users understand exactly what they would be getting are part of their subscription. This helped enable our sales team and reduced our inbound significantly.

Final screens from the Demyst Public Catalog (Product Detail Page on the Left, Sample Data Viewer on the Right)

Website Group.png

Screenshots from the Demyst website with updated design, copy, and product positioning.

The Launch

The Most Radical Update Since 2017

On April 12, 2019, the Demyst Marketplace launched globally. It was a major achievement by the team, considering that it was a complete redesign and rebuild from scratch. The new platform was leaner, faster, and featured a public data catalog, hosted python notebooks, and 100s of new data products. 

This was the culmination of every initiative we were working on, contributing to our vision of external data driving improved decision making, processes, and compliance.

In the 4 months after the launch of the Demyst Marketplace, the team continued to tweak designs and workshop additional features. We also continued to collect user feedback that would shape the future vision for the Attribute Search experience. 

The Impact

Positive Results and Next Steps

The redesign of the Demyst Labs platform and the introduction of the Marketplace has had a positive impact on the Demyst user experience. In the 12 months since we started the platform redesign, we had increased our organic search traffic as well as discovery usage. 


Enabling users in Python and using targeted CTAs to capture user intent allowed us to better enable our sales team, increase transactions and empowered us to pursue partnerships with leading AI and Machine Learning platforms like DataRobot and Snowflake. 


New user sign-ups increased by 300%

Organic traffic increased by 200%

Conversion rates increased by 2%


However, touchpoints between users and our customer success team had not significantly decreased. The redesign of the platform opened the gates to new types of users with unique challenges. 


If you want to learn more about how we addressed these challenges, read about the pilot launch of the Attribute Search Experience.

More Case Studies

bottom of page