Analytics is having a UX crisis.
Why user experience is crucial to analytics success
👋 Hello! Robert, CPO of Hyperquery and former data scientist + analyst. Welcome to Win With Data, where we talk weekly about maximizing the impact of data. This week’s post is our inaugural guest post from the one-and-only Matt Blasa. If you enjoyed this post, I’m sure he’d appreciate a like/share/follow. 🙂
I grew up in the 90s. And putting aside the more egregious missteps of our generation (beanie babies, pogs, and rollerblades), one of the defining characteristics of my generation was a ubiquitously terrible sense for UX (user experience). We had great excuses, of course – the Internet was nascent and our culture has simply not yet recognized the value of form, particularly when function alone seemed to suffice.
Fortunately for us, technology from the 90s didn’t stay unchanged: UX has attained some level of primacy over the last few decades. Recently, it’s become the standard flown by modern tools: Slack over Skype, Apple over Blackberry, Google over Yahoo, Spotify over Napster. Every modern innovation has been replaced by a service with the same functionality, but a better user experience.
That said, there’s one industry that Prometheus’s fire hasn’t yet reached: the data industry. And given data work is so interdisciplinary, this is baffling. The success of our analyses, our pipelines, our ML models is dependent on consumption by other people, yet our deliverables are embarrassingly devoid of any consideration to the experience of those involved. Let’s talk about some of the problems in our toolchain.
UX problems all the way down.Â
When folks think about analytics, they generally think dashboards, tools, or appealing interfaces, but these are the deliverables, not the objective. In reality, analytics is about cultivating data-driven conversation, not just dropping raw data on stakeholders. For analytics to be effective, therefore, it's crucial that everyone - business users, analysts, stakeholders - grasps the context of the narrative being told. Where mediocre analytics answers questions, great analytics enables tighter feedback loops that spur faster change.
The key to effective analytics, then, isn't just crude delivery of the data, but a focus on how seamlessly that data is created, shared, and consumed – that is, the user experience along every step of the way. And the problem? There are experiential blockers. Every. Step. Of. The. Way.
For instance, one of the first UX problems that we as analysts often create for stakeholders is our choice of deliverable – we too often share esoteric or context-poor shareables, expecting our stakeholders to do the heavy lifting to interpret the data themselves. As a new data analyst, I loved analyzing data, and it was this love that made me nearly fall into this trap. Fortunately, I had prior experience that helped me avoid focusing too heavily on the technical work. Prior to my career in data I had been a marketing strategist, so I’d developed a keen bias towards presenting insights and contextualized knowledge, not just raw answers.
But even with this solved, I noticed a gap – I knew how I was supposed to work, but it started to feel unnecessarily difficult to do what I was supposed to do: present insights.
This difficulty didn’t come because my team lacked clean data or due to the quality of our storytelling. Far from it. We had great presentation skills and knowledgeable analysts, engineers, and product managers.
The real issue was workflow inefficiency. Too much time was spent switching tools, fumbling around with shareable links, and, consequently, transferring context manually in meetings, leading to longer, more expensive projects.
And this brings me to our second UX hurdle: our tools were designed for tech users, not business users. We had data-rich notebooks, but they meant little to non-technical team members. Ironically, these tools and notebooks, meant to democratize data, were widening the communication gap. Work was not only difficult for stakeholders to follow, but difficult for us to even share.
The power of solid user-focused analytics tools lies in their ability to unify. You need to connect diverse groups - from analytics teams and business users to decision-makers. Their strength is in providing immediate context, communicating simply and swiftly. Moreover, analysts will spend less time wrestling with functionality and more time analyzing data. This makes the essence of context easier to understand — which drives faster decision making and feedback loops.
A rubric for good analytics tooling UX.Â
So what does optimal analytics UX look like? Having used a number of different analytics tools, I’ve found there are four factors that consistently lead to greater consolidation and communication of analytics work:
Ease of creation. Analytics processes ought to be accessible to users with varying abilities. It means taking the intimidation factor out, and cultivating experiential learning and community. Good analytics initiatives grow by democratizing data access, But great ones simplify the experience of using data, making it easy to focus on what matters - insights.
Intuitive design. A tool that's easy to navigate makes the sharing process a breeze for analytics teams. Primary users should be able to dive in, without spending endless time trying to decipher complex and twisted documentation. Clear labels, logical workflows, and consistent design elements are non-negotiable.
Centralization and collaboration features . Tools should unite diverse user experiences, which can go a long way in speeding up analytics processes. When your work is silo’d, communication pathways get blocked. Collaboration and centralization features make that process smoother - everyone involved in ingesting analytics work can leverage a centralized source of truth to access resources and communicate.
Customizability. Not every user is the same, and communications styles differ. Even within the same company or industry. Being able to cater analytics experiences to the stakeholder improves communicationand rapport. Being able to communicate insights often isn’t as important as how you communicate them.
These might seem like small tweaks, but they're massive as they add up. By seeking toolchains that satisfy these criteria, we can make the analytics process not just more multi user-friendly, but more efficient and effective. This paves the way for not only more effectively communicated insights, but better decisions, and, consequently, the trust and credibility that fuels a thriving data culture.
Why we should care
Good UX Brings Builders and Users Together
Well-designed tools make loops. A great tool can bring analysts, business users, and stakeholders into one place. The analysts are able to build their insights and code in a way that is natively shareable. Business users and stakeholders are easily able to understand the context of the code and explore it. A well-designed tool can reduce friction and bring organizational attention to what matters:Â the underlying insights.
Roughly speaking, I’ve found that there are three levels of collaborative maturity:
Level 1: Data Transfer. This is where most organizations get their start. Think interactive dashboards, offering data exploration and drill-downs. But this level hits a big wall: restricted data manipulation and rigid exploration paths. Business users can't venture beyond what analysts created, and questions answered are often not particularly valuable.
Level 2: Knowledge Transfer. The gap between users and analysts gets narrower. Think notebooks or dashboarding tools that allow some data modeling and deeper customization capabilities, and consequently, these tools produce more careful alignment between insight and stakeholder needs. Users are able to flexibly manipulate data, use querying or programming languages, and handle the data pre-processing steps needed to answer precise business needs.Â
Level 3: Embedded Analytics. This is the holy grail of analytics collaboration. We're talking about seamless switching between tech-focused and business-focused interfaces. Analysts are empowered to query, build, preprocess data, while switching between languages and data sources with ease. But at the same time, business users can be deeply involved from the get-go, ask analysts real-time questions, and easily integrate documentation and notes. This level helps enhance communication efficiency.
Each level serves as a bridge, closing the divide between users and analysts. And as tooling progresses up the ladder, analytical development picks up the pace, and user insights come faster. Analysts can swiftly incorporate feedback in a digestible format for users. While users get the data they need in a user experience custom-fit to their needs.
As organizations progress through each level, they build environments that increase both user experience and usability for creators and stakeholders alike, improving both communication and trust.
User Experience Democratizes Data
Great UX can also help to democratize data access. In particular, documentation, data management, and data governance rules are often siloed from the analytics process. A great user experience would unite these governance and basic data management functions with points of data access. Analysts should be able to peruse documentation and access the data from one place, reducing context-switching costs.
Unification of this sort would nurture consensus, trust, and credibility within analytics teams by connecting important contextual info with consumption pathways.
Moreover, with large scale implementations, effective user experiences in analytics tools do more than make data accessible - they educate users and builders, enhancing a data-centric and data literate culture within an organization.
Final Thoughts
User experience has been too long overlooked in analytics, and our work has been hamstrung as a result. Poor UX of existing tools perpetually slows the pace at which insights can be communicated and understood. And I think we’re on the cusp of something. Modern tools like Hyperquery, Mage, Motherduck, Snowflake, and dbt seem to clearly recognize the value of reducing friction in our tool-stack. Our industry is approaching its Apple moment, so let’s not pretend we’re happy with our Blackberries. We deserve better.
I write frequently about Data Strategy, MLOps, and machine learning in the cloud. Connect with me on Linkedin, YouTube, and Twitter.
 | A guest post by
|