How to Build an Analytics Front End that Doesn't Suck: Presentation-Purpose Fit
End dashboard spam forever and stop building stuff nobody cares about.
Too often, the valiant efforts of data engineers and business intelligence professionals are wasted in a limp dashboard-only front end that fails to show ROI and makes their work look trivial and uninspired. You know this feeling - when you work hard as a team on the pipelines, tables, metrics and orchestration for a ‘critical’ data need, only to find that the business looked at the dashboard a handful of times and moved on. Why does this happen, how can you prevent it?
By better aligning the presentation style of your data front end with the way it’s supposed to inform or influence people - ie, its purpose, you can deliver more timely, better received and more important insights. I call this presentation-purpose fit. Embrace it as a key to justifying data spend in the tough economic conditions of 2023.
What is Presentation-Purpose fit?
This isn’t a new concept exactly. ‘x-y fit’ is most commonly seen as ‘product-market fit,’ which describes the alignment between a product’s features, pricing, delivery model, etc… and the needs of the specific market it is trying to serve. If you’ve ever worked at a startup - or been within 50 feet of someone who has - you’re familiar with this term. Extending this idea to the data realm gives us presentation-purpose fit.
Presentation - Purpose Fit is the alignment between the presentation style of data and the specific business needs the data seek to address.
When presentation-purpose fit is high the data is well received, easily understood and does its job supporting business goals. These presentations of data satisfy needs of the business and lead to high value ad-hoc questions and business action.
When fit is low, data is poorly received, confusing or seldom used and fails to support business goals. These presentations leave the business no better off than before and lead to low value ad-hoc questions and frustration on business and data teams alike.
This emerges from product and design thinking, in which you focus tightly on your audience, understand them as people, build positive feedback loops between users and developers and iterate through hypothesis about what will serve them best until you achieve a tight coupling between what they need and what you actually produce. There are dozens of books about this process - I was first introduced to it in The Lean Product Playbook by Dan Olson, which changed my thinking about BI forever.
We’ll be discussing presentation-purpose fit and how to build useful analysis that go way beyond dashboards on this week’s episode of Super Data Show, Thursday, March 9th at 12:00PM Eastern.
Advantages of high presentation-purpose fit
This isn’t something you can achieve overnight - it’s a process that requires a shift in your delivery model, attitude and relationship with users. It’s hard, but worth it. Here are just a few things to expect as the benefit of all this effort.
Drive actions with data
The first thing high presentation-purpose fit delivers to the business is obvious next actions. This is the holy grail of data teams IMO, because the impact - and thus the justification for your budget - is immediately felt. It’s easy to identify the big wins here - running a successful new marketing campaign based on a clever analysis, for example. But sometimes the impact is simply to spur an intelligent follow up question someone wouldn’t have thought of otherwise. That’s an important action too.
Reach the right audience with the right message
When you align the needs of your audience with the content and style of your reporting and reach the right person at the right time with the right message something magic happens - a heady mixture of joy, relief and action.
Reduce wasted effort
Too often data teams put enormous effort into a front end that, ultimately, nobody uses - most commonly an abandoned dashboard. By finding presentation-purpose fit your efforts align with the needs of the moment. Complex dashboards are fine when called for, but in many cases you’ll find a simple data set or even single chart with a written explanation takes way less time and has immediate impact.
Align high value ad-hoc vs low value ad-hoc
I often talk to data teams that feel so inundated with ad-hoc questions that they can no longer contribute to the strategic mission of the company. They only exist to pump out simple answers to repetitive, one-off questions. One person told me they felt like a ‘chart monkey,’ which I’m sure is a horrible feeling. The first step to solving this is learning to distinguish between high and low value ad-hoc questions.
High value ad-hoc questions deliver insights that materially improve business outcomes. They may require new ways of thinking supported by new data pipelines or metrics. They lead people to take action. High presentation-purpose fit leads to high value ad-hoc questions.
Low value ad-hoc questions aren’t unimportant, but they don’t have a large impact on the business. These are often reconfigurations of existing knowledge to show a slightly different slice of the business. Think, ‘Can I see this metric by quarter?’ Low presentation-purpose fit leads to low value ad-hoc questions.
Low value ad-hoc drives a huge amount of data team frustration and wasted data spend. Improving presentation-purpose fit will lower the volume of these questions by heading them off at the pass. Once reduced, the rest can be shunted off to self-service while your team focuses on high value ad-hoc.
Finding presentation-purpose fit
Finding presentation-purpose fit is a process that applies product thinking to evolve how you deliver business intelligence and analytics. The unsatisfying - to some - answer is that it’s both art and science. You need to develop a repeatable process that starts with user needs and works backwards to how you present the data. You also need to deeply understand the communities that consume your analytics to develop intuition about their needs.
The single most important thing you can to do improve presentation-purpose fit is to work backwards from user needs to data, rather than forwards from whatever data happens to be available!
While it’s best to approach this in a systematic way as part of a rethinking of your analytics practice (which I can help with, message me!) there are practical questions an individual can ask to start developing better presentation-purpose fit today.
Do I know how this data requests contributes to the company making or saving money (the ultimate ROI justification!)
Am I working backwards from user needs to determine my design, rather than forwards from what data happens to be available?
If I had to sell this report as a product, would my end users choose to buy it? Why or why not?
Have I really understood the motivations of my users and what drove them to request this data?
What real world action will they take based on what I’m presenting?
Do I know the ultimate purpose for this data - or just a step on its journey?
Does everything in my report communicate something important? Is there fat to trim?
Have I reckoned with how Excel fits into what I’m building (It’s always the elephant in the room?)
Ultimately, product thinking in data is about creating data products that delight end users. ‘Delight’ is a very specific word and feeling that describes the joy felt when one unexpectedly discover exactly what they needed. In the context of BI, our product is the front-end presentation of what may be a long, complex data transformation process. If you fail to build a delightful front end, all of your backend work may be wasted.
Presentation-purpose fit for dashboards
Alright enough theoretical stuff - let’s look at an example. This is probably the single worst offender of presentation-purpose fit in the analytics industry - the dashboard. Analytics environments are often so littered with derelict dashboards they resemble the aftermath of some great, grinding battle. Why?
Why do you build dashboards nobody cares about?
Because that’s what people ask for, what they expect, what they’ve been trained to know and understand for over a decade. The logic behind the typical dashboard is simple:
Charts are good. We like charts.
However our screen is too big for just one chart!
If we put multiple charts on the screen, and they are at least sort of related, that will be even better!!
And if we can do some simple filtering, that will be best!!!
This algorithm is already running in the heads of data and business people alike, just waiting to be called forth to spit out a collection of charts that you hope will move the needle. The challenge is that a dashboard is a tool; when perfectly applied to its intended uses, it works brilliantly. When applied to things it’s just okay at, your results are just okay. When used irresponsibly and haphazardly, you get maimed.
What are dashboards for anyway?
To properly use dashboards, you need to understand when they are most effective. A lot of this thinking goes back to Stephen Few among others. What are the key aspects of dashboards and what sort of usage do they suggest?
Dashboards give a high level view of many metrics at once.
The use of position, color and sequence can show relationships between the metrics
When showing time variate data, they can suggest correlation between metrics
Basic filter controls allow for a degree of governed self-service exploration
Considering these characteristics in the context of presentation-purpose fit, we see following:
Perfect fit: Operational oversight
This is the grand purpose of the dashboard - to give you a single screen to that tells if something is going wrong in the real world that contains the information you need to take action. The further you get from tangible, understood, actionable processes - whether physical or digital - the further you get from the platonic ideal of the dashboard.
Acceptable fit: Executive reporting, guided slice-and-dice
You may be surprised to find executive reporting here. I find that executives often want deeper answers than fit easily into a dashboard, and they don’t typically use them to make decisions. Instead I would create a standardized notebook of the metrics an executive cares about and prepare written summaries of what changed and why that can be presented to them at a regular interval. Still, it’s not a terrible use of the dashboard format.
Looking to take your BI to the next level? The Super Data Brothers are a dynamic duo that applies presentation-purpose fit to every project and builds business intelligence systems that have real impact. Contact us to learn how we can help.
Slice-and-dice is another area where dashboards are an okay solution when paired with the ability to easily jump to self-service BI or excel. A well built dashboard with a reasonable number of filters provides an analyst the ability to identify the right direction for further inquiry. High level analysts will be better served going straight to the data, but for some a dashboard is a reasonable place to start. There are better solutions though.
Poor fit: Everything else
The worst offender here is using dashboards as a storytelling interface for high value ad-hoc requests. It’s just hard to explain what was done, why and what should come next in a dashboard. On the other end of the spectrum, sometimes you really should just send someone a screenshot of a chart, or an excel sheet, or give them a table. A dashboard is an extremely overcomplicated response to simple data requests.
How to know if you’re building a bad dashboard
There are a few telltale signs that the dashboard you are building has poor presentation-purpose fit.
You have no idea what real world actions someone would take from looking at it
There doesn’t seem to be any theme or relationship between the visualizations
You struggle to find charts to fill the space or
You struggle to fit all the charts into 10 tabs
You add dozens of filters
You’re just building a dashboard because the business explicitly requested one
You know everyone is just going to dump this to excel anyway
These are all individually red flags and in combination are probably a sign that you need to step back and ask yourself - should this be a dashboard? Do I fully understand what real world impact this data demands? Is there a more elegant, focused way to present this data?
Presentation-purpose fit in action
This blog post is long enough - hopefully I’ve introduced you to this concept, convinced you that it’s important and spurred some thoughts on how you can improve presentation-purpose fit to deliver more impactful analytics, save yourself wasted effort and improve the ROI of your data team. I’ll write more about this topic in the future, but for now here are a few examples of use cases and the presentation style I would recommend to quickly increase presentation-purpose fit for your analytics front end.
This is an Amazon affiliate link - remember, I gotta pay the bills somehow.
I spoke with a friend today who recently presented a comprehensive executive dashboard to a client CEO. His reaction was, ‘This is great. It looks amazing. I can’t believe I can see all these numbers in one place… but what the check am I supposed to do based on this?’ It’s a common response so brace yourself to underwhelm if you aren’t prepared for it.
This is really great, adding it to my ever-mounting to-read list so that I end up reading it again!