How AI will transform the UX of Analytics & Performance Measurement
AI is over-hyped. It won’t be the holy grail for every type of product. However, it will be a game-changer for analytics & performance measurement products.
The user experience of these products has become too obsessed with data and forgotten that users most care about answers. AI will transform this by giving analytics & performance measurement products the ability to support users with business decisions in language that is precise & contextual.
Why AI will transform the Analytics & Performance Measurement user experience
The global big data analytics market size was valued at $271.83 billion in 2022 and this doesn’t even include the countless analytics and performance measurement products that are critical to software in the worlds of AdTech, MarTech, CRM, HR, DevOps, and almost every other.
Creators, businesses, and enterprises all rely on these products to make business decisions. Data powers decisions today more than ever. But decisions need more than data alone, they need an idea of what to do.
Data visualizations: The best solution to an age-old problem
Visualizations came out of the era of enlightenment when big ideas had to be conveyed simply. Graphs, charts, and maps explained the statistical data that was being discovered across the sciences. The visualizations made it easier to interpret the data and the finding.
In the 18th century, William Playfair published a book that incorporated graphical representations of data that transformed how economics could be explained.
As you can see, these visualization types are still in use today, 250 years later.
Data visualizations make it easier to interpret large sets of data and interpret differences/changes in the data set. They’re the best solution to present data.
Presenting ideas is more challenging than data
In 1490 Leonardo Da Vinci created The Vitruvian Man, a visualization that is more iconic than any graph. Oxford professor and British Art Historian Martin Kemp called it the world’s most famous drawing in his 2019 book, Leonardo da Vinci: The 100 Milestones.
What Da Vinci accomplished in 1490 has eluded many great thinkers and product designers since: the ability to simply communicate ideas. Ideas are different than data.
Successfully conveying ideas requires communicating a point of view about data. Da Vinci’s diagram did all of this in an illustration that can be understood without words and annotations. Da Vinci sparked the idea of human-centred philosophy and the subsequent human-centred design thanks to a single illustration. It did all this while being minimally executed.
Why conveying ideas matters more than data
Data may change people’s minds but ideas spark action.
In Ben Decker & Kelly Decker’s Communicate to Influence: How to Inspire Your Audience to Action they present a range of research, case studies, and lessons about why communication fails. One of the biggest reasons: placing too much importance on the content and information of your presentation and not enough on why it should matter to the recipient.
Analytics & performance measurement is guilty of this. The data presented on dashboards and reports are usually centred on the data, not what it means.
Without presenting a point of view and idea of how to act, analytics & performance measurement are only informational.
Why people use analytics & performance measurement
Our team at PH1 Research has worked on many projects to improve the UX of analytics & performance measurement products. And regardless of which industry or use case, the majority of users are seeking answers. They need help solving business problems.
Our research found that while users dig into reports to better understand the data, most don’t know how to act on what they find. In fact, many take unsubstantiated actions because they create their own, false hypotheses about what the data means.
Examples:
- Decreased conversion might not be a signal to spend more
- Increased performance might be a sign of issues elsewhere
- New audiences gained might be a low-yield distraction
- Products hitting KPIs may have nothing to do with your update
Why we need to redesign the UX of analytics & performance measurement products
First off: There’s nothing inherently wrong with dashboards, charts, and reports. They are the best solution available for presenting and consuming large data sets.
There are also many resources for how to create visualizations that really work and thought leaders (like Aurélien Vautier) outlining the gaps and opportunities of analytical user interfaces.
There is however an opportunity to better align to user expectations thanks to advancements in front-end and back-end technologies.
Just look at data science — enrollment in Master programs have seen 20% enrollment growth between the 2020–2021 and 2021–2022 academic years alone & 650% job growth since 2012 — to see how the data maturity of organizations will continue to surge. This all results in better data and insights for designers to work with.
Why AI will transform the UX of Analytics & Performance Measurement
Large Language Models (LLM) — (OpenAI’s ChatGPT, Google’s Bard, etc.) — work by analyzing large sets of data and recognizing patterns. They help users know what matters most and also can recommend actions — exactly the gap that analytics & performance measurement products need resolved.
Some of these opportunities:
- Users are told about an increase/decrease, explained why, and recommended an action to optimize performance
- Users running multiple campaigns or monitoring multiple work streams are explained which is performing best, steps to optimize further, and provided a guide for future efforts
- Users creating a new campaign/asset are recommended launch plans based on historical and market performance
What needs to be solved before this is possible
Requirements for this level of strategic analysis and recommendation:
- Know what’s being requested: Comprehension of common requests, relational data being requested, and ability to ask for additional info
- Know what happened: Situational data and contextual metadata
- Know what is important: Well-structured and sufficient data about user’s history, plus of marco/market
- Know what to do about it: Lexicon and grading of potential outcomes, including risks/rewards
Depending on the data maturity of your organization and the depth & quality of your data, these sorts of requests may be possible next year or next decade. One of our clients is planning to launch these capabilities within a few years and based on our research with users, it will be an absolute game-changer for their business.
What you can work on today to improve your analytics & performance measurement product
- Create an ethics charter for appropriate data usage
- Research how users make decisions using data
- Research which data will be necessary to boost recommendations
- Prototype work flows to evaluate the quality of answers & outcomes
If you’d like help with the above, please contact me so I can walk you through the process.