Why I want to kill analytics dashboards and you should too
They promise to give you insights in the blink of the eye but most analytics dashboards end up taking far more than 10,000 hours of your time. SessionCam’s Chief Product Officer, Richard Churchill, thinks there’s a better way…
In the early 2000s, the idea of ‘thin slicing’ — making fast assessments and discerning patterns of data based on narrow windows of experience — became pretty fashionable. The man to blame was Malcolm Gladwell, whose blockbuster book Blink shoved the concept into the mainstream.
Three years after Blink burned its way up the bestseller list, Gladwell’s next book, Outliers, put forward an entirely different perspective on expertise. This time, the idea that caught the public’s imagination was the 10,000 hours rule— a claim that the key to being world class in any field is practicing the skills required for that long.
Of course, if you dig into the underlying studies that Gladwell built both of his books upon you discover that the reality is a lot more complex. But the pleasure of pop psychology and sociology is that they take that complexity and pare it down into easily shareable ‘wisdom’.
You can encounter the same issue with analytics dashboards that promise to present you with blink-like insights into customer behavior. The fascinating complexity of what your website visitors are doing is polished down to a set of comforting assumptions.
Your field guide to diagnosing customer frustration.medium.com
There’s a kind of joy in complexity
User behavior is complex. That’s one of the reasons I love studying it. What constitutes an extreme in one area may not be in another. That’s why SessionCam’s Customer Struggle algorithm establishes an intelligent baseline so you know what’s important relative to the average website visitor.
Measuring behavior using one metric or even a handful of metrics is a bad predictor of customer frustration.
Leaning on one gauge will always lead you astray. That’s why we call the vanity measures promoted by some solutions Unmarvelous Metrics — they make a lot of noise but actually tell you very little that’s useful.
Trying to pin down the ebb and flow of customer behavior with a few specific metrics is like thinking you understand the motion of an entire sea by analyzing the way buoys bob in the harbor. Yes, it gives you a sense of what’s happening but there’s a lot more going on in the depths of your data.
It doesn’t matter if the metrics you choose are business-orientated, such as measuring how many visitors click ‘add to cart’ on an important product page, or goal-orientated, such as determining the level of ‘engagement achieved’ based on scrolling behavior alone (shudder!)
These approaches on their own only give you access to a small part of a much broader and more exciting picture.
Is your analytics dashboard a playpen or a prison?
When you lovingly build out a configurable dashboard, it’s easy to feel like you’re constructing a playpen. But if you’re not careful, you’ll look up one day and realize you built a colorful cage which doesn’t let you get to grips with lots of valuable data.
How do you know what you’re missing when you always let the dashboard decide what’s worth looking at?
To ensure that you’re looking at the right data — data that will help you achieve your goals — you need to step back and ask yourself this:
Can I look at these metrics, digest what they’re telling me and make sound decisions quickly?
Dashboards can be brilliant at offering up information that’s interesting at a glance. For example, seeing the likelihood of stock running out based on current buying behaviors is very useful, assuming that what’s feeding the dashboard reflects how that metric has ebbed and flowed over time.
The most important thing is that the information displayed to you is relevant to your specific needs. That’s what bugs me about dashboards that claim to show “key” pieces of behavior on a page. Very often those are actually just the behaviors you have decided are key rather than what’s actually important in your website visitors’ activity.
Say you set up a dashboard to watch a shopping cart page: what’s the most important behavior there?
Is it product deletion, entering a promo code, the number of users who attempt to progress to checkout but hit an error, promo code failure or the number of users that exhibit signs of cognitive struggle?
That previous sentence was long and with good reason — there’s a lot of factors to watch in even that simple example.
Promo code errors screw up your conversion rates. Here’s how to fix that and the data prove it.medium.com
Let go of your analytics ego
Our human egos persuade us that we can take in a vast range of metrics, digest them and make blink smart conclusions. But most of the time, you simply can’t and dashboards are just acting as a crutch for that delusion.
Bad dashboards offer the illusion of knowledge.
To compare metrics, you have to have a good understanding of what’s normal in any given situation. If you don’t have a benchmark, you end up making bad decisions, driven by a partial view of the data.
Think about what you’re trying to achieve. I understand the comfort in looking at a page with lots of figures but if you want to instantly grasp what’s going on, you’re probably asking too much of the dashboard and yourself.
I want to kill off the traditional dashboard. It’s what we’re working towards at SessionCam. Valuable insights should be delivered to you with as little effort on your part as possible.
You can already get insights delivered to you in SessionCam — the most valuable routes through your site with Key Journeys, the most troublesome areas surfaced by the Customer Struggle score algorithm and your mostly costly errors automatically identified by Error Reporting — but our road map goes far beyond that.