Live user-on-page activity, the analytics feature that wasn't

Way back in the mid 2010s, so well over a decade ago now, the business of "Big Data" and data science was in full boom mode. Everyone was running out to become a data-driven company and vendors were popping up left and right, fighting for VC money to get in on the action.

With that money came all sorts of experimentation with technology as the industry tried to figure out the best way to go about understanding users. There were a bunch of companies fighting to be the mobile analytics platform for the then-popular "mobile first" strategies of the time. There were various frameworks for running A/B experiments on the frontend and analytics collection frameworks that tracked user interactions. A lot of those companies were highly valued and have now faded away into relative obscurity as their ideas managed to find a niche or not.

One of the more curious analytics ideas to bubble up during that churn was the idea of "what if you could track, in real time, what users were doing on your webpage right this moment, down to the literal pixels on the page they were interacting with?". Essentially, an intensive browser tracking framework that would report user activity at a level of detail that was pretty much unheard of at that point. I most remember the concept in the NYC startup scene under the company Chartbeat (they're still around, and I'm not throwing shade in their direction today because I have absolutely no idea what their business and product are doing now).

As a nerdy tech and data person looking at that kind of technology in the 2010s, it was pretty fascinating. If you had the analytics suite running on a very popular website where you could guarantee multiple people simultaneously viewing a given page, you could see interesting data points like "X% of people scrolled past halfway" and what they were clicking on in real time. Slap on a slick visualization interface on top of that and you would have an interesting product demo to pitch to investors and potential clients. If memory serves, one of the main target audiences for this product would be major news sites.

But now that it's 2026, I'm sure that the vast majority of you are thinking "wait, we don't use any of that anywhere these days." I wouldn't be surprised if anyone younger than 30 have encountered such a product before. The reason is pretty simple – I'm not sure if anyone found a long term use for the information.

While at first glance, looking at the visualizations is impressive in a "wow, you can see people doing stuff!" kind of way, the immediate next question was always "so what can I do with this information?" You can handwave a bit and say that if you know your users are on average not scrolling deep into a page, it has implications on your ad placements, your article formatting, your overall site layout. But you could've gotten that with much more simple event-based tracking triggered when users scroll to certain positions. There wouldn't be a need to pay a VC-backed company SaaS level subscription fees for just that feature alone, and I'm pretty sure very few clients were willing to sign checks for what is effectively a vanity metric.

Since I didn't keep track of what the company did over time, I'm not familiar with what other analytics features they implemented to prove their usefulness. I knew one or two people who worked there over time and even back then they were aware that the interesting feature they built was a solution in search of a question. From the looks of it, Chartbeat used the initial hook of an interesting visualization feature to get clients, and then built out a more complex analytics platform that actually provided something useful enough to justify their continued expense. I have no idea how successful they are now at it – their homepage exists but everything pricing related is just "talk to sales".

What's also amusing is that I'm positive that some other analytics platforms of the era had taken note and implemented similar "live stats" features to compete. As far as I know, none of those features found any traction. It goes to show that when vanity metrics are obvious enough, even overly credulous investors and analytics purchasers will eventually figure out that some things are too useless to pay for.

Chartbeat's analytics framework at the time represents one of the more data intensive tracking that could be done. The browser is sending a constant stream of little messages showing where user 1234 is on a page and what they're doing. It's about as "Collect everything and sort it out later!!!" as you can get and that was something of the philosophy of analytics of the era. You can still see bits of this in more modern day analytics with things like how Kindles were reported to track every single page turn and a bunch of other information back in 2020. While there may be legitimate analytics and product development uses you can use that data for, it is definitely a very intensive way to track how someone is interacting with your product.

Web analytics in the 2020s, doing more with less

Back then, people were collecting every scrap of data they could because it wasn't clear what would prove useful. New and exciting NoSQL frameworks finally let analysts use that mountain of data, so it was worth trying. Eventually everyone slammed into the same issues with cost, scale, and slow analytics and so modern best practice is to scale back and collect roughly what you expect to need.

But eventually, the public got a bit tired of having all their information sucked up and brokered everywhere by every site. It was common to visit a random page on the internet and trigger tracking pixels for Google, Facebook, Adobe, and a whole host of other analytics services.

So privacy efforts started taking hold. GDPR passed in 2018. Firefox, Safari and some other browsers started blocking/restricting use of 3rd party cookies in the 2020s. Ad blockers and Javascript blockers had always been popular even prior to those efforts but over time they grew increasingly popular with the general public. This means that the current trends in innovation around analytics is working with and around the constraints. As you can imagine there's technological barriers and innovation involved with various vendors trying this or that way to identify users in order to advertise to them. There's also been product changes like trying to encourage users to use a single sign-on identity across multiple sites which on the one hand provides users with convenient account creation methods, but also provides data companies a firm identity link. And finally there's various analytical hacks and models that need to be implemented when a solid chunk of your user base nowadays refuses to accept cookies or finds ways to block your analytics efforts.

While there's still lots of innovation going on in the analytics space, a lot of it is less visible infrastructure and policy work. It's important work that affects billions of users across the globe and potentially shuffles billions of ad funds around, but at the same time is significantly less eye-catching than what it used to be.

My guess is it would take the opening of a new interactive medium to trigger another wave of frothy innovation churn – like for example if VR/AR things actually took hold. Health and fitness data triggered a bit of related churn when those devices started taking off too. I'm sure humans will invent some newfangled thing that demands new analytics at some point.