.jpeg)
In any given domain, ‘meta’ knowledge is key.
Everything’s a game, and being good at playing games is the most important one of all. Obviously, the most important thing to know before playing any game, be it FIFA, commodities trading, or product development-are information and knowledge.
Let’s say you want to get fitter-you can read particular books about running, lifting, or calisthenics. Can’t hurt, right?
Well, literally speaking, it can’t. But functionally, it’d likely be a better use of your time to just do those those things, and to learn by doing.
On the other hand, let’s say you read a book about physiology. This would likely inform you about all three: running, lifting, and calisthenics, and be complementary with whatever you’d be learning by just practicing those activities. It’d be complementary because it’d affect every micro-decision you make, it’d be changing the framework upon which you make your decisions in the first place.
It’s a change from a first principles perspective.
In an ideal scenario, that’s what one ought to be doing when developing products as well. You don’t want information thats only helpful in any one narrow context, but information that make you a better decision maker in general.
This basically translates to the fact that data isn’t where the value is at, insights are. Insights will change your overall decision making framework, whereas tidbits of information may just do that for specific decisions.
You want to learn about how your body works, not how running works.
That’s precisely why SaaS companies could greatly benefit from recalibrating their approach to product analytics and shifting the focus to insights.
Even in 2023, there is still a significant gulf between sourcing data and then actually leveraging that data to make good decisions. Indeed, 41% of executives have reported that they find data driven decision making to be extremely challenging.
As it stands, data is more than abundant-but insights, and thus quality deliberation, are lacking.
Clearly, this is not good for the bottom line.
So why’s this the case?
And more importantly, how can it be tackled?
Firstly, let’s look at how analytics tools themselves may be responsible:
The point of tools like Mixpanel and Amplitude is to provide data, but the ultimate point of data itself is to gain insights. For this, the data needs to be easily digestible. As it currently stands, it is not. We're talking about a tsunami of numbers, charts, and figures, all streaming in real time from various data sources. While the availability of large quantities of data is in itself a good thing, it can easily turn into a hindrance if not managed properly. It's not enough to merely have data; it needs to be relevant, manageable, and directly applicable to the challenges at hand. Too much data can be just as paralyzing as too little, if not contextualized or filtered properly.
The other problems that we’ll discuss basically run downstream of this one crucial one.
In large part necessitated by the sheer volume of data, analytics tools do not have the most elegant or user friendly UX. Indeed, Mixpanel themselves seem to have tacitly admitted this earlier this year in this article, in which they are announcing their new toolkit for Custom Buckets. In this feature, users can ‘bucket existing user segments into new, more meaningful ones.’
Given that the industry leaders are building this functionality, it's reasonable to assume that this is a pressing issue.
The most pressing issue this UX issue causes is the fact that knowing how to use most analytics’ tools is a skill in and of itself. What this means is that the potential for cross-collaboration between teams is affected, as roles that directly involve interacting with tools like Mixpanel then become necessary conduits to disseminate data across teams.
If the flow of information is impeded, it increases the chance that team members across different verticals will not be on the same page. So, when it comes to making sense of raw data and translating it into actionable insights, there’s a problem. This is because people in different roles bring different lens’ to the same problems, so the way they might analyze raw data and refine it into insights is likely different.
Now, given how fundamental the aforementioned problems are to any SaaS company, its apparent that fixing them ought to improve profitability, which would of course be beneficial for the toolmakers themselves.
So what are some ways this could be implemented?
Making data more accessible is a relatively low-hanging fruit. Nevertheless, it would go a long way in making the pipeline from raw data to insights easier. This could be on parameters such as customer behaviour, demographics, or product usage. Indeed, the step described above taken by Mixpanel is a decision exactly in this direction.
The ability for multiple users to be able to present data in the form of insights would also lead to new, different metrics being given more attention. Consider Figma: collaborative features enabled non-designers to be a part of the design workflow, something which has been greatly beneficial for companies everywhere (and was of course pivotal for Figma’s own growth).
In this context, let’s consider a CRM SaaS company. Conventionally, analytics would be handled by the Product Manager, who may focus on data points like user engagement, churn rate, user acquisition cost etc.
Now, let’s say engineers also integrated analytics’ tools into their workflow. Given their likely superior understanding of the granularities of their product, they may be able to understand in real, practical terms how much incremental upgrades they make actually affect the end user experience. For example, they could observe correlations between server response times, or app loading times. While PMs can of course do so as well, they would not be as acutely aware of the engineering efforts/bandwidth required for any give uptick in user behaviour.
The root cause of the issues spoken above is the fact that the first and final filter for the large amounts of data points involved is the analytics’ platforms user interface itself. So, it stands to reason that having a sort of automated filter ‘behind the scenes’ would be a huge aid in improving this issue.
What does this mean?
Aside from some basic manual inputs, the data you see would first be analyzed by an LLM, which would then assess the content, assess your role within the company, and then present the insights in whichever visual format you have demonstrated to be most at ease with.
An added benefit here that LLMs provide is the fact that your preferences (to the degree that they must be manually inputted) can be input in plain text. Given the degree of developer bandwidth that is currently required to both setup and configure similar tools today, the advantage this brings cannot be overstated.
Let’s paint a picture here.
First, you input (in plaintext, of course) your broader goal-let’s say Doordash wants to decrease their cart abandonment rate by 8% over the next quarter.
A model then goes and backtests data and ascertains variables that have correlations with cart abandonment- autonomously.
It provides these recommendations, provided in a succinct manner and easily digestible manner, which the model has assessed to be the optimal way to present information to any particular user.
It would also be a lot easier to personalize this presentation. Given the option for a natural language interface with which to interact with, the model would over time learn which methods lead you to have the most clarity from the get go-this could be figured out quite simply (for example, the less questions you ask, the better the presentation was).
The demand for clear, actionable insights takes precedence over raw data-a dynamic only gaining momentum as the amount of data at our disposal gets increasingly vast and complex.
This prompts the need for intuitive, AI-powered platforms to bridge these gaps. Crunch is the solution for this era, enhancing ease of data comprehension, enabling efficient cross-team collaboration, and presenting insights tailored to individual roles and preferences. With an AI backbone, the platform simplifies the journey from data collection to actionable strategies. Don't merely collect data; leverage it. We invite you to explore our platform and revolutionize your approach to product analytics. Discover insights, not just data.
Signup for out waitlist here
