I like to create and invest in data analytics companies. Over the years, it’s become clearer how to think about organizing the product to deliver the most value for customers. I call this organization the “data/value pyramid.” Essentially, each layer of value builds on the one below and customers are likely to pay more and see differentiation for each layer up. The actual volume (# of items) at each layer however is much smaller.
To explain further, the first focus needs to be on “aggregating” as much data as makes sense and is relevant to future value. This usually comes mostly from underlying systems like CRM (e.g. Hubspot), website traffic (e.g. Google Analytics), sales data, customer success tools (e.g. Zendesk), product logs, or external APIs like Facebook or Twitter. Getting this into a database where you can organize, normalize, combine, compare, trend…is actually no small task and requires real engineering skill and focus. Even something like normalizing timestamps can be annoyingly hard, but this is a critical first step. Some people might call this “data engineering.”
Next, create some ways to “analyze and visualize” the aggregated data. This could be Tableau, Google Data Studio (up-and-coming), Chartio (my current favorite), or other tools depending on what the end-user is trying to understand, how sophisticated the visualization needs to be, how many people need to see it, what form factors (web, mobile, exporting to PPT/Slides, email outputs…) are required. This layer of visualization was so important at RivalIQ (a marketing analytics company I helped start), we created our own visualization and output tools, which are very cool, but were a heavy engineering lift (see a sample report for comparing outdoor brands here). This is often where analytics companies stop and while customers are initially excited to finally “see” their data, the excitement quickly wears off as they ask, “so now what should I do?” This is where insights come in.
“Insights“ are where it starts to get valuable and answer the kinds of questions that many business users will pay for. Why is something happening? Why did that important metric go down or up? Even better are insights generated from the product which are surprising or were not possible before the data aggregation or visualization made them more apparent. For example, from the Rival IQ Outdoor Brands Report, “Consistent engagement tells social media algorithms that followers want to hear from you, which makes Instagram, Facebook, and Twitter more likely to serve your posts to followers again.” I am an investor in Remarkably where this kind of thinking is applied to commercial, residential real-estate (think apartment buildings) and they consistently deliver unique insights to their customers on how to market and lease apartments better. What content, what channels, what days of the week…drives the best customers to any given apartment building…
Insights can be found in commercial products like Rival IQ and Remarkably, but your team can apply the same methodology to your own internal tools, like CRM. Don’t just look at the data or reports, ask yourself, what is the data telling us? What are we learning?
Next, and now we are getting much more valuable, “suggested actions” or “now that I know something, what should I do?” Tweet more on Tuesday, add this hashtag for better engagement, call this prospect in the morning vs. afternoon…to get better results. Many analytics systems are starting to produce these kinds of suggested actions. In many cases, these are done with the help of humans at a company but surfaced in the software tools to customers. That’s just the start and as we get better data, better visualizations, better machine-learning, the humans need to do less and less. I am an investor in Sentinel Healthcare, we provide remote patient monitoring for people with hypertension (aka high blood pressure). Our software collects blood pressure readings, correlates it with other factors like a patient’s weight or medication, and looks for bad trends or anomalies. If negative situations are detected (insights), the software triggers a nurse to engage with the patient (suggested action). Then, depending on the situation, the nurse can solve the problem and/or add more data to the system.
Finally, and most valuable is “measured results.” The system found something unique (insight), then told the user to do something specific to improve and outcome (suggested action), and that action was proven to have the desired result. This is when it gets really exciting because that result is actually new data and can be fed back into the system. As the measured results improve, the system can speed up, making more suggestions, and providing a positive feedback loop.
Data is critical to almost every business and its importance is increasing. But, data is just the start. Using the data, software, tools, humans in the loop to get to a full data-value system is where the real gains will occur.
If you are building a system like this, especially in education or healthcare, let’s talk about how to work together.