80/20 Rule: Why Do So Many Spend 4x More for Analytics than They Planned?

Zdenek Svoboda's picture
Vice President, Platform

Follow on:

Have you heard of Vilfredo Pareto? Even if you haven’t, you’re likely familiar with his work. More than 100 years ago, this Italian engineer invented the popular 80/20 rule, originally for economics. I bet that he had no idea how important the rule would become for other areas, such as product management.   

How does the 80/20 rule relate to product management?

There are plenty of product management examples where this rule applies. For example, getting to the point where your new product reaches 80% levels of completeness and adoption will ultimately take 20% of your team’s time. Consequently, building out the remaining 20% of your product’s capabilities will take 80% of your team’s time. Good to know when you are planning your go-to-market strategy. 

Knowing this tendency, it might seem like a good option to essentially ignore the “last 20%” long tail and focus on getting to 80% completeness. And sure, everything sounds great at the beginning. You achieve fast time to market, and you’re seeing high initial adoption rates. Unfortunately, if you don’t chase the 20% long tail, you’ll pay your debts later with low retention rates as your customers get quickly behind the 80% threshold, and start struggling. 

Why focus on the 20% long tail?

Business intelligence and analytics have few great examples of why you should focus on the 20% long tail of your customers. I wanted to share one in particular that’s related to performance and scalability—essential for almost any analytical solution and are even more important for solutions that deliver analytics to many customers. 

I see many customers who focus on the initial implementation and testing for average customers, forgetting about their large customers’ long tails. Because these larger customers tend to have different needs than the average customer does, we often see that significant infrastructure changes and additional unplanned costs are needed before implementation. In analytics, revenue is strongly correlated with data size, so ignoring 20% of the largest customers—who generate more than 80% of their revenue—is one of the costliest mistakes they can make. My estimate is that this approach leads to 4x the initial budget overrun. 

want to see what gooddata can do for you?

Get in touch and request a custom demo

How GoodData can help

We’ve met with quite a few potential customers who have already been burned by failing to account for the issues I referenced above. You can immediately identify them, because they always start off by asking us about this problem.Typically, they’re meeting with us because they know we’ve implemented two essential capabilities to address the 4x cost problem. 

The first one is data distribution. We partition data to workspaces that contain data for one customer only. Achieving predictable performance on the individual partitioned workspaces that are encapsulated from the HW bandwidth perspective is far less costly. Besides the performance-cost ratio, the partitioning is also far more secure from the data privacy perspective. 

Another feature that we’ve implemented are the pluggable analytics back ends. The GoodData platform can seamlessly switch between back ends that are optimized for the 80% of small and average customers. The high-performance clustered, columnar back end provides blazing-fast performance for the largest  20% customers. Our customers don’t need to do anything to migrate between these back ends; we flip the switch and their analytical solutions just work as they did before.  

Combined, these two capabilities bring huge value to our customers. They can build one solution and use one infrastructure for all of their customers, regardless of size. They don’t need to do any lengthy and costly optimizations or even start implementing entirely new solutions for the 20% of their largest customers. 

With GoodData, cost planning is very predictable. You won’t see any price multiplication, and you pay what you planned. Moreover, we just introduced even more predictable pricing for the first year: pay a flat price per each customer you have, no matter what bandwidth you need. 

Do you want to know more about GoodData pricing? Just fill in this form and we’ll get back to you. Are you interested in more technical details about our platform architecture? Download our platform paper.

August 28, 2019
Blog Bottom - Subscribe Newsletter

Want to ask about something specific?

Contact us