Why take the risk of starting a software company?
I could answer this question with something trite, like “I want to save the world from crappy products”. But the reality is a bit more nuanced. I’ve had 3 formative experiences in the last 12 years that convinced me there was a big problem worth solving with a packaged application.
At Frictionless, we built a SaaS company from pre-revenue to its sale to SAP in 2006. One of the most interesting things we did was to share adoption and usage data with the sponsoring executives of our customers during account reviews. It wasn’t pretty; we wrote SQL queries to extract usage metrics for each customer. We made charts and graphs in Excel, then pasted them into Powerpoint decks. But the data didn’t lie. And it provoked all sorts of healthy conversations about the expectation gap between executive vision and actual user adoption. This experience made me an early believer in usage data and its healthy effect on customer relationship management.
RSA, the Security Division of EMC
My flagship product was a security appliance that was distributed to four corners of the world across many thousands of units and customers. The product was getting “long in the tooth” (my kind words) and we experienced availability issues at an increasing rate. The problem is that we didn’t have any data about the use of the product in customers’ environments. This lack of data inhibited our ability to spot root cause drivers across our customer base. And it was made worse by the growing variety of customers by size and use cases. There was no “archetype” customer that we could build and test for.
Another challenge was to depict the health of my product line during each quarter’s internal business review. Was the average selling price the same in Germany via the re-seller channel as it was in Brazil via direct sales? Was our customer renewal rate the same across every geography, customer tier, customer vertical and distribution channel? Of course not.
We struggled to de-construct the global business to spot the outliers that inevitably exist. For both the sake of spotting things to fix and finding the most successful segments in order to replicate the winning formulas.
AVG is a provider of freemium security and related products to consumers and small businesses. Most of its revenue comes via online distribution. When I arrived, I had a team member inventory all of the customer data sets we had; reducing churn and increasing cross-sell of our newly expanded product suite were strategic imperatives for me. I was certain the data held the clues. What I found was a lot of data, but living in silos. Marketing had lots of clickstream data from the website. Engineering had lots of product usage data from constantly tuning the anti-virus product’s protection algorithms. We also had a huge self-care community online. And an e-commerce system.
I spent 2 years building an internal analytics team, starting with hiring. Then we collated all of this data into some emerging technologies with which we had little experience (“Hadoop”, “Datameer”, etc.). Then we started producing basic reports and metrics out of it. Then we built some early statistical models to discover relationships in the data. 2 years later and we were still scratching the surface in terms of understanding the user behaviors that drove churn and cross-sell, and how to operationalize the findings. There had to be a better way.
What’s in common across these experiences?
Each experience was the same in several ways:
- important data about our customers was locked in many silos
- the effort to overcome that was expensive, time consuming and required real executive commitment
- as a company, we lacked key insights about who our best customers were
- we were trying to improve business metrics that everybody cares about: customer retention, Lifetime Value, ARPU, customer success and customer advocacy
Ultimately, it was these common themes that led me to believe this was a widespread issue for companies, and that a packaged solution could be the answer.