CTO Corner #8: Incorrect decisions are not the enemy — stagnation is the enemy
In this edition of the CTO Corner I talked with Drew D’Agostino of Crystal about the path to building a startup and improving the customer experience.
How did you get into engineering?
I started building websites as a way to pay for college. Eventually I dove deeper and deeper into the stack out of necessity, then realized that I really enjoyed building software and pursued it full time.
I had a goal to pay my way through school without taking out loans, which was definitely a struggle early on. I had some scholarships, but still a huge monthly tuition bill plus living expenses. I used every spare hour to make money, and that led me into a bunch of gigs — freelance writing, brand ambassador, guitar lessons, tech internships, pizza box ad sales, bed bug extermination.
But in the process of trying to cobble together my own websites for these mini-companies, I fell in love with coding, even though my skills were super elementary. Luckily, I also got a much bigger scholarship for my last year after making a viral video about my school. I decided to sink my extra time into development. I worked on my own ideas, and soon was able to take on website jobs for clients. The pressure of delivering on time forced me to learn much more quickly, and ultimately I went from simple websites to full-blown applications.
During my last semester at Northeastern, my roommate Greg started working on the prototype that became Attend.com, and asked me to be part of it. After a mind-clearing post-grad road trip, I decided to stay in Boston to build the company with him.
Over the next 2 years, we grew to about 30 people and raised millions of dollars in venture capital to create Attend.com. Over that time, we rebuilt the product three times, each one more scalable and effective for our customers.
After we left the company, both of us were exploring next steps, and I landed on Crystal after an experiment that used NLP (Natural Language Processing) to predict personality. I decided to go full-time on Crystal in January 2015 and hired a few friends to build out our core product. We launched in March 2015, and over the past 18 months we have acquired hundreds of thousands of users, thousands of customers, grown to a team of 12, and raised money from investors like Salesforce and CEB.
What drove you to found Crystal?
Curiosity — I was digging into DISC and natural language processing simultaneously, then realized that the combination of the two could be very powerful.
Our personality analysis engine can detect anyone’s DISC type from text samples that they have written. It’s powerful because DISC allows you to predict a ton of behaviors and preferences that are typically unknown if you have never met a person. It’s game-changing because we can now predict the relationship dynamic between any two people that have never met.
What’s the key to get users to trust a machine/system to make decisions for them?
Consistency — you need to set clear expectations for what system can and cannot do. If it’s too “magical” you will eventually fall short.
How do you get user feedback to keep improving customer experience?
We get feedback for our personality data, the accuracy of our analysis, and the usefulness of our product.
Personality feedback is collected from upvotes and downvotes, as well as self or peer-assessment responses that can be submitted on any profile. That information is saved to the associated profile and modifies Crystal’s results.
Accuracy feedback is collected when new users sign up and select an accuracy percentage for their own profile. When profiles are confirmed as accurate by our users, that user’s data is weighted more heavily in the training data for our analysis.
Product feedback is a qualitative process — we get ideas and complaints sent to us all the time via live chat, email, and phone support. We have a master Trello board for all requests, and it allows us to log frequency, importance, and alignment with our product roadmap, and feasibility. We try to bring our users as close to the product development process as possible.
What advice do you have for up and coming founders and CTOs?
It’s tough to come up with blanket advice, but one thing I’m learning now is that there is huge value to decisiveness. Most of the time, you’re not going to have enough data to know if you’re making the right decision, but the faster and cleaner your decisions are, the quicker you can move and keep your team nimble. Incorrect decisions are not the enemy — stagnation is the enemy.
