Sometimes it takes a bit more time to get something right — especially when it’s technically challenging.
That’s why Amiato, a Palo Alto, California-based startup, labeled itself as one of the “off the record” Demo Day companies when it graduated last spring from the Winter 2012 class of Y Combinator. But today Amiato, which was previously known as Nou Data, is coming out of the shadows with a product that seems like it could be well worth the wait.
Amiato has built a tool that lets companies, websites, and apps perform comprehensive A/B test analysis on their products at big data scale. What’s especially compelling is that from end to end, Amiato can be used by relatively non-technical product managers, requiring no assistance from engineers or IT administrators to make the most of the tool — all that’s necessary is a knowledge of SQL, which many PMs have, or the ability to work a translation tool that speaks SQL.
The company, which currently has 7 full-time staffers and has up until now been testing with a handful of lighthouse users, has attracted $2 million in funding from some big name investors including Y Combinator, Data Collective, Andreessen Horowitz, Ignition Partners, and others. Today Amiato is coming out of stealth mode to open up to more private beta customers to test the product for free.
The chasm in A/B testing tools
Amiato is looking to hit an unaddressed market when it comes to A/B testing, cofounders Mehul Shah, Stavros Harizopoulos, and Nathan Binkert told me in an interview this week. At the one end of the spectrum are tools such as Optimizely, Mixpanel, and even Google Analytics, which provide close to real-time testing feedback. They’re valuable and incredibly useful tools that are user-friendly even for non-technical people, but they can be very limiting in the types of questions that can be asked. Those who are ready to dig deeper and find out the “why” behind what’s happening on a site could be left wanting a bit more at this level.
At the other end are sophisticated data processing tools such as Splunk, Google BigQuery, and Cloudera’s Impala. These let companies turn their unstructured data logs into SQL-searchable formats, but they’re not exactly for technical beginners. Amiato says that these tools typically require an engineer or an IT admin to “massage” the data logs into the correct format before they’re loaded into these systems — a process with some friction that can take a product manager a bit more time to see A/B testing results. These kinds of tools can be intimidating for mid-level companies with lean engineering operations.
At the moment, there’s a bit of a leap required between the low end and the high end when it comes to A/B testing.
The Goldilocks porridge of A/B testing
Amiato aims to fill that chasm with something that’s “just right” for its crowd. Amiato lets users load JSON data files or other forms of semi-structured data activity into its system without any “massaging” required — they simply upload files or drop the full data dump into a bucket on Amazon S3. Then they share that S3 bucket with Amiato, which automatically figures out the structure of the data and maps it all into a table. All that users are required to know is SQL to run any kind of complex query against this now-structured dataset and receive answers back within a matter of seconds.
“We have a UI that makes it easy for people to start using the product, so you don’t have to sit there and write low level code to make it work,” Shah told me, adding that the results can be very meaningful for users. “I think many companies right now are scared of doing rigorous big data A/B testing because they don’t know what to do, or how to get started. What A/B testing allows you to do is stay one step ahead of the competition, do a series of experiments with your audience, and optimize your site to make it more meaningful.”
For data-driven companies that want more
The target market here, Shah says, are product managers or advanced product managers at companies that deal in realms such as gaming and online retailing — anything that is data-driven.
“These folks have outgrown their off-the-shelf tools, but they’re scared of figuring out how to put together a huge data team and work with huge pieces of infrastructure. But this could also work for advanced people who have a team in place, but are frustrated because they’re not getting their data in time to keep a real edge.”
It seems like a very promising product, and Amiato has put together a very seasoned team to do it. All three co-founders hold Ph.D.s and are alums of Hewlett Packard, where they met while working for years in the company’s research lab arm as big data experts working on data management infrastructure. Amiato is certainly going up against some big names in the data processing space, but sometimes it takes a fresh team to tackle something in just the right way.
To check Amiato out and sign up for its private beta, go to the company’s website at Amiato.com.