Tidemark founder Christian Gheorghe grew up in communist Romania, and his background helped form his thinking about the three-year-old startup, which aims to bring sophisticated data analysis to the masses.
"I do see a lot of conversation around consumerization as being sort of equal to BYOD. It's more than that. It's certainly at the core of what we believe in. For me it's very deep down based on my experience growing up, where you're told what to do."
He continues, "Analytics used to be the domain of the few. IT purchased this thing and is going to tell everybody how to plan or forecast or analyze your performance of your company, then IT will provide you this book of Excel or pivot things."
Now, the power has shifted, Gheorghe believes.
"For the first time there's this pull for a better kind of experience, a better kind of actionability, and it comes from the user, not from the vendor. … It's almost like a perfect storm. People are asking for things because they want to be involved, but they also realize that what they had was not sufficient."
Gheorghe knows about data analytics -- he was the chief technical officer at business analytics company OutlookSoft, which SAP bought for an estimated $200 million in 2007. But now, Gheorghe believes that his old company, and its competitors like Hyperion (bought by Oracle for $3.3 billion) and Cognos (bought by IBM for $4.4 billion), took the wrong approach.
When Tidemark started out -- the company turned three years old last week -- Gheorghe bucked the trend of building the product first. Instead, he and cofounder Tony Rizzo, who ran field services for OutlookSoft, went on the road and spent six months interviewing prospects before they wrote a single line of code.* (Another cofounder, Nenshad Bardoliwala, had a technical background. He is no longer with the company.)
"What we did for six months is talk to about 100 customers. We said, 'You have 7 to 8 tools you are using, some of which I just sold you in my previous company. And I'm here to tell you that was great, we've done it, cool, but I think there's a better way and here's why. What do you still struggle with?'"
They found that customers were struggling with the flood of data from new sources -- sources they hadn't imagined would be available for analysis five years ago.
"It's not they need what they have but a little better, they just cannot accomplish some of the things they're trying to do with the new world of much more data and volatility and data."
Older BI systems use a process called ETL -- extract, transform, and load. Data is extracted from a series of predefined sources, transformed into a format can then be consumed by analytic tools, then stored in a data warehouse. To set these systems up, companies need to figure out what they're going to try to measure before they ever start the process, so they can figure out which data to extract and how to formulate it. That means consultants and slow rollouts -- implementations often take a year or more. It also means that end-users are bound by the restrictions built into the data transformation process from day one -- if they want to measure something that the original designers never thought of, they're out of luck.
This used to be a reasonable way of doing data analysis back when the most relevant data lived in one or two systems, like a general ledger.
"Now, most of the data people care about is no longer in their databases," says Gheorghe. "They still need to take transactional data, like ERP systems – both on premise in and in the cloud, like Workday, which is a partner, also SAP, which is the biggest company we work with. But a lot of it is outside the organization. It's public info from Bloomberg or Reuters telling you how this particular industry is doing. It's Twitter feeds that talk about sentiment associated with the product you just introduced. It's weather data and RFID data that tells you how your production levels are doing in the field. It's all that unstructured data along with the structured data…put together for the benefit of the many."
To help companies make sense from this flood of data, Tidemark aims to swap the order of the "T" and the "L". In effect, data is extracted from a wide range of sources, then put into a big unstructured data store in the cloud (Tidemark uses Cloudera's Hadoop implementation).
"When you do ELT – that's what cloud as a computing platform enables. You're putting everything in a grid and the transformation actually happens at runtime, when you ask the question," says Gheorghe.
Paradoxially, to make data more accessible to end users still requires a lot of up-front customization and thought. That means a high-touch sales process. This is no user-provisioned "land-and-expand" kind of product.
That meant a long period of testing, refinement, and working with early customers. But finally, last August, the product launched to general availability.
The company has already gathered a handful of sizeable customers, including Chuck E Cheese, Pabst, and US Sugar. That's a pretty big coup considering that this is a rip-and-replace scenario -- you don't add Tidemark to an existing performance management system.
The average implementation time is down to 82 days -- a lot shorter than an old-fashioned BI implementation -- and the company hopes to bring it even lower.
Gheorghe calls hiring Rizzo his first heretical moment. His second was building the client-side app on HTML5 instead of Adobe's Flex, which was a popular cross-platform programming model back when the company was starting out. It turned out to be the right choice -- the web-based app looks and works the same way across all kinds of devices, including tablets and phones.
That fits right back into the company's core mission: making data useful for all.
"It's not one particular advanced user. It's a person in the field -- one of our customers uses it for the ability to sell to everybody in field who's growing sugar. Instead of getting this several hundred page PDF report that corporate sends you, you'll have day to day information on how weather patterns are affecting your crops, what are the chemical levels are, what are the currency associated with selling crops per day. All this data is at the point of action."
*This story has been updated to reflect that Gheorghe and Rizzo were not the only cofounders.