Back in 2014, as a new member of MuleSoft’s product management organization looking to prove himself, Anton Kravchenko had no idea of the nerve he was about to strike when he inadvertently tapped into an explosive nexus of tech industry sea-changes; cloud computing, the democratization of IT, and the unquenchable corporate demand to economically bridge previously difficult-to-bridge data silos. The resulting invention would not only go on to win first place at a prestigious hackathon, it would join an exclusive club of industry innovations that start at tech companies as skunkworks projects and end as major revenue generators.
“The problem was surprisingly simple,” as Kravchenko described it to ProgrammableWeb, “that it was hard to believe it was still unsolved by existing solutions.” Kravchenko was referring to a category of software that’s commonly referred to as “Export, Transform, and Load” software or “ETL.” In a nutshell, ETL is largely associated with the programmatic (scheduled) or manual movement of data in bulk between dissimilar but specialized systems. Thanks in part to that dissimilarity, the process often involves cleansing or transformation of the data as it is moved between the source and destination systems.
For example, if a national retailer must move data between its NCR point-of-sales systems and its Oracle Financials system on a nightly basis (after its stores close), it might rely on an ETL tool to unload the data from each of the geographically distributed POS systems, consolidate that data into a single batch, adjust the data’s integrity (ie: normalize redundancies) for compatibility with Oracle, and finally upload it to the financials system.
As Integration patterns go, ETL is practically an anti-pattern to modern day API-led integrations where the same data and transformations might happen in real-time as each sales transaction is completed. But ETL is so commonplace and in many cases perceived as bulletproof, that organizations are loath to consider replacement. In fact, many IT analysts view ETL as the biggest driver of growth of the data warehousing market — a market that’s expected to exceed $50 billion in revenue by the end of this decade.
But as hot as ETL software was at the time that Kravchenko was hatching his invention, it was also an enigma to a new breed of concerns. Not only did most of the existing tools focus their attention on on-prem systems (or the ETL tools themselves were on-prem solutions), they were also very much the domain of IT departments. The idea that a line-of-business knowledge worker might attend to an ETL-process, perhaps manually getting involved in the movement, cleansing or transformation steps, was rare.
Meanwhile, these shortcomings presented a very particular challenge to the customers of Salesforce. By 2014, across the many tenants of its multi-tenant CRM solution, Salesforce was unquestionably the largest single custodian of customer data on the planet. Relative to the previous generation of on-prem CRM solutions, all of that customer data was in the cloud. And the majority of Salesforce users were non-IT people.
“There was no question in my mind that the customer data of the sort that Salesforce was managing for its clients was also the sort of data that organizations needed to both import and export to and from other systems” said Kravnchenko. “Even if the only reason for moving it was to manually perform a bulk data update and then to re-import it.” At the time, Salesforce was good at many things. But easily exporting and importing data, mapping fields between dissimilar systems, and doing bulk updates (oftentimes the equivalent of a global search and replace) were not among them.
According to Kravchenko, Salesforce had a downloadable tool. But he suspected that similar to the way Salesforce’s customers were drawn to the cloud for CRM, they would also appreciate an ETL-like tool that itself was totally cloud-based.
The idea for dataloader.io was born; a purely cloud-based solution that democratized ETL by making it possible for non-IT people to do little more than visit a web browser in order to easily move and manipulate data between Salesforce and other systems.
From there, dataloader.io took off. After first being honored as the winner of a Salesforce-produced hackathon in 2016, Salesforce’s customers acknowledged the utility of dataloader by making it the top-selling solution of its type on AppExchange; Salesforce’s public marketplace for third party solutions designed to work with Salesforce’s various clouds. “99 percent of the users pay nothing” remarked Kravchenko of dataloader’s freemium business model. “But for the one percent that pay, it’s a good revenue stream for MuleSoft.”
Since its invention, dataloader’s star has continued to rise among Salesforce users and at MuleSoft. At the time this article was written, dataloader had an average 4.5/5 star rating across nearly 600 AppExchange reviewers. And as dataloader’s star has risen, democratizing integration for thousands of Salesforce customers, so too has Kravchenko’s. Whereas his journey started with MuleSoft as an associate product manager, he is still with the company, now a part of Salesforce, as a director of product management focusing next on the company’s innovations to democratize automation. But the passionate technology professional is still very modest about his accomplishments. “I’m just glad we stumbled upon a big pain point and eliminated it for thousands of people,” said Kravchenko. “It really motivates you to do it again.”