What do you do if you have real-time bidding software that's non-core to your business, but could still be an asset to young ad companies faced with building their own bidders from scratch? You take it open source.
That's what Montreal-based Datacratic has done with the release of a new RTBkit framework, available at RTBkit.org. Engineers who decide to use the software package, offered under an open source Apache license, can avoid some of the coding headaches required to put in place a bidding system capable of handling tens of thousands of queries per second. The framework's booster say developers will be able to focus instead on differentiated features such as unique optimization algorithms and bidding logic.
"We think this allows people to get RTB out of the way quickly and focus on optimization, data driven bidding strategies, and the dataflows that a particular company might require to bid in a very unique and innovative way," says Datacratic CEO James Prudhomme.
RTBkit is a framework only. Any company that decides to use it will need to pay for servers or rent server space in the cloud to actually bid on exchanges. But it does represent a significant milestone for RTB infrastructure, according to Neal Richter, chief scientist at The Rubicon Project and co-chair of the IAB's OpenRTB committee, who has been briefed on the project.
Richter says RTBkit contains the architecture and scalability "table stakes" for buy-side platforms to get involved in bidding – and that it supports video and mobile from the outset. He pledged support from The Rubicon Project and the IAB OpenRTB group in the form of system testing, code contributions and advocacy.
"The real promise of something like this is people can start with much more of a framework and then get immediately into implementing their data strategy," Richter said. "From an RTB perspective it's a big step forward because you have a referenceable implementation."
Even so, Datacratic is a relative unknown as a bidding platform. It is a predictive modeling specialist that has leveraged its optimization algorithms in a few ways, including through lookalike modeling on the BlueKai data platform, and product recommendation for e-commerce. "We see ourselves as an algorithmic optimization company, not a company that does ad tech components," says Prudhomme.
To be successful, RTBkit must prove it can go toe-to-toe with robust and widely adopted bidders such as that offered by Iponweb, which has been implemented by dozens of Demand Side Paltforms.
And RTBkit still must show it can attract a community of developers that will engage with the software package, and contribute code back into it.
Such a community is what keeps an open source package from obsoleting itself and becoming a quiet corner that no one plays in, according to Richter. He gives the RTBkit project good odds.
"Some open source communities bootstrap themselves. Other open source communities are founded and supported by a company. Datacratic definitely fits into the second category. What they're doing is open sourcing technology that's already in production. It has a much better chance of having a community form around it," he said.
A previous attempt at an open source RTB framework, called Java-Bidder Framework (and built for the AppNexus platform only), may have suffered from a lack of community involvement. The code reportedly hasn't been updated in quite some time.
So how is Datacratic building community? It's setting up a website, email list, and other resources to support developer interaction around the framework, and has pledged to contribute to the code base. For instance it will open source reporting platforms and "augmenters," which let a data provider attach their data to a bid request.
While no one has yet begun implementation on the RTBkit framework, Prudhomme says Datacratic has gotten verbal commitments from new and established companies that lack a bidder but want to add one to their stack.
And then, what of the potential vulnerabilities associated with open source as it relates to digital advertising? Just last week OpenX just shut down its open source ad server, OnRamp, citing a sharp rise in malvertising.
Richter points out that most of ad tech stands on top of open source software, whether it be Linux (operating systems), Apache (web servers), or Hadoop (data applications). While the case can be made that these are "oxygen" required for many web businesses -- in contrast to the "ad mechanics" of RTB and ad serving – Richter still sees no special risks inherent in open source RTB.
Prudhomme goes a step further. Contrary to endangering RTBkit implementations, he believes open source status could will confer security and stability – for instance through a community-based blacklisting system.
"Any real time bidding system would have the potential for security vulnerabilities and other vulnerabilities. If RTBkit is successful in creating a community around it, then the risk is reduced substantially as compared to proprietary bidders," he said.
Asked why Datacratic decided to create the RTBkit framework, Prudhomme says both altruism and self-interest play into it. He says the primary aim is to give a leg-up to young companies that lack the time or wherewithal to develop their own bidder.
But the company could also benefit in a number of ways. Assuming RTBkit manages to scale implementations, Datacratic will glean PR as the flag carrier of a robust open source platform – a halo effect Linux provided to Red Hat.
And RTBkit could provide lead gen for Datacratic's other offerings. "Our real time bid algorithmic optimization platform happens to work very well with RTBkit," he says. "They are free to use it standalone, but RTBkit does make it easy for people to work with our optimization platform."