Having worked in managerial roles in tech ranging from tech startups to a stint at Microsoft, where he worked on the Windows operating system, Stephen Purpura was convinced by 2001 that data products were going to be the next big wave.
Since that time, he’s been working on nothing but big data-related projects – even plumbing the depths of graduate studies at Harvard and Cornell to discover the most effective methods to enable big-data analytics throughout corporate America.
“At the time, I was watching Google and Yahoo begin to benefit from measurement of interactions with customers,” said Purpura, who pointed to use of simple A/B tests. “The data that they got back was analyzed and incorporated back into the company.”
This idea wasn’t lost on Purpura for the last decade, so his new company, Context Relevant, founded in 2012, built a platform that allows clients to construct experiments and analyze them in near-real time so they can dynamically change how they price things – such as display ads. He pointed out that the company’s ad products are still in the early stages but that the firm already is working with a handful of unnamed customers to “prove out” a product line.
As of late, the machine-learning software company employs 20 people, mostly engineering types, and recently announced a multimillion-dollar round of financing with a pitch that pricing across several high-value verticals remains a significant challenge.
“The initial go-to-market strategy involved online travel agencies (OTAs), such as Concur, where there is a huge amount of disruption occurring,” Purpura said. “The reason that OTAs are interested is that they have a need to reprice inventory at a rapid rate because their competitors are doing it. Priceline is repricing in seconds. If you’re not keeping up, you are losing revenue, which translates into profits.”
Given the shifts in that market, online travel agencies provided Context Relevant what amounted to a “test bed” that also led them to ad exchanges and the analysis of Web logs and Adobe Site Catalyst data. Speaking from a Web publisher’s perspective, Purpura said, “It helped us understand editorially how to build content that will sell as well as how to maximize engagement with their customers.”
Though software-as-a-service is one option for customers, the company’s on-premise business model mixes with customer data which clients might not normally given competitive and privacy reasons, for example. Dealing with the sheer scale of data is another factor for on-premise installations for large website customers.
In drawing out an example of how his company handles the pricing of display ads, Purpura kept it at a high level. “People who advertise teeth-whitening products spend a lot of money for placements,” he said. For publishers, “there’s always a trade-off between serving ad buyers who are willing to spend a lot of money and what your site’s visitors may or may not want to see. Our software helps dynamically calculate that.”
The company is quick to point out that its foray into ads isn’t about replacing ad exchanges, but rather informing buyers or sellers how to price, given the need to show the right ad in the right place at the right time. Content is a part of the mix, too, as the software can automate the ability to A/B test whether specific content appeals to people.
Keeping his company’s options open, Purpura said he doesn’t see the software as an advertiser or publisher or vendor solution. “It’s about giving the people in control of a site the capability to understand how to best engage with their customers, whether it be through advertising or through content,” he said. “It’s about maximizing growth and revenue. The only way you can do that is by running high-velocity programs to experiment on what works with your customers and analyze the results. We allow companies to increase their cadence of that, hopefully, by 10 to 100X. That’s our goal.”
In the ad-tech “stack” conversation, Context Relevant sees itself working on behalf of a marketing analytics team and inserting a measurement layer at different points to help people understand and predict where they’re getting value out of their own ad stack.
Reviewing the company’s client opportunities, Purpura explained, “It turns out that our technology is useful in finance. It’s useful in analyzing pricing on the Internet — because Amazon is repricing everything every three seconds, and other companies need technology to compete with that. Similarly, Priceline is repricing constantly, and all the online travel agencies need the ability to compete with that. Likewise, in the ad exchange business, pricing trends shift quickly. It helps to have technology like ours to understand the underlying factors that contribute to reader viewership and what they’re happy with and not happy with.”
He added that “futures” may become part of their ad exchange pricing product line. “Helping companies understand how to engage in not only the current markets that exist,” he said, “but also in forward markets is clearly a use case for our product.”
