Parveen Jain is CEO of data technology company, JovianDATA.
AdExchanger.com: On your site, you say: "The more dimensions, data sources, and KPIS - the better." How can that possibly be true? Agencies, for example, are sitting on mountains of data. Are you suggesting they need to harness all of it.
PJ: JovianDATA technology creates multi-dimensional models with 10’s of dimensions and 10’s of measures (key performance indicators) on massive amounts of data (100’s of Terabytes) which processes multi-dimensional queries in seconds. As a result we can run expeditious and accurate analytics to any level of granularity without sampling the data with the lowest total cost of ownership (TCO). This allows us to invite our customers to send us “more data with more dimensions, data sources and KPI’s”, something that no other vendor can claim.
We know that once customers have to sample or aggregate the data, they lose the granular nuances that are critical to identifying and acting upon opportunities in the data, and thus prevent them from maximizing the return on their advertising spending. JovianDATA has removed this barrier, and our customers discover insights that allow them to take in-flight actions to optimize their campaigns, during the campaign.
Analyzing data at its lowest granularity allows advertisers to discover large segments or phenomena that might get lost when data is aggregated to higher dimensions. Analyzing data across more and more dimensions helps us expose the correlations between various dimensions.
How is JovianDATA differentiating itself from other data providers?
JovianDATA is bringing a game changing technology to handle large data across the entire analytics stack at a very low cost. Analytics is exposed to end users in the form of applications that provide responses to their business questions.
- Applications (solutions), and not just a platform or tools like other vendors. We have built ‘Campaign Effectiveness’ and ‘Media-mix Optimization’ applications to help marketers answer the business questions essential to running their campaigns most effectively.
- End-to-end solution. We take away the needs for CapEx and an army of engineers to assemble, manage and run a multi-vendor analytics stack. We offer the solutions through SaaS offering that gets a customer in production in weeks, compared to multiple quarters that it takes with the alternate solutions.
- Unprecedented low TCO through dynamic cloud provisioning and an integrated SaaS model. The dynamic cloud provisioning is one of JovianDATA’s disruptive innovations, as it gives us 70% to 90% cost savings versus conventional technology stacks.
- Ability to handle massive data volumes (e.g., tens to hundreds of terabytes of data). We thrive on massive data sets as it allows us to extract actionable insights for our customers, which is not possible when you sample data. Our customers (Advertisers, Agencies, Ad Networks and Publishers) now have the ability to answer questions about their business, campaigns and customers that either technology or economics prevented them from asking in the past. This is allowing them to optimize their campaign performance and media mix allocation.
- An integrated ETL layer that allows us to align, normalize and cleanse disparate data sources. JovianDATA also provides data enrichment capabilities that allow customers to create advanced data models for segmentation and attribution, and advanced measures such as unduplicated reach, frequency, recency, ad sequencing and n-tiling.
- Integrated multi-dimensional analytics that can handle any number of dimensions, data sources, measures, and key performance indicators. We’ve patent-pending technologies that allow us to use MDX (multi-dimensional modeling) on 100’s of terabytes of data. Historically seen as an SQL problem – which needs to pre-aggregate data and pre-build specific indices – this can now be answered using a spreadsheet-like MDX environment. For some complex business questions – like understanding the ad survivor rate for a video ad across multiple video ads on a second-by-second basis – we have taken what would have required 36 hours in SQL and reduced it to less than 5 minutes in MDX.
- Very fast query response time. Most of JovianDATA’s queries, across terabytes of data, come back in 5 seconds or less. We do not have to pre-build specific indexes based upon the types of questions that users want to ask. Users can ask completely ad hoc questions and get the same level of query response. This level of performance lead one of our customers to call it “analytics at the speed of thought”. To them, this meant they could have an interactive question-and-answer dialogue with their data.
How do you view offline data? Can it provide huge performance benefits for online advertising and are you selling (or re-selling) offline data today to marketers?
JovianDATA can handle data from any number of sources, representing on-line or off-line activities. We believe that offline data is important, especially demographic profile data and sales data, to completing the full view of the customer engagement funnel. At the same time, we believe on-line campaigns can provide insights that could be important to planning and tuning the overall advertising spend, including the offline spend. Whether it is offline media spend for TV, radio or print advertising, or the offline sales data, you need to have the data that represents the breadth and depth of the customer engagement process. This is critical whether you are focused on lead acquisition, conquest effectiveness, customer retention or customer maturation.
What are you seeing in the demand-side platform space today? Still early? Are they using data to its fullest potential?
We believe the limitations in data analytics in terms of accuracy, granularity, timeliness and cost will remain serious obstacles in providing effective advertising solutions. Statistical models based on sampling and/or curve fitting will peak out in their abilities to deliver actionable analytics. As the marketers demand more-and-more accountability and addressability, vendors like JovianDATA with business rules driven and action oriented solutions built on game-changing analytical platforms will be the only ones to fulfill those requirements.
How does the online publisher work with JovianDATA?
JovianDATA is engaging with several customers today. We have a very simple, three-step process for engaging with our customers:
- Give JovianDATA access to your marketing data. That includes data from the full customer funnel including impressions (from your ad server), clicks (from your web site analytics) and conversions (from your point-of-sale system). The data can either be ftp’ed to us, or give us your user name and password and we’ll pull the data directly from your ad server or web analytics provider.
- Help us understand the business questions that you are trying to answer. This allows JovianDATA to model the data in the most user-efficient structure, and to build some initial analytics (guided analysis) to answer those questions.
- Get insights. When we provide access to the customer’s data, we bundle our Campaign Effectiveness application so that the customer has a complete solution on day one.
Today we can get customers from Step 1 to Step 3 in 4 to 8 weeks, depending upon the complexity of the business questions. And as we build out our library of data enrichment models and analytics, we expect that timeline to shrink to 2 to 4 weeks.
What is your specialty in terms of data sets? Auto intenders, for example?
JovianDATA has direct experience in dealing with all sorts of different data. We’ve been working with Ad Networks, Agencies and Publishers and all of their respective data sources. At this point in our development, we really don’t have a specific area of industry focus.
How does your pricing model work?
We price based upon the total volume of transactions (total of impressions, clicks, engagements and conversion transactions) that you want to keep for analysis purposes. Some customers want only 3 months of data, while others want a full year. For a 3 – 5 billion transactions, that cost would be in the $10k to $20k monthly cost range.
Do you see benefit from data exchanges currently and/or in the future?
Since we do not actually own the data - the data still belongs to our customers – we do not sell through data exchanges. But on the topic of data exchanges, we are a big fan. Anything that can bring more data into the digital marketing intelligence processes, the better. As we mentioned above, we want as much breadth and depth in the data as possible to tease out audience behavioral insights and optimize campaign performance.
What role do you think search retargeting data will play in shaping the data landscape in 2010?
Search retargeting adds a layer of relevancy to banners served to searchers after they leave your client's site. By recognizing a user's initial need based on the search keyword used, the value of a search listing should increase by enhanced likelihood of a conversion.
We believe search retargeting data is valuable for increasing the efficiency of display advertising. The interplay between display and search is one of the most complex problems in analytics. In our work with agencies, they have had broad goals of coordinating search and display campaigns but so far they have not had an analytics solution which exploits the interrelationships between search and display. Our Media-mix Optimization application has changed that. In this application, we analyze data at a per-user granularity to compute the lift in a display campaign from prior paid search clicks. This enables Advertisers and Agencies to co-ordinate their search and display campaigns to retarget visitors who have shown search click propensity.
Follow AdExchanger.com (@adexchanger) on Twitter.
Email This Post