How MiQ’s DPI automates the bidding process
Digital, is one of the most sought after businesses around the world and that has remained unchanged for over fifteen years. Certainly, the market is all about identifying and reaching the right audience. Since the advent of programmatic technology in advertising, several methodologies have been developed to track user behaviour across the web and identify them as prospective targets for any digital advertising (ad) campaign.
At Media iQ, our data scientists explore different dimensions of available data and innovate constantly to identify the target audience.
But is that enough to reach out to these viewers in a single go? The simple answer is `NO`.
What the market needs is a data provider instance (DPI) to automate the bidding process. Sumit Srivastava, Software Engineer at Media iQ, throws more light on our DPI.
While most players in the market, rely on manually setting up bids directly on relevant ‘consoles’ (MediaMath, AppNexus etc), the Product and R&D team at Media iQ Digital have already engineered a unique way of automating the whole bidding process without manual intervention. Our DPI allows our analysts and traders to use a ‘bidder’ flexibly.
We layer over 500 variables on every ad impression. In other words, we use a huge number of variables to differentiate consumers of a given product category. To extract the most accurate results, we need a ‘ sound method’ to determine which variables will be most valuable in distinguishing different groups of consumers.
At Media iQ, we analyse incoming impression request details forwarded by DSP’s (Demand Side Platforms) to extract parameters such as site-domain, keywords from the URL, IP address and corresponding geographical details. For each of these parameters, AiQ mines predefined ‘modifiers’ (if any), put together by our analysts.
These modifiers are packaged and then sent back to the DSP as responses which in turn manages the bid by multiplying it according to the `modifier` defined. What gives us a competitive edge is that this whole process approximately takes less than a millisecond.
Most importantly, this piece of software can handle around sixty thousand requests per second per instance making our campaigns immensely efficient and delivering high performing results for the business.
For example, we recently ran a campaign for a truck rental company in the US. We wanted to see which states in the US were most likely to convert, and found that consumers in DC were most likely to visit the site.
Using the DPI, bids were modified up in regions like DC to FL. Also modifiers were not applied across some regions and were modified down in the rest. When the three groups were compared, we found that the regions where the bids had been modified up performed better in terms of conversion and efficiency than the ones that had no modifiers or where the modifiers were lower.
Therefore, modifying the bids using our DPI, not only enables accurate analysis of region wise performance; ensures the message reaches the right audience, but also impacts the overall campaign performance significantly.