The Data Behind Demand: How The Information Revolution Is Powering Programmatic Advertising
:: By Venkat Krishan, YuMe ::
There's a hidden world behind today's website advertisements. Extraordinary feats of analysis and computing power, along with lightning-fast bidding wars for crucial online marketing real estate, help determine what Internet and app users see when they turn on their devices.
It's imperative that marketers understand exactly how this complex process works -- and how big data technology can help them identify and reach their potential customers within a dauntingly large population of several billion Web users.
With programmatic marketing, due to its size, scale, and speed of information, the complexity of the decisions that advertisers face has increased at an unprecedented pace. And marketers continue to struggle with how to best harness their data to take full advantage of programmatic technology. In fact, late last year, Forrester reported that only 23 percent say they understand programmatic buying and use it to execute campaigns.
To help marketers evaluate their seemingly ever-growing list of options, there are a few important considerations to keep in mind:
Know Your Audience: The first, and perhaps most important, challenge advertisers face is identifying exactly who their potential targets are online. Though a few large publishers can boast billions of users within their direct network, some demand-side platforms (DSPs), or companies that facilitate real-time bidding for online ad space, can identify the similar audience pools using advanced software algorithms built on data sets that aggregate anonymous demographic, contextual, behavioral and psychographic insights. By pinpointing the profile and characteristics of their ideal audience, a marketer can maximize ROI and reduce waste.
Think Quick: Demand-side platforms (DSPs) see an incredible volume of activity, even within miniscule units of time. The number of real-time bidding transactions can now exceed 1-2 million queries, or bid requests, every second. This is a substantially greater number of bids than even some of the world’s most popular websites receive. In contrast, it is estimated that some of the largest publishers process only 40,000 search queries every second. DSPs have the difficult task of scanning through 2-3 billion users to find the “right” users for those 1-2 million bid requests.
The Eight Second Rule: The DSP brokering the sales of programmatic ads must select the right ad campaign at the optimal price within a very short time window. The aim is to successfully render an ad on the user’s screen within one second, because if the ad doesn’t load quickly, there is a risk of ruining the user experience. Once 8 seconds have passed, certain research indicates that the viewer has likely been lost.
Assess Speed at the Margins: The time margins on these transactions are incredibly slim. Websites selling ad space expect to receive bid responses from DSPs in less than 100 milliseconds in order to run an auction on all bid responses. The communication between advertisers and DSP servers takes 50-60 milliseconds due to network latency, meaning a DSP’s internal operations must be completed within a lightning-fast 3-5 milliseconds. The user on the Web page or the mobile app may not even realize the complex yet nearly instantaneous process that goes into the ads they see.
While the nuts and bolts described above may be more information than many readers need, it is increasingly critical that marketers digest and fully understand it. Programmatic ads are bought, sold and distributed based on the real-time analysis of billions of Web users. Not every DSP has the same degree of processing power or the same level of mastery of big data analysis methods. It’s essential that marketers understand the hidden world behind Web advertising in the big data era – and how to make it work to their advantage.
Venkat Krishan is the SVP, Product at YuMe, the global audience technology company powered by data-driven insights and multi-screen expertise.