Demand Side Platforms: Making The Most Of Your Big Data In Display
Big data seems to be the buzzword around our industry these days, and is there any wonder based on the recent explosion of data in terms of growth, availability and use? IBM does a pretty good job of explaining big data and I’m not one to mess with a good thing. According to the company’s […]
IBM does a pretty good job of explaining big data and I’m not one to mess with a good thing. According to the company’s definition, the three core components of big data are volume, variety and velocity.
Volume refers to the growing mass of terabytes and petabytes being generated and converted into analysis and real time decisions.
Variety refers to the structured and unstructured data pouring in via thousands of sources that is analyzed in new and creative combinations.
Velocity simply refers to how quickly the data is being produced and the speed at which it must be processed to serve useful tasks central to marketers’ needs.
With all of this big data pouring in, a question remains, how do I make the most of big data?
Today many demand side platforms (DSPs) are working with brands of all sizes to redefine what is possible when integrating big data across marketing channels.
The net result is that marketers have granular control over every impression that is served resulting in faster scale, improved ROI and significantly deeper insights into their target audiences. This is all due to the centralization of key vital components:
- Data sources, volume and structure
- Auto algorithmic machine learning and campaign decisions
- Inventory availability
- Reporting transparency and insights
Centralizing Valuable Data
Data can come from many sources at varied volumes and in varied formats. In an effort to deliver rock star marketing campaigns that are data driven, DSPs are working to centralize data sources into single portals for analysis, and advanced customer intelligence that is actionable. Data sources you should look for in a platform include:
First-Party Data – We all know the feeling of ordering from that barista at Starbucks who always remembers that you like a grande white chocolate mocha with skim milk, no whip and extra white chocolate. It makes you feel important, not a bad feeling at all.
Take a lesson from the barista — use existing site traffic behavior to show your consumers that you know them. The most common first party data is online and offline CRM data.
Historically, a data management platform (DMP) was needed to sift through, categorize, and segment first-party data so it could be received by a DSP.
Today, the right DSP can marry all of this data without needing a DMP to clean it.
Additionally, many of the newer DSPs allow publishers to place data collection pixels on their site to collect a huge range of element level data including: pages visited, SKUs viewed, searches performed, referring traffic sources and keyword searches, time on site, etc.
All of this data can be leveraged exclusively across publisher campaigns to customize marketing message, frequency, channels and much more.
Third-Party Data – This is the most common data used in targeting new audiences for branding, engagement and direct response campaigns.
Third-party data is purchased and passed through as data costs, as it is acquired through non-exclusive and exclusive data brokers.
While this the most common data used by first- and second-generation DSPs, it has performance limitations. This data varies greatly in transparency and most often leaves publishers taking the good with the bad as they buy, optimize and report on a large data set (fixed segment) levels.
DSP Sourced Data – Often third-party data segments inhibit the ability to effectively leverage data to achieve the desired outcome.
For this reason, newer DSPs are developing data agreements across the exchanges and publishers to source their own “element-level” data. By sourcing their own data, they can gain transparency into every data element in its raw “unstructured” form. This is the opposite of prebuilt data segments.
One example of this is DSPs managing intent-based campaigns leveraging known consumer searches performed on the web.
DSP-sourced data in an unstructured format allows element-level DSPs to treat every keyword search individually, resulting in keyword-level bidding, messaging, recency and reporting. In short, you can manage the display campaign at the keyword level — just like search.
A Window Into Trends – Auto Algorithms
It was not that long ago that marketing teams were up to their nose in Excel documents analyzing mounds of campaign data — manually reading the tea leaves to connect the desired action to the right customer.
Today the effective DSPs are leveraging machine learning in a real-time environment to find data connections previously hidden and translate these insights into actionable campaign decisions.
Marketers no longer need to labor over campaign details and tactics. Set the budget, define the goals and the best platforms will engage in multi-variant campaign optimizations that get smarter with every impression, leading to optimum performance in a relatively rapid timeframe.
A DSP allows you to centralize media buying into a single portal unlocking billions of impressions across hundreds of thousands of domains and sub-domains while capitalizing on big data. Everyone knows that DSP real-time bidding (RTB) inventory is commonly accessed via exchanges. This has led to a common misconception that DSPs somehow only have access to remnant inventory that publishers and networks could not sell.
The reality is more and more publishers are making their own inventory available to platforms as DSPs expand their own supply side platform (SSP) functionality. Even leading ad networks known for “premium inventory” are expanding their SSP capabilities to allow DSPs to rapidly access their audiences in real time. Publishers have amazing control with the SSPs and exchanges, including setting floor prices for select pages, time of day and more. With this kind of control, there it’s less likely that pages will be withheld as private or premium.
Leveraging data to serve the ideal impression to the ideal consumer, with a personalized message on pages containing specific content and keywords sounds like the definition of premium.
Transparency & Insight
The key to making big data actionable is turning the data you collect into a two-way conversation with consumers. Think about Facebook and the ads you’ve seen for Nike’s newest tennis shoes ever since you added “running” to your likes.
Element-Level Causality – Furthermore, the real benefits come from gaining insights into what content is the most successful with your target audience and how to turn the target audience into customers. The era of black box optimization and campaign decision-making are rapidly coming to an end, as CMOs begin to require transparency.
Real-time Interpretation – Part of maximizing big data is leveraging reporting on an impression level to discover key insights, tactics applied as a result of data interpretation in real time. With transparency of data in reporting there can be large scale cross channel marketing decisions and improvements.
Maximizing Big Data Example
Let’s say you are a publisher or network with 10 properties focusing on college football. Because the content of your websites are laser-focused, your targeting can be also, but only if you have instant access to data during the execution of the campaign.
Through sending unstructured first-party data to a DSP who can combine and analyze it alongside DSP sourced data, you can execute each campaign to the exact target, and you can see how each element is performing during the campaign.
For example, let’s say you want to target only SEC fans in Georgia and Alabama. If you were to build a segment with your first party data, you would only see how the entire campaign is performing. However, if you loaded your raw data into the DSP, you would be able to see if Auburn fans were responding better than the University of Alabama fans. This would allow you to continually optimize, not simply watch a static segment run.
Next, you would be able to incorporate any offline data that you may have in your customer relationship management (CRM) system. Perhaps you also have subscribers to a print magazine or newsletter. Importing this CRM data into the DSP means that you can target the audience that is reading your content offline when your DSP sees them online browsing websites outside of your own properties.
In short, big data is most powerful when it is most actionable. Today, there are some incredible platforms that allow marketers to harness this power to deliver amazingly effective marketing campaigns and contribute insight throughout their entire company. The era of big data is now.