The Digital Anthropologist: Dstillery’s Tom Phillips on the Post-Cookie World

Tom Phillips was CEO of Dstillery in 2014. This story originally appeared in The Advertising Technology Review in 2014.

Advertising technology company Dstillery, formerly Media6Degress, has raised $26 million in funding and worked with more than 400 leading brands across the globe. CEO Tom Phillips and Chief Scientist Claudia Perlich discussed key principles marketers need to keep in mind when looking at Big Data in a post-cookie world.

There’s been a lot of talk about a post-cookie world.  How likely is it that this would occur sooner rather than later? How will it impact marketers?

The post-cookie world is already here, and is evident by the rise of apps, as well as browsers that limit the use of third-party cookies. Companies have already started deploying alternative approaches that use a “recipe” of tactics to find and target audiences. In addition, numerous companies are looking to provide a true multi-screen marketing solution for advertisers trying to engage with consumers, who are moving seamlessly between devices wherever they go. The cookie represents a declining part of the recipe, as other tactics such as statistical IDs, registered user data and location-based targeting gain traction. However, as much as most people think in terms of absolutes i.e., the “death of the cookie,” the reality is that the cookie will be around for a long time. The biggest challenge for marketers and ad tech vendors is to get non-cookie-based identifiers synched with the tools that measure performance of advertiser campaigns. If analytics providers and ad servers measure only cookie-based conversions, the impact of non-cookie-based targeting technology will be understated. For instance, today’s ad servers measure post-view conversion results using cookie-based technology. In most cases, post-view conversions on a Safari browser go uncounted. This problem has been recognized but considered on the fringe for some time now given the low penetration of Safari. Consumer adoption of non-cookie-based devices and digital media is making this an issue that needs to be addressed more urgently.

We hear so much about data these days, many marketers have become data-blind, believing that any data is good data. What do marketers need to look for when determining which data is relevant to crafting strategy? How should they address the data quality issue?

Claudia Perlich, Chief Scientist, Dstillery

There is an overload on the term ‘data.’ In the world of data science, data is the beginning, the raw ingredient. Imagine making a cake. Data is simply one ingredient — the flour. In order to turn the flour into a cake, you need a baker -a data scientist- and a recipe, which is the algorithm along with other ingredients. The marketer’s strategy is to offer your future mother-in-law a cake. Now it is true that if the flour, the data, is bad, the cake won’t be great. But the right cake may require a whole different recipe. Raw data is just a recording of facts about the world. For instance, did someone go to a website, did they click on an ad, etc. And as such, data is not good, bad, clean or dirty; it simply is. When it turns out that the data doesn’t lead to the results we expected, we consider it low-quality. For instance, what technically might qualify as a click, may not actually be a person that deliberately clicked on the ad in order to check out the offer. It may just be that the computer system recorded something that by technical definitions was a click. Software that has infected the browser could have generated that click, giving someone far away with no interest in the ad power to control the browser.

That relates to what we actually mean by “data,” it seems.

Most often when non-data scientists question the quality of the ‘data,’ it is really a question about something derived from the raw data, something considered an ‘insight.’ When we move to insights, or more generally any form of derived data, things start to get really tricky. At this point a lot more manipulation has gone into it, typically done by someone else. Now you have to figure out whether that means what you think it does. For instance, if someone gives you a number and tells you it’s the average of the age of your customers’ cookies taking action on your site, the number they’ve shared almost surely doesn’t mean what you think it does. It depends on many factors like the percentage of traffic coming from a specific device or browser, the percentage of cookies disabled, whether one removes the zero ages or not, etc. Any number between one hour and 90 days is plausible, but only one is consistent with what you think it means. So the only solution is a strong push for data literacy. Everyone using data has to really know what it means and buying data or getting data from a third-party makes this nearly impossible.

How will the possibility of more screens, increased privacy regulations and the growing importance of content marketing impact traditional models of behavioral targeting? What is Dstillery’s value proposition in reference to the changing needs of marketers in an uncertain ecosystem?

Dstillery CEO Tom Phillips

Traditional models of behavioral targeting help bring scale and relevance to campaigns, regardless of the screen. Traditional advertising models use estimates to try to get a message in front of appropriate audiences. And while mass communication is one tool for building brand awareness, intelligent use of data can create much more efficient solutions for marketers, who are looking to build brands and drive sales. Whether it is television, print, radio or digital, marketers have focused on content as a proxy for trying to find the right audience. Fragmentation of media has helped push endemic advertising. But there is still a huge amount of content consumption, where marketers are looking to hit high compositions of an audience because it’s the best that they can do. In other words, if you’re trying to reach cooks, it’s a lot easier to hit your target audience by advertising on The Food Network rather than American Idol. Understanding consumer intent through tracking and then deploying advertisements through digital means allows marketers to connect with the right prospects for their brand more effectively. This has been done in the desktop space for some time now. Mobile efforts are growing rapidly, and soon we’ll see significant growth in digital TV, radio and outdoor. This shift is resulting in more data, and with it a more complete window into the consumer beyond anything that previously existed. At Dstillery, what we have built for the past five years allows us to capitalize on this shift and effectively activate the data on behalf of marketers. We find ourselves at an exciting time in marketing, one that is great for the marketer. And with the right balance of innovation and respect for consumer privacy, it will be great for the consumer as well.


You may also like