Real-Time Real Talk: Log-level Data 101
“Real-Time Real Talk” is an ongoing blog series that seeks to clarify the “what”, “why”, and “how” behind ad-tech innovations.
In our latest edition of “Real-Time Real Talk”, we wanted to cover a topic that is not name-dropped at every ad-tech conference, seminar, or webinar but is critical to digital advertising success: namely, the topic of log-level data.
What, in plain English, is log-level data? How exactly are log-level data points different than other, more “traditional” kinds of data reporting out there? Why is there a growing need for buyers to learn its fundamentals – as well as to master its intricacies? Also, and while we’re on the subject, what’s a good, clear example of how log-level data can help a marketer succeed in their campaign.
There’s a lot to unpack here, so let’s get started.
If “log-level data” were an actual term in the dictionary, how would we define it?
Until that happy day when “log-level data” becomes a household word, here’s a description we think works well: Log-level data can be defined as the the collection of an abundant set of data points gathered from the moment a programmatic advertisement first appears on-screen to the moment it disappears off-screen.
So if that’s what log-level data is, here’s the next question: what does it do?
Log-level data does more than just one thing. Likewise, and it needs to be said, one or two log-level data points don’t really do much of anything! But once those data points begin to add up, they also begin to serve a practical purpose: The better an advertiser can understand their audience, the better their ability becomes to serve the right advertisement to the right, individual end-user at the right place, moment, and circumstance. And, in doing so, the greater the chances become for that ad to engage the interest of that user.
So how does this work in practice? For starters, advertisers can use log-level data to see which people – at the individual level – are engaging with their ads the most in near real time. This lets advertisers separate those customers most likely to buy their products or services from other consumers less likely to do so. And knowing which individual users to target more frequently helps advertisers determine their return on ad-spend (ROAS).
Seems to make sense, but how is log-level data different from the data most marketers receive and analyze?
Here’s the thing that sets log-level data apart from other types of data: its real-time actionability. Unlike the aggregated data sets that most marketers receive and review from their agencies or DSP partners, log-level data is raw. It’s real. It doesn’t just represent a set of common and basic stats that can help inform the generic, day-to-day decisions of media traders. It’s fast-breaking intelligence that presents trends even as they’re developing, peaking, and subsiding.
Aggregate data is still important to have. But it doesn’t show nearly enough of the full picture of what’s happening. It shows things like the average number of CPCs or CPAs occurring over the course of an hour or so. It doesn’t offer individual-level performance or show an expanded set of data points that marketers can use to engage meaningfully with individuals.
Log-level data gives advertisers access to information such as a user’s operating system, device type, or a particular domain or app they’re visiting. It provides inventory data that pertains to a site, or an exchange, or where an ad is serving. It affords geographic specificity: a user’s longitude, latitude, and zip code – the kind of granular data that marketers wouldn’t have any use for when processed in bulky, hourly, batch datasets.
Furthermore, when using proprietary information available on the AppNexus platform, log-level data shows other, more complex attributes as well: factors including ad viewability, estimated average pricing, estimated clearing price – not to mention specific creative details and price quotes.
Aggregated, batch-processed data lets marketers only see the forest for the forest. It doesn’t let them see the individual “trees” of data fluxes, peculiarities, and anomalies. In contrast, log-level data lets marketers “toggle” between seeing aggregate patterns and individual trends. It lets them see the forest for the trees – and then see the trees as purely forest again. In short, log-level data can be shaped and sculpted in ways that aggregate, bulk-package data can never be customized.
Advertisers can then use log-level details and inputs to ask a lot more questions than they ever previously could as to how smartly they’re putting their data to use – questions that can only lead to subsequent development of ever-more thorough, ever-more constant, ever-more reliable opportunities to put data intelligence to good use: for example, in real-time dashboard reporting and in real-time updates. While aggregated data is still a must-have “big picture” tool, only log-level data gives an advertiser pointillist intelligence that lets them act on new, emerging audience opportunities and trends – or follow up if they notice their campaign isn’t performing as they predicted or expected.
That’s a nice overview. But let’s dive more into the details. What are some of the specific benefits of collecting and analyzing log-level data?
Here’s where we get into the fun stuff. Regardless of whether a marketer’s in automotive or online retail – and regardless of their level of technical sophistication – the tech and tools are in place right now to collect, analyze, and make actionable decisions based on log-level data.
Sophisticated advertisers can apply log-level data to tie back ROI to every single impression in a given campaign. They can see the cost they paid for a single impression and what that single impression delivered back in return. They can see the value that individual publishers place on their creatives in relation to other, different creatives from other, different advertisers.
They can witness the point-by-point journey of individual consumers from click to conversion – giving them the consumer-centric focus that’s been so pivotal to the success of companies like Amazon. They can see what users saw what ads, how those users chose to engage with those ads, and whether those users’ levels of engagement were sufficient (or insufficient) enough to reassess how much they should spend on ad campaigns in future, similar circumstances.
Log-level data – combined with proprietary, pre-existing data – lets advertisers optimize their campaigns to engage more and more users at the individual level with greater individual precision. For example: using CRM data, to segment customers who demonstrate a specific attribute – say those who selected a specific upgrade to their new car - automotive advertisers can now tie complementary upgrade offers to those same customers using log-level data points to optimize customer lifetime value.
What are questions I might want to ask internally and of my vendors?
We totally get why advertisers would want to keep to prior agreements, obligations, and business arrangements with longstanding partners. But that doesn’t mean you shouldn’t be asking questions of your agency or DSP if they aren’t able to stream log-level data using their own technology. If their data analytics teams aren’t privy to log-level, real-time insight, then how can one be completely sure the insights they’re offering are, in fact, of any actionable worth?
AppNexus is one of the few tech providers to offer its clients and partners access to their own log-level data. In the programmable advertising economy of tomorrow, we believe log-level data will be the building material of choice for buyers worldwide. It will be the glue that’s able to hold together a rich, holistic portrait of a consumer across multiple and disparate devices, apps, websites and usernames. We offer all our customers the option to sign up for a download of their log-level data not only for the sake of transparency – but because we genuinely think that the greater insight they have into their own business, the greater their business will be. We also believe the more directly connected an advertiser is to their data, the greater the chances they’ll be able to connect in resonate ways with their consumers.
AppNexus offers log-level data to its clients – as well as an in-house technical support team that can help ingest all the information you need to get your campaigns up and running at the “log level”. We also can provide you with the right amount of data science and how to best take action with the data you have – as well as a world-class services team to show you how to maximize the impact of your data.
And for those prospective clients and partners who possess more advanced technological wherewithal, we’ve invested heavily in streaming data capabilities that let you make faster, data-informed decisions to drive future performance of your campaigns. You can find out more about our data-streaming capabilities by having a look at numerous articles in this publication, and downloading our e-book.