Getting started with log-level data in digital advertising
Two weeks ago, when the ANA released its Programmatic Media Supply Chain Transparency Study, it shone a light on the importance of log-level data (LLD). But from my experience, very few marketers are using LLD to improve their programmatic advertising.
Log-level data provides a very granular view of the data from each individual impression in your campaign, surfacing many fields of data that may otherwise not be available. With it, marketers can understand more about each individual impression they serve, such as its viewability, its geo data, the device and operating system it appeared on, and its price.
However, by default marketers do not get that level of information from their AdTech partners. DSP or ad-server platform reporting is aggregated impression data. But by going to the effort of analysing individual rows you can open up advanced capabilities and drive greater efficiencies in your programmatic advertising.
Use cases of log data
There are many ways LLD can help marketers improve programmatic advertising. Here I will look at just three: financial transparency, improving your route to supply, and custom algorithms.
1. Financial transparency
Most digital marketers by now know about the ‘unknown delta’ of unattributable ad spend first reported on in the ISBA/PWC programmatic transparency report. In the ANA’s most recent report, it showed that this discrepancy can be virtually eliminated by utilising a full path log-level analysis. Accessing log data allows you to do this yourself to understand where in the supply chain your budget is being spent before it reaches the publisher.
2. Improving your route to supply
LLD can be used to improve understanding of how you access programmatic media supply. For example, in work carried out by TPA Digital, we discovered brands often face different SSP take-rates based on the publisher, format or device an impression is served on. LLD can help brands bring this to light, so they can optimise their supply paths towards higher quality and lower costs. 3. Custom algorithms
Custom algorithms allow brands to enjoy bespoke functionality in their programmatic media buying that is aligned with specific goals. If you were to just rely on aggregated data your custom algorithm wouldn’t be able to benefit from knowing which impressions weren’t optimal. By using log-level data, marketers can fine-tune their buying based on the data held within each individual impression.
What are the challenges in starting out with log-level data?
Talent and tools
Using log-level data requires the ability to process and analyse enormous data sets with often hundreds of millions of rows of impression data, combined with more than 100 columns per row. For brands, this means a well thought-out ‘Extract, Transform and Load’ (ETL) process is required. Different sets of log files will have their own data formats. There won’t be a consistent format for log data across vendors; each will have its own fields and its own naming conventions. So data scientists will be required to make this data usable.
But data scientists alone won’t make this data valuable. You also need people who understand media, what you should be looking for and how it can be used. This requires better collaboration between teams and clear briefs and objectives set.
Legal and Compliance
Getting access to log data takes time. We’ve become accustomed to the necessary hoops to jump through when sharing of data is required. This will be no different. But once completed, there may still be challenges out of your hands. For example, once you get SSP access, some publishers might be missing because they’ve opted out with the SSP. Prepare to have blind spots.
How to start using log-level data
Despite the above challenges, marketers can begin to make the most of the opportunities of log-level data. It does require the right talent and experience (and is something TPA Digital can assist marketers with).
Understand what log data can do for you, and what you need it for
Who are the providers that can offer it to you? Identify where you can access it already
Prioritise access to LLD across your supply chain:
Get LLD access from your DSPs first, as this is the most important
Then look at your SSPs. First, understand which SSPs are providing market, publisher and format coverage, and transparency
Then have conversations with them to get the log-level data
Include log data when sourcing AdTech
Implement this advice when looking for new AdTech partners. Within each type of advertising technology, different vendors will have different defaults and data on offer. For example, some DSPs will provide more data fields in their log data (breadth). But others can offer more granularity in their data e.g. full URL vs domain for contextual targeting (depth).
Therefore, understand the use cases of log-level data you want to employ, and the data you need to make that happen.