The long term success of any successful analytics platform relies on two items – the usefulness of the data and the time/value costs associated. At Metamarkets we have a dedicated team of experts in the AdTech space who can help programmatic marketers squeeze the most ROI out of their integration. Here are a few tips from our team on how to maximize your investment in analytics:

Work Backwards from Actions

When thinking about how you can best leverage Metamarkets, it’s important to start by focusing on your challenges and questions then work backwards. In the OpenRTB standard there are dozens of different fields to examine, but how many can drive positive changes for your business? It may not make sense to pull in every field if there aren’t direct actions one could take using that information. Knowing who the mobile carrier for a device may be useful to some people, but if it’s not directly contributing to troubleshooting or increasing revenue, it may not be worth the ongoing data storage cost for something “nice to know.”

Go Lean

With any data pipeline, it’s always important to look at what each event sent contains. As part of a standard OpenRTB auction, there are several different dimensions captured detailing a range of subjects. With Metamarkets, one of the billing variables is the total size of data posted, not just the number of events. In order to get the most bang for your buck, go lean and only include fields that you plan on examining in the Metamarkets dashboard. By including excess fields you don’t use, you’re artificially inflating your total volume and bill! Keeping a minimalist record works best for everyone.

Pre-Trim

On a similar note, it may also be advantageous to do some light ETL work on fields before sending to Metamarkets. A good example of this is the Device User Agent field – (“Mozilla/5.0 (Mac; U; Intel Mac OS X 10.6; en-US; rv:1.9.2.16) Gecko/20140420 Firefox/3.6.16”). There’s a lot of info in here, but how much is relevant? If you only care about the operating system (Mac OS X 10.6) why not pre-trim and only send that value? A key theme of any improvement activity is how many small changes add up to big differences. Removing the excess information here in User Agent may only return a 1% or 2% reduction, but done across several items the results add up quickly.

Leverage Lookup Tables

Metamarkets can support using lookup tables to further add onto the savings seen above. A common example of this is App IDs and Names. For instance, you may be working with a Publisher with the following App:

{“app_id”:123,”app_name”:”Metamarkets’ Guide to Adtech and Programmatic Analytics”}

Lookup tables allow you to post a file that matches IDs to Names so you’ll no longer need to include it in your production data feed. Once sent, Metamarkets will load the table into the query engine and do the lookup for you. The end result on the dashboard ends up being the same, but using these lookup tables will allow you to remove the name from the stream. With the Publisher/App example above, there would be an 85% savings! 68 characters may not look like a lot, but multiplied by 1 billion events per day, it adds up.

Cleanup Junk Names

Finally, an often overlooked data strategy to improve usability is intelligently cleaning/grouping fields. The most obvious issue we commonly see is different text cases used. By default, Metamarkets will not group two items with the same text but different formatting. For example – OS Version values of “ios 10”, “iOS 10”, and “IOS 10” would be displayed as three separate items, making it challenging for an analyst to include everything he needs. Bucketing lesser used values may also help end users see important trends. Using the same OS Version field, there are values from iOS 1.0 to iOS 10.2. However, a vast majority of the values may be contained to the past 2 or 3 large releases. Some quick cleanup work could combine the rarely examined iOS 1.0 to 7.0 values to create a single dimension that is much more usable.

There are a few different suggestions in here, but like with any data set some may make more sense than others to try. With Metamarkets, you have a dedicated Data Engineer who can help guide you through these decisions and propose other ideas that are unique to your dataset!

If you’re a current Metamarkets customer and would like to discuss additional tips on maximizing the value of your data, please reach out to your account manager today. If you’re not a current customer and interested in working with Metamarkets – please reach out! 

Eric Albanese is Director of Data Engineering at Metamarkets. He’s responsible for ensuring the stability, scalability and performance of Metamarkets’ big data ecosystem – including data collection, pipelines, ETLs and the data warehouse.