“Data-Driven Thinking” is written by members of the media community and contains fresh ideas on the digital revolution in media.
Today’s column is written by Marc Goldberg, CRO at Method Media Intelligence.
Supply-path optimization (SPO) is yet another vague ad tech initialism, the definition of which changes depending on who is speaking.
Most people think SPO is exclusively a tactic that publishers can use to maximize their revenue, but that obscures the reality, which is that the buy side also needs to optimize for quality sources.
The fact is buyers can – and should – also do SPO. The problem is it’s difficult to understand, and therefore optimize, the supply paths that are coming to them. In order to do that, they need more sophisticated tools and access to data – much more data, and much more specific data.
And then they can start to do some interesting things.
Cutting a path
But there’s a problem: in most cases, buyers don’t actually have the data they need and their ad tech vendors don’t give it to them.
For example, sampled quality data in seven-day intervals aggregated to a percentage without a breakdown of supply pools by their level of quality, is like trying to solve a whale-sized problem, with goldfish-sized memory.
Without specific information, it’s easy to confuse domain optimizations with SPO. You might notice that a domain has a high invalid traffic rate, so you block it. Some other domain has low viewability, so you block it. Low click throughs? Block it.
But this is not really SPO. Don’t forget that the “p” in SPO stands for “path,” and the path is how you get the granularity you need to make meaningful optimizations. To do that, you have to connect information sets from DSPs and ad servers, and this requires having common identifiers.
You need the auction ID (using DSP macros as the key values) and impression ID (using ad server macros), and then you can start to get a fuller picture of your spend.
The true magic happens when you combine this with quality metrics, such as invalid traffic, viewability and other signals. This opens up your ability to optimize your supply paths and answer simple, but vital, questions, like, what is waste and what is working?
Be aware that doing this kind of measurement on a sample basis and using the incomplete reporting can give you misleading results.
Here are three supply-path optimizations that most brands aren’t doing right now but should be.
1. Size matters
There is a lot of useful data available out there – if you know how to connect it.
The events from your ad campaign, for example, can and should be measured. From that, optimize for improvement and keep an open mind. Optimizations exist that you might not have even considered.
Maybe your ad creative is too big, so it’s loading slowly and delaying the rendering. It’s entirely possible that your non-renders are due to a creative size issue. So, look at the data if you have it, and look at creative IDs and performance by creative.
2. Look deeper
Michael Scott from “The Office” put it well: “Wikipedia is the best thing ever. Anyone in the world can write anything they want about any subject, so you know you are getting the best possible information.”
If you are exclusively looking at your DSP log files, you are forgetting that the information is all server-side measurement, meaning that it’s self-declared by sellers. This information needs to be merged – and made mergeable – with the-ad serving records, since those are the truth.
Once they are merged, you can see which sellers are driving most of the spoofing or wasted spend. Remove from those sellers!
3. But don’t throw out the baby with the bathwater
A typical optimization is done by looking at quality metrics by site/domain and then placing low-performing sites onto an exclusion list or removing them from an inclusion list.
Remember, a site might be entering an exchange via multiple sellers. With 100% measurement and supplier data from DSPs, you can try to break out the quality metrics by seller to get a proper granular breakdown for any given site. This allows you to remove a seller rather than just the domain.
For example:
- 3,000,000 impressions from Exchange A and Seller123 (Seller.ID 5409383) – 80% viewable, 2% IVT
- 3,000,000 impressions from Exchange A and Seller456 (Seller.ID 5410388) – 10% viewable, 30% IVT
- 2,000,000 impressions from Exchange B and Seller135 (Seller.ID 4419772) – 70% viewable, 4% IVT
- 2,000,000 impressions from Exchange B and Seller246 (Seller.ID 4419857) – 15% viewable, 40% IVT
Blocking Seller 456 from Exchange A and Seller 246 from Exchange B prevents the low-quality supply path from being purchased while allowing the high-quality supply paths to continue.
Spend your budget, remove what’s not working and reallocate to what is.
It might look complicated on the surface, but it’s actually quite simple, and if you’re curious and motivated enough, you can really help your campaigns.
Follow Marc Goldberg (@MarcJGoldberg) and AdExchanger (@adexchanger) on Twitter.
