Programmatic Inventory Isn’t Black or White

During the last few months the industry has experienced yet another tidal shift in how buyers and sellers think about transacting on digital media. The rise of the private marketplace (PMP) and the fall of the open exchange has been caused by fear and chaos created in part by the media, start-ups looking to exit, and of course sellers looking to create advantage for themselves with clients/agencies. However, all of this fear is not unwarranted.

All inventory is not created equal, nor are exchanges, and worse, nor are direct PMPs leveraging a standard DealID. How is this possible? Buyers on behalf of clients spend painstaking hours creating well-tailored whitelists to ensure delivery on only those domains and brands that their brand is comfortable with being associated. Conventional wisdom is that a whitelist is the ultimate protection for a brand for both reducing fraud and bots and for brand safety. But turns out, that couldn’t be farther from the truth.

There are more than 200 exchanges representing more than 300,000 domains, and there is massive duplication of these domains across all the exchanges. Most people think that when you apply a whitelist in the open exchange you are buying directly from the domain you have on your whitelist. This may happen, but the reality is that given the number of times a single impression can be bought and resold across these exchanges, you are likely not getting the quality impression or an impression at all on that domain. This creates immense complexity, price instability, opportunity for fraud, and risk for the brand. Therefore, while you may be getting access to the domain you’ve selected, it is far more likely that you are not due to domain obfuscation, redirects, adware, overlays, and nasty iframes all designed to make the DSP whitelist believe they are paying for a quality impression.

In addition to a whitelist, buyers are now working with a smaller number of exchanges and building PMPs (using DealID functionality) as a way to more directly access a single publisher’s inventory. Enter: non-human traffic and fraudulent activity. Unfortunately, similar to the open exchanges, there are similar problems with DealIDs. There are massive swings in the amount of non-human or fraud associated with a single domain on DealIDs from one exchange to another. This means that accessing the same domain using a DealID on one exchange vs. another will expose the buyer to varying degrees of non-human traffic. DealIDs are very good for a single brand that wants to reduce price competition within a specific publishers portfolio, but it will not eliminate fraud.

The last problem that buyers try to solve for with whitelists and DealIDs is viewability. When selecting inventory at the domain level and allowing inventory to be bought run of placements within a single publisher, there are no guarantees that the buyer will get highly viewable inventory. Buyers need to be very prescriptive and work with the publisher to define the viewability metric, the proper placements, and what will qualify for a payable impression. The challenge here is that if transacting on a DealID and the impression is taken by the DSP to satisfy the buyers campaign, it is considered a “payable” event by the DSP and SSP or Exchange that provided the impression. Viewability is not considered today, and neither the DSP nor the SSP have any incentive to apply viewability as metric for payment because that could reduce their vendor fees by as much as 60 percent and on some placements as much as 95 percent. Buyers that want to transact on viewability need to calculate vCPMs and explain in detail to their clients how there are impressions that are bought out of view. To calculate the true value of the viewable media, you need to subtract the out-of-view impressions and then determine real cost of the viewed media (eg. 1 million impressions purchased at $1,000, 60 percent in view, converts to a $1.67 vCPM instead of the original expected $1 CPM).

Wondering what to do in this complex landscape of deception and chaos? It isn’t all that bad, but it does take focus, dedication, and more analysis than you may have done before. Here are a few steps to get started:

  1. Select a vendor that provides fraud, viewability, and brand safety reporting on all inventory purchased across sellers
  2. Evaluate each of these metrics by seller/exchange to determine which are the highest offenders
  3. Review duplication of domains by seller/exchange and compare pricing alongside the safety metrics and use this data to limit buying of domains to a single, quality seller
  4. Remove any sellers that don’t meet a minimum standard for the metrics to limit exposure and risk for brands
  5. Continually evaluate these metrics by seller and placement, sharing information with the sellers to continue to optimize and improve the quality of the inventory

With these steps, buyers will go a long way to improving the quality and consistency of the inventory being acquired for clients, and they will make huge strides in the war on bots and fraud.

Related reading