Value per customer – a new way to value cookies

This article was first published in AdExhchanger in 2010.

I think it’s truer today than ever:

Real-time audience buying is currently being heralded as the next paradigm shift in our industry. Empowering agencies to leverage their own data as well as the data of third party vendors in a liquid marketplace should deliver the holy triad for media buyers: right audience, right price, right ad. But there still exists a clear value difference at the moment of delivery – the buyer needs to find/impact/acquire that customer at the most cost effective price possible. While bid prices may in fact drive up prices in some instances, if the buyer has access to massive amounts of scale (aka long tail inventory) and cares only about the user and not the context, the ads will find the users in less expensive inventory. Reach and frequency will be more cost effective to manage and scale against in less expensive inventory, thus short-tail (aka premium) publishers may in fact see only incremental benefit from DSP’s vs. their current ad network daisy chains.

These publishers need to find alternative value models in order to maintain their ability to command premium prices for their audiences. One idea worth considering is Value Per Customer. If audience-level targeting across real-time environments is perfect for the buyer, and impactful, brand-oriented sponsorships are best for the seller, then it sounds like we need a hybrid model to allow for the page itself to be valued and sold. The concept of Value per Customer (or VPC, since we need more abbreviations in this industry) is one that I’ve been ideating on for some time – and one that I believe addresses the fundamental gap between what buyers need and what sellers want.

VPC is a basic concept that operates on the principle that ad serving decisions on the page should be viewed holistically, not just as individual units on a page. And part of that value needs to be the known on-site value of that user. Targeting against a third party data set is of critical importance, undoubtedly. But a users interaction with a site leaves even more valuable breadcrumbs that should be used to target that user. An example data set for an online content provider, in this case let’s say a newspaper, will likely include: average time on site, average pages consumed, bookmarked landing page, links shared, number of comments left, path of visitation throughout the site….and that’s before we get into ads clicked, registration data (which will likely include basic demography), ads viewed per session, etc.

The next step is having the publisher look at the gross margin of the actual ad suppliers. Assuming the typical remnant inventory ad stack inclusive of ad nets, exchanges, and SSP’s, as well third party seller agreements, publishers can use tools like YieldX to assess the ad calls based on what is actually producing the highest yield. While a single supplier may provide higher CPM’s, due to revenue share changes the margin on that ad call may actually be lower than another provider.

Compare the margin of the ads this user sees vs. other users. Is he a high value user or a low value user?

Now as the user starts to peruse the site, he can be served these ads in line with his media consumption. The advertiser can reach the right user at the right time ON SITE, and the seller can start to identify which users should see lower margin/higher volume ads from the daisy chain, which through third party cookie targeting will not allow them to know the true value per user but will allow them to still achieve maximum value per user.

Publishers could then sell against High, Medium, and Low value customer at the page level.

Existing buy:

  • Custom placements for $50K targeting anybody who happens to hit the page
  • Sports section targeting = $15K
  • ROS against user reg data = $10K

Versus Value Per Customer Targeted buy:

  • 1mm impressions at $15 CPM for HVPC = $15K
  • 5mm impressions at $10 CPM for MVPC = $50K
  • 10mm impressions at $1 CPM for LVPC = $10K

Same buy, but targeting the right users at the right price. That will equate to stronger ROAS and less waste. Over time the site can dynamically adjust the content that “low value” user sees to achieve higher VPP, increasing their profit margins and the responsiveness the advertisers want. Plus the user will be presented with more relevant content (and in turn advertising), and spend more time engaged with that property. Advertisers achieve better on-site targeting and can serve the right ad to the right user across a number of different CPM’s at the same time, and sellers are able to both understand and adjust user value over time. This allows advertisers to buy different users at different prices through a truly transparent buy, achieving more appropriate value for both buyer and seller.

It’s a bit of a lengthy process, but in the long run it’s taking the basic precepts of our industry: data, targeting, and value – and delivering them to both the publisher, the advertiser, and MOST importantly, to the end user.