Frequency Capping Won’t Get Solved Anytime Soon

State Farm is wasting money because of frequency cap violations… and that’s okay

The ability to implement a universal frequency cap is one of the major selling points of consolidating programmatic buying with a single DSP. And in practice, DSPs are pretty good at frequency cap enforcement. But they are not perfect, and solving those last few fringe cases that cause frequency cap violations just isn’t a priority.

Here’s what’s happening in the screenshot to the left. When I visit weather.com, Weather’s SSP conducts multiple concurrent auctions — one for each ad slot on the page. State Farm’s bidder evaluates each bid request independently without considering the possibility that it might win all four auctions. Because I am in State Farm’s retargeting pool, State Farm bids aggressively and wins all four ad slots. It is entirely possible State Farm implemented a 3-per day frequency cap, and in a single page view, this frequency cap is violated.

Could DSPs solve this problem? Sure. There are lots of potential solutions — sequencing bid responses, checking for concurrent bid requests, or potentially utilizing optional bid parameters to link multiple auctions. But in the context of broader ad tech development opportunities, this fringe case just isn’t a priority. Both buy side and sell side technology developers simply have more interesting problems to solve.

State Farm could probably save some money if its bidder had smarter logic to prevent roadblocks. But it can also unlock lots of new use cases by allowing its technology partners to continue pursuing needle-moving development. For now, let’s focus on real innovation. Tweaking can come later.

16,000 Wasted Pixels

While watching an interview with “Selma” director Ava DuVernay, I opened my phone and visited imdb.com to learn more about her career. At the bottom of the page, I saw an ad promoting Amazon’s new streaming video service, Amazon Prime Instant Video. The ad had a clear call to action and was positioned next to highly relevant content. There’s nothing obviously wrong with the ad unless you know that I’m already an Amazon Prime member and already have the Prime Instant Video app installed on my phone. Why would Amazon waste their money promoting a product that I’ve already purchased?

It turns out that iOS makes it very difficult for advertisers like Amazon to make smart targeting decisions when delivering ads in a mobile web browser. The issue stems from two different tracking mechanisms that coexist on iOS devices.

1. Native iOS apps have access to a mobile device ID (called IDFA, or “ID for Advertising”) that is consistent across all installed apps.
2. Websites that are loaded through a browser cannot access the device’s IDFA, but instead access cookies that operate very similarly to desktop browser cookies.

When I installed Amazon’s Prime Instant Video app, Amazon’s marketing database recorded my iPhone’s IDFA (let’s call this 1234). But when IMDB sold a banner ad within my browser, Amazon’s bidder was passed a cookie ID (let’s call this 6789). To Amazon, cookie ID 6789 is a brand new prospect, not an existing customer. Amazon purchased the 320x50 ad slot and wasted an otherwise productive 16,000 pixels.

The cookie vs. device ID issue is problematic for both for targeting and attribution. Ads like Amazon’s are poorly targeted, resulted in wasted ad spend. In other cases, advertisers miss opportunities to engage warm leads, resulting in poor publisher monetization. And when an impression served in an app drives a conversion on the web, its impact is often not measurable, reducing the perceived effectiveness of mobile advertising.

Start ups like Tapad and Drawbridge are leveraging probabilistic modeling to link multiple IDs, and scaled platforms like Google and Facebook are well positioned to use login data to provide more robust ID matching products. But we are still in the early innings of people-based marketing, and until the industry embraces a consistent ID linking solution, the mobile ad economy will suffer.

It’s Going To Be A Powder Day! A Banner Ad Told Me So

View fullsize

I’m heading to Vail on Wednesday, and according to this banner ad, it’s looking like I timed my trip well. Vail is getting socked with snow, and I’m getting real time updates as I browse the web. I took the screenshot below at 6pm on February 28th. The ads I’m seeing now tell me that the snowfall total is up to 24 inches, and the tally keeps rising.

So how is Vail accomplishing this little bit of marketing magic? Is some poor ad ops soul swapping banner ads every time another inch of snow falls? It turns out Vail has a clever solution for automating these ads. By working with a dynamic creative vendor called Spongecell, Vail is injecting real-time snow report data into their ads. A few components need to come together for this to work:

• Vail’s website has a page that includes the latest snow report. You can find a consumer-facing version of that page here and an XML version of the raw data here.
• Spongecell has a simple tool that retrieves raw data from a third party domain. Combining the Spongecell tool with Vail’s XML document yields a URL that can send live snowfall data to any object hosted on Spongecell’s domain.
• Spongecell can then build a Flash file that retrieves data from Vail’s XML document and uses this data to populate the ad’s message.
• Because the overall process relies on Vail to maintain a live data feed, there is a risk the dynamic snowfall data fails to populate. As a backup, Spongecell also loads a static ad in the background. That static ad can be served if the connection to Vail’s XML document fails.

The final experience is a dynamic ad that is populated on the fly with Vail’s latest snow report. Fingers crossed for another few inches!

Why Site Should Definitely Not Be A Mandatory Object In A Bid Request

@simonjharris recently argued that the IAB should require publishers to disclose their site URL or app name in RTB bid requests. This would be a bad idea for two reasons:

1. Transparency requirements would push premium publishers out of the market
   The rapid influx of premium publishers to the RTB market is largely driven by the flexibility to provide limited transparency into inventory. Premium publishers with direct sales forces fear that RTB channels will cannibalize revenue from reserved sales. The option to mask inventory details on open exchanges greatly reduces the potential for channel conflict, making RTB testing more palatable for premium publishers. The RTB ecosystem has benefited greatly from the growth of premium inventory, and this growth would have been stifled if publishers were required to provide full URL transparency.
2. Economic incentives are already in place
   Regulating URL disclosure is also simply not necessary because there are already strong economic incentives for publishers to sell inventory transparently, a point Simon acknowledges. Due to brand safety concerns, many advertisers will not bid for blind bid requests (those that don’t disclose URL). With lower bid density comes greater price reduction and therefore lower closing prices. By providing URL transparency in bid requests, publishers benefit from better RTB yield, and as premium publishers become more familiar with the RTB landscape, many are choosing to provide greater transparency in bid requests.

The IAB’s goal should be to enforce the minimum possible level of standards that enable a functioning advertising marketplace. Free market economics should handle the rest. The decision to make site an optional bid request parameter maximizes marketplace freedom, while incentivizing behavior that benefits both buyers and sellers.

Find the complete OpenRTB 2.3 specs here:https://github.com/openrtb/OpenRTB/blob/master/OpenRTB-API-Specification-Version-2-3-FINAL.pdf

Audience Syndication: A Peek Behind The DMP Curtain

In a recent post, I highlighted some strange behavior of retargeting campaigns. One of the more surprising experiences was the way Jeep delivered ads to me following a visit to the jeep.com website. As expected, shortly after my visit to jeep.com, I began seeing ads promoting Jeep’s SUVs. But two things were unusual about these ads. First, the ads were sponsored by a local Jeep dealership (Cherry Hill Jeep), not the national Jeep brand. Second, the ads were served by a company called AdGear, which does not fire retargeting beacons on the jeep.com site.

So how did Cherry Hill Jeep know to target me, and how did AdGear execute the ad buy? While hard to say with certainty, I suspect Jeep is using a data management platform to syndicate audiences for local dealership campaigns. Here’s how audience syndication works:

1. Jeep partners with a data management platform to track visitors to jeep.com. Based on tags that load on Jeep’s website, it appears Adobe is integrated as a DMP. When I visited jeep.com, an Adobe beacon loaded on the site and recorded my visit in an Adobe database.
2. At any time (before or after my visit to jeep.com), Adobe can perform an ID sync with AdGear. The details behind ID syncing are complex, but the outcome is that Adobe and AdGear share with each other the anonymous ID that each company uses to track my online behavior. (“Hey AdGear, it’s Adobe. The guy you call user ID 1234 is our user ID 6789.”) Once an ID sync is in place, Adobe can send information about my jeep.com visit to AdGear, and AdGear can place me in Jeep’s retargeting pool.
3. AdGear can then activate a campaign that targets consumers who (a) recently visited jeep.com and (b) are within driving distance of a particular Jeep dealership. Without any technical integration with the jeep.com website, Cherry Hill Jeep is able to run a retargeting campaign.

Retargeting has been around for a while, and this seems like a complex way to accomplish a relatively simple advertising task. Why not just load the AdGear beacon on jeep.com? Assuming Jeep wants to allow all of its local dealerships to run retargeting campaigns, a traditional approach would quickly overwhelm jeep.com with retargeting beacons. Jeep has 40 dealerships in New Jersey alone. Add another 112 dealerships in New York and 27 in Connecticut, and we’re looking at almost 200 beacons just to cover the tri-state area’s dealerships. Scaling nationally, we quickly hit a point at which jeep.com becomes unusably slow due to the presence of hundreds of retargeting beacons. By adopting a single DMP and then syndicating audiences to each dealership’s preferred media buying partner, Jeep is able to execute local retargeting campaigns while preserving a seamless consumer experience on its national website.