It has been a while since I’ve said that here, in fact it has been 8 weeks since I took (unexpectedly) early paternity leave. Lots has happened since then & it’s good to be back. As always this weeks newsletter covers off the key stories shaping the industry we work in.
We start with Amazon who recently announced they’re acquiring Sizmek’s AdServer, we look at what their motivations might be. From a business that loves AdTech we turn to one that doesn’t, Apple. This week they announced a number of privacy features focussed on apps at their worldwide developers conference that will likely challenge the market we work in, we look at the detail.
Next up we turn to an academic study that asks the question whether behavioral targeting is worth it for publishers? The study suggests the answer is not really, but there are a number of questions over the methodology used to get to this answer, we’ll take a look at these. Finally we turn our attention to Bayer who have saved $10m by taking their programmatic buying in house with the help of MightyHive.
Amazon Acquires Sizmek’s AdServer
We start where the last newsletter left off, Sizmek. You’ll recall that the troubled AdTech business had filed a Chapter 11 bankruptcy in New York & its future was very unclear. At the point of filling it had 1,000 - 5,000 creditors & owed over $39 million to the key online advertising exchanges.
Many forecasted a fire sale & I said at the time if that were to happen the ad server would be considered the crown jewels because a) it remains popular & b) is one of the few credible alternatives to Google’s Campaign Manager. Last Friday Amazon announced it was buying Sizmek’s AdServer & Dynamic Creative products:
So how much did they Amazon pay? In their filing Sizmek estimated their assets were worth between $100 & $500 million, quite a spread. According to WSJ Bankruptcy Amazon is only paying $30 million for the products it’s acquiring, suggesting the market values Sizmek’s assets towards the low end of the above estimate.
Why the acquisition? I suspect long term Amazon will look to create a link between the AdServer they’ve acquired & their DSP. This will allow audiences & creative to sync from one to the other. This is how market leader Google’s adserving & buying suite works currently & it affords advertisers a number of efficiency & cost benefits so this type of integration would also be a very Amazon thing to do.
However in their public announcement Amazon said the business would run as separate entities for now, so whilst this seems to be off the table in the short term, it is one to watch for in the long term potentially more plausible than some of the rationales that have been proposed:
1) Some postulated that the DCO product was the driver as it gives Amazon an improved ability to recommend products to people offsite based on their onsite behavior. This likely true but an acquisition sure seems an expensive way to achieve this objective.
2) Others hypothesized that Amazon bought Sizmek for market intelligence, to see “every detail of the buy including rates” this is a touch far fetched given a) very little cost data lives in an AdServer b) Amazon’s marketing spend rivals that of P&G & Unilever so they already have a very good view of what things cost online, offline & in social c) They own enough AdTech to know what an given user ID is worth. Log level data for conversion events could be interesting though.
As I said I suspect this is a long term play with the Seattle based giant looking to replicate Google’s playbook, Sizmek is their DoubleClick & Ad funded fire TV their YouTube. This will all make for an interesting 2020, especially as its rival Google looks like it will increasingly come under anti-trust scrutiny.
Apple’s New Privacy Features
Apple has battled & defeated AdTech on the mobile web over the past 18 months & this week at their Worldwide Developers Convention in San Jose they announced yet more privacy based features, this time focussed at Apps. This looks like it will make life here for many AdTech companies a little bit more difficult:
The first area they beefed up in App privacy features was location, Apple already offers the option for location to only be tracked whilst the App is in use but iOS 13 released in September will bring with it an “Only Once” option that’ll share a users location when they first open the App but then not again. This is going to really limit many of the location data companies ability to build out their data sets for users who select this option & one can see it impacting device graph business as well.
For those who allow tracking whilst an app is in use, iOS 13 will sporadically check if if the app still has the user’s consent. This consent renewal dialog will include a small report the with the locations your app has been tracking/sharing so if a user deems this excessive then they will have the option to revoke access, which again could hurt AdTech companies.
But the big feature was “Sign in with Apple” which they suggested was an alternate to being tracked by business that profit from advertising (Google & Facebook). Amongst other things this feature will randomly generate a different email address for every app or service that person logs in to as a way to limit tracking.
Whilst this sounds like Armageddon for AdTech it is worth noting that Apple is not getting rid of IDFA’s yet (the identifier which advertisers use instead of cookies in iOS) which would have caused massive problems. The key word here though is “yet” & I’d imagine there are a number of businesses that will be exceedingly worried that in 2019 & 2020 Apple will become as hostile towards advertising tracking in App as they were with mobile web advertising in 2017 & 2018.
Behavioral Targeting: Is It Worth It?
This week a study by academics from Carnegie Mellon, California & Minnesota had some of the AdTech world up in arms. The study titled “Online Tracking & Publishers’ Revenues” found that an impression with a Cookie had only 4% more value than one without. The WSJ had the scoop that kicked it all off:
Many pointed out there were a number of flaws with the research. Before we dive into that I’d recommend reading the draft of the study here. There’s also a great tweet storm on the potential flaws of the study by @pbannist which you should check out here
The veracity of the findings are important given there is an ongoing debate around consumer privacy & if it’s conclusively proven that AdTech doesn’t help publishers better monetize their content, the argument that the juice isn’t worth the squeeze is much easier to make.
So what look to be the issues with this study? Firstly the data set is old, three years to be precise. Lots has happened since then Safari’s ITP has made it tough to target & track users impacting pricing around trackable users. The study found that 73% of Safari browsers had no cookie ID associated but given that almost all Safari now has ITP I’d suggest this will be now much higher. This matters, as do the consent frameworks GDPR has bought with it which again were implemented after the study was done.
Not only is the data relatively old it is also a relatively small sample. The study says it used a sample millions of impressions taken over a week, for context most DSPs handle more bid requests than this each minute. Therefore the potential for sample bias is there, despite the best efforts of the authors, as highlighted in this excellent AdExchanger article.
Despite its flaws the study is super interesting & worth a look. I’ve long felt it’s odd that a display market worth circa $60bn in the US this year alone has very little independent information on it & so my hope is we’ll see more similar studies & as this field of research grows, the kinks that have rightly been highlighted are ironed out.
Bayer Goes In House & Saves $10 Million
As you all know In-housing programmatic has been a hot topic for some time now & over the past 12 months we’ve started to see some real success stories come to light. Bayer’s head of programmatic recently shared the benefits they’ve realized by taking things in house at a DigiDay Programmatic summit in Austin. You can read the write up here:
As with many other in housers Bayer said they still used an agency for offline buying but that it wasn’t working with anyone in digital, save for MightyHive which it views as its transition agency.
Critically Bayer said they had successfully worked through the challenges often associated with in housing 1) finding the right people 2) escalating costs of platforms & has said the process has helped them realize massive cost savings of ~ $10m & no that’s not a typo (!).
Industry leaders at the same event commented that their clients who had taken programmatic buying in-house still kept their “best programmatic people on retainer” which being a viable business model in the short term, but some online pundits are questioning if this is a sustainable long term strategy. Why is this?
Well with programmatic buying so much value is driven from insights learned from the execution layer (trading). Some are now arguing that the danger for agencies is that if more & more big clients take control of this they might lose their edge here, which in turn could weaken any consultancy offering they build out.
How this plays out remains to be seen, but many are now saying the long term/bigger challenge for agencies is what happens when programmatic platforms become infrastructure for delivering audio & TV ads? “Offline” is one area that currently almost all advertisers (even those who have gone in house) retain an agencies services for, but will they still do this when they can buy TV ads as easily as online display? Again what happens here is still to be decided, but how agencies respond to this challenge will likely determine their future success.
Each week this newsletter gets forwarded to over two thousand people, if you’re one of them & want to join the three thousand plus marketing professionals who get the newsletter direct, you can sign up easily here, by hitting this button: