Commentary
Do You Know How The 'Black Boxes' In Our Industry Work?
- by Dave Morgan , Featured Contributor, February 21, 2019
The worlds of media, marketing and advertising contain a lot of “black boxes" -- systems and technologies that perform functions without a lot of obvious visibility into how they do what they do. The list could include how ad servers make decisions, how location-based tracking projects where someone is (or was), or how a research panel balances its output.
I’m naturally curious. My father was a prosecutor, so being inquisitive (or interrogating) is how I was raised. And, I’ve always believed that those who know how things work have a big advantage over those who don’t.
To be clear, the lack of self-evident transparency into operations is not the result of a strategy on the part of the system creator or operator to deceive or obfuscate. It’s just that most of them involve very complex systems, disparate inputs and complicated algorithms — any or all of which may be modified at any time.
Thus, it’s just not simple or practical to make it easy for all to understand, particularly when the purpose of these systems is to drive a particular market output or result.
advertisement
advertisement
Net, net. It’s probably true that most of the folks working in our industry are relying every day on the outputs of less-than-transparent black boxes. Here are some examples. Test yourself to see how much you know about them:
Cable zone ad insertion. All those who watch television have grown up watching a smattering of localized ads during national network programming — the local car dealer ad during the Super Bowl, for example. Do you know how those ads were delivered? Do you know how the Q Tone works? Do you know why the ads are sometimes cut off?
TV upfront contracts. A massive amount of the entire national ad market is negotiated and committed over a two- or three-month time period. Do you know how those deals are structured? Do you know what options either buyers or sellers have? Do you know how those deals are guaranteed?
DSP auction bidding/header bidding. So you want to buy some digital ads on an exchange, maybe on Google or MediaMath or Xandr’s AppNexus. Do you truly (be honest) know how the auctions operate? Do you know how bids are accepted? Do you know how second bids or Dutch auction bids are calculated?
Header bidding has been an enormous issue over the past two year for digital publishers, demand-side platforms (DSPs) and supply-side platforms. Do you really know what header bidding is, how it works and why it’s so important?
ACR TV ad spot detection. Many of you have seen the results of automated-content recognition of TV content and ad spots: an output of properly equipped “smart TVs.” Do you know how the TV knows (ostensibly) what ads you watch? Do you know the difference between “fingerprinting” and “watermarking”?
Commissions, discounts and rebates. Yes, this became a sore spot in our industry over the past years, and may still be for some time to come. However, if over the previous decade or two everyone in the U.S. ad market had worked to learn about what these were and how they operated by geography, vendor and partner, the transparency work of the ANA might never have been needed.
Addressable TV ad delivery. You hear and read all the time about how big and important addressable ad delivery on TV is becoming. Do you know how different TV ads are actually delivered to different households? Do you know how systems across AT&T, Dish, Verizon and Altice do it? Do you know which have to store ads on personal DVRs in the homes, and which can deliver the ads dynamically? Do you know how they differ from the system employed in Europe by Sky and its AdSmart unit?
CTV ad delivery. Connected TV ( CTV) is pretty buzzy these days. Do you know what CTV actually is? Do you know the formats of ads that can be delivered? Do you know the difference between CTV and addressable TV ads and AVOD (ad-supported video-on-demand) ads?
Why does it matter if you know how these black boxes work? Isn’t one of the perks of management that you don’t have to know how the sausage is made? Isn’t that why you hire technologists and experts?
If this is the way you see the world, I suggest you're taking enormous risks both with your career, company resources and decisions you’re making for your employer.
Knowing how black boxes work means tens of billions of dollars a year in margin for companies and people in our industry who get them right.
What about you? Do you know how the black boxes in your business work?
All criminal activities.
The problem, Dave, is that most higher-ups, especially on the advertiser side don't see a need to immerse themselves in what they regard as tech or research trivia. They like to believe that the agencies are on top of that---which often, these days, isn't the case. This is not because the agencies are derelict in their duty. Rather, it's because they are squeezed so tightly on their fees and person-power assigned to each account that their expert, generalists, such as the media research folks, are expected to do this kind of thing without charging the client for their time. And, to be fair, unlike TV's early days, when advertisers developed and sponsored their own shows and subscribed to Nielsen, the agencies know that their clients, despite what they say in public, don't really care---so long as the agency isn't totally screwing up the media buys.
I should note that my comments refer primarily to traditional media, not digital, where there is much greater pressure by clients and lots of them are going in-house to save money ( questionable ) and/or as a result of agencies who specialize in digital, not being "transparent" as has been revealed so many times recently. Here, there is a need to know but, though not to learn all of the intricate details but, mainly, the concepts and what benefits they mey bestow on the advertisers' sales promotion activities which are the main thrust of the in-house movement.
I think that trying to make senior advertiser and agency management types well versed in the subjects you raised as well as others, is a lost cause. However, making them understand that they need to fund and support those in their organizations who not only should know this stuff but would find much of the learning process very interesting might gain support---if you can find a way to connect with the big shots and convince them to pay real attention to the media function.
Dave, many kudos on addressing the pervasive black box issue. It does point to a critical transparency issue in our business. It also highlights, per Ed's comment about infusing C-suite/agency leadership with a sense of a how tech works as a lost cause, the tremendous disconnect or even apathy at higher levels. No, senior execs don't need to bathe in algorithm granularity, but they are responsible for knowing what they are paying for. And sometimes that bill is pretty high.
To justify lack of knowledge with an "at least X performed better than Y" mentality shows lack of fiduciary responsibility to brands and clients. At some shops the adtech/martech budgets are higher than the base IT budgets. How is all that money being held accountable? When you start digging into how many intermediaries are taking % cuts along the path, it prompts brands to bring processes in-house. Can you blame them? When it's in-house, at least it's their own black box and they know what's inside and how it works.
Great points and perspective Ed. I totally agree that there are a bunch of issues, both structural and cultural that prevent the leaders in our industry from learning how the blocking and tackling is conducted. I believe that it is critical that we find ways to reverse this, or we will continue to see too much energy, time and resources in our industry chasing wrong-headed strategies.
Great post Dave.
Dave,
Great points. And while we wish that executives in the ad business fully understood just how things work, I fear that they simply never will.
Even the basics of viewership measurement. Ask a president of a programming service how Nielsen’s sample is drawn, or how potential panelists are recruited, installed and then incentivized. What in-tab rates are for various demos? How the people meter actually works or how do the editing rules work to produce viewership?
In my job, over the past many years, I have been asked to explain these types of things to senior executives hundreds of times. But very few actually took much interest in this information. My honest assessment (and I’m a consultant, so I am always honest) is that fewer and fewer people care about this stuff.
Dave I'd like to add another one.
Cross-media measurement. In a word - impossible. Cross-media modelling metrics - possible.
We have to realise - and accept - that if you COULD accurately measure all media exposure, that (i) the metrics (especially time) need to be standardised along with geographic and demographic definitions, (ii) the measurement methods need to be standardised and calibrated equally (well, fairly rather than equally), (iii) in order to get longitudinal measurement you would need a panel, (iv) given advertisers stated needs the panel would need to be comprehensive, massive and representative, (v) it would need to interface with various other thrid-party data a.k.a 'big data' (which we're already doing) (vi) no-one wants to fund it and no-one can afford it.
However modelling 'currency' measurement data is possible.
Oh ... and one more if I may. How can someone get 2,864,902 fewer votes and still win the Presidency?
I'm with you Marshall. The issue is not just that most senior execs don't know how some of the kys metrics in our industry work, it's that fewer and fewer today care or want to know.
Electorial college and gerrymandering and whole other explanations are how the popular vote lost. Deeper dives into the wrong questions asked from the get go way before the election probably have something to do with it. However, what Dave is discussing and the election system disection are in a rather different end path.