Commentary
Automating TV Ad Activation Is Hard: Adobe Exits The Business
- by Dave Morgan , Featured Contributor, June 18, 2020
In its earnings call Friday, Adobe President and CEO Shantanu Narayen announced the company was shuttering Adobe Advertising Cloud transaction-driven offerings, the TubeMogul TV and digital video ad activation business it acquired in 2016. Narayen remarked that its Advertising Cloud transaction-driven offerings were “low margin,” not core to Adobe’s business, and “in fact are extremely resource-intensive.”
With the announcement, Adobe joined a long and venerable list of companies that have met their match trying to automate the activation of TV advertising. Remember eBay TV-Ad Auction? Spot Runner? Google TV Ads? Microsoft Admira? Aol/Adap.tv? Videology?
Why have so many companies tried and failed to build profitable, scalable businesses automating TV advertising with digital-born techniques?
The reason for trying is obvious: TV advertising is where the money is, generating more than $70 billion in U.S. spend last year. The repeated failure part is a bit more complicated, but having had a front-row seat throughout, I have my opinions.
Here are the main reasons why so many have failed to automate the activation of TV advertising:
It takes more than pretty dashboards. Too many of the TV buying services were mechanical Turks: pretty dashboards that just handed off insertion orders to conventional buyers and traffickers to buy the old-fashioned way.
Trafficking TV spots conventionally is very time-consuming, and the faux precision of tight targeting and fast turnarounds as suggested by digital dashboards created expectations that labor-intensive “managed services” couldn’t deliver on, certainly not without being “extremely resource-intensive” and quite unprofitable.
Retrofitting digital buying systems to TV doesn’t work. The workflow of TV and digital are about as opposite as you could imagine. In TV, demand for inventory exceeds supply. In TV, planning and buying have little to do with each other, with planning now more of a Kabuki dance that occurs the better part of a year before many spots actually air.
Rather than retrofitting, automating TV ad activation means integrating into the TV ad workflow, particularly the internal trafficking, many times using software to connect into TV media owners’ human management systems.
That way of operating is foreign to how most digital ad systems were created, intended more for real-time bidding, largely unchecked access to inventory and rates, and advertiser-managed creative delivery.
Without the capacity to have true automation from planning to buying, ideally integrated and automated to the network spot level, buyers/advertisers are left with a better plan, but no better chance to execute it; basically, too many cards short of a full deck to be useful.
Commingling economics with agency buyers doesn’t end well. When buyers are incented through shared fees (frequently undisclosed) to use an ad activation platform, the platform tends to underinvest in technology and overinvest in client management. As volume grows, advertisers want fees to go down, agencies want bigger cuts, and the number of account managers (costs) keep growing.
It’s a vicious circle, and ultimately death spiral. All you need to do is remember Videology’s bankruptcy. WPP’s Group M was both one of its biggest debtors and creditors. These days, transparency rules.
Access to broad pools of inventory is critical. A digital ad planner for TV that doesn’t connect into a broad and diverse pool of inventory is at a big disadvantage. Google TV Ads was way ahead of its time in many ways, but its reliance on inventory primarily from Dish, with little bits from a couple of the NBCU networks, meant it couldn’t do optimized audience-based campaigns in any significant way. It just ended up being an easy way to buy pre-bought inventory on Dish.
The same thing has happened to those “platforms” that were just dashboards on top of limited pools of pre-bought, bartered or daisy-chained inventory from other intermediaries (frequently non-disclosed to hide provenance and mark-ups).
Truly automating the world of TV ad activation is hard. It has defeated many who have tried. What do you think?
Very good points, Dave. Also, the buying of TV time, especially on a national leve,l is not merely a matter of garnering the most targeted eyeballs per dollar, no matter what the content is or the daypart or the network type. Many other intangibles are involved that automated systems can't handle, therefore they are relegated to rather costly data and price chomping, leaving the actual buys up to people who, often, must disregard what the computers are saying. As you said, it's almost the exact opposite of programmatic buying for digital media.
The main reason that attemtps to automate national linear TV have not worked is that most, if not all attempts have focussed on the buy side with no real benefit for TV sellers. What you need is automation that allows the same amount of control over pricing and inventory management that linear TV sales departments have always had.
Google TVads, TubeMogul, Videology and many many more have all tried to make TV buying just like digital, but have negected to focus on the very people who own and control the inventory: The TV networks! Clypd has been the only company that has tried to solve the sell side challenge. WideOrbit? All local TV, which is a completley different animal.
For more than 20 years I have been looking for an approach that will work and I really think I have found it. Offering crosss CTV and linear TV campaigns, while optimizating on both sides will be the key. More to come.....
Thanks Ed. I totally agree that many of the digital attempts have failed because they didn't understand how TV works, but I don't agree that the only answer is buying conventionally. At my own company, we are activating/buying between 5000-15000 national spots a day with only 1 buyer and 4.5 traffickers. it can work because our systems predict not just what spots we need, but which are likely to clear at the right prices and times. The instructions are now being dleivered in portals that the sell side logs into, and they can accept/reject in the application. The trafficing then happens automatially, with the capacity for them to have it delivered direclty into the API of their stewardship and reconciliatoin systems. Software can now automate 98% of the TV activation transaction, but it has to be purpose built for the TV ad workflow, not repurposed from digital ad programmatic systems.
Long, very good points. I agree that it requires significant sell-side development to make it work (see my comment above to Ed). I also agree that the audience data and plan optimization needs to incorporate CTV/OTT ads as well.
Dave - What is simulmedia doing to help TV sellers optimize their inventory yield?
Long, currently, we are helping networks automatically accept, manage and yield optimize orders that we generate for them. Once that is established, we will likely open the platform up for them to use with other demand generators.
Dave, I didn't mean to suggest that buying conventionally, mainly via meaningless "demos" such as adults aged 18-49, was better. I was referring to the many intangibles such as program content that's related to the image the advertiser wishes to convey, or content that generates higher degrees of commercial attentiveness, or content which has less commercial clutter, or highly merchandisable content---to the "trade", for example--or buying into content, such as sports, because of promotional tie-ins with the leagues or the stars, etc. As you know, I advocate a two-upfront system whose lynchpin calls for a special upfront for individual brands only. By definition, this would use all sorts of metrics and involve many considerations that are ignored in "conventional" eyeball counting TV time buys.
Ed, yes. Totally understand. I agree that a two-step upfornt is the way to go, and hope that the current disruptions give folks a chance to consider it. My point was that having automated systems making both spot-level selection but spot level activation and trafficking is where the TV ad activation platforms need to go. Adobe and others failure to do the second part doomed them to low margin, high resource, high cost managed services that were inevitably going to fail.
Thanks Dave.
IMHO, I do not think that having the buy side pick and choose the inventory and then simply helping the sellers "fit that inventory in" for the best yield optimization will be good enough for the TV networks, unless I am not understanding your approach.
Long, I'm not suggesting that the best yield optimization is for the buy side to cherry pick what they want. Quite the opposite, you have to start with a strong, actual and predictive point of view on the likely audience volume and composition and price and availability for every single spot on every single network at least three months forward. Then, if your platform has demand from doens of advertisers, you can interact with each and every network to both fullfill each advertisers demand on an audience/performance/price basis as efficiently as possible and significantly increase the yield of both their scatter, rotators and DR availabilities. The networks do not have visilbilty into the fact that one adverter (console video game) might pay 4X for the same spot as the incumbent automotive that they typically rotate into the spot. Their inventory management techniques are way too dated and they don't have strong views into the diversity of the market demand. Plus, their audience prediction models are terrible; almost always built on year over year guesses.
Ouch
Sorry to push the point so hard Long, but you know better than anyone, probably, that TV media owners have been slow and resistant to proactively embracing change and much-needed yield optimization. I don't know of any way other than tying it into revenue delivery at improved yields - didn't mean for my passion on this point to come off so sharply. But, as you know, i tend to be pretty open and direct on issues :-)
Dave, I have to agree with Long---assuming that I understand what both of you are saying. Basically, the sellers must find a way to unload content that is not so desirable on advertisers in package deals, and share out the goodies to hook as many buyers as possible. The problem is that only a small percentage of TV shows really attract the most desirable audiences---younger to middle aged with above average incomes---as their primary viewers. Instead, most TV shows are, at best, "flat" in their demos or, worse, heavily loaded with older and low income vewers. These GRPs must be moved, somehow. Sure, it's possible that one might use sophisticated targeting methods to single out a few shows from the pack that give your package a slight positive demo lift, but once the networks see that this is happening, by virtue of the buyer's requests for inclusion in their packages, they are certain to charge more for the desirable spots so where does that leave you? As Long has said a number of times, and I have also pointed out, in theory, its desirable that both sides---sellers and buyers---"win" in these deals, but unless the sellers have a lot to say in creating their packages, I don't see how they can come out as winners.
Ed, very good pionts. However, it isn't just about finding those few shows; that is relatively easy. It is about finding small pockets of those audience in lots and lots of shows in places and volume that folks don't normally look to. This is tough for media owners because so much of their information comes from talking to endemic advertisers for their content and trheir high indexing audiences.
Also, I don't advocate media owners selling it to buyers in a bundle, In fact, I believe that selling wholesale has dramatcially hurt TV networks and is not good for them, which is why I am a big advoate of your idea for a two-stage upfront with selling audience-based programs handled specially.
Dave no worries. I'm not offended in the least. It is the TV Networks I would be sensitive to.
It's nothing that I don't say to all of them regularly, both privately and publicly ;-)