• by Joe Mandese  @mp_joemandese, August 28, 2023

As the ad industry stands on the precipice of what many believe will be its most disruptive transformation ever -- an AI-powered one -- it's worth stepping back in time to recall how it was changed by its first big tech shift -- computerization -- and why it hasn't evolved much since.

Perhaps no piece of pop culture captures that better than Episode 4 in the seventh season of "Mad Men."

The episode, entitled "The Monolith," has multiple meanings -- a nod to the human epiphany-inspiring monolith depicted in the opening of Stanley Kubrick's film from that period: "2001: A Space Odyssey," as well as the beginning of the transition from the "Mad Men" era of the 1960s to a "Math Men" one.

The episode opens with Don Draper arriving late at the office amid more than the normal agency hustle and bustle and getting flagged by Roger Sterling.

Draper: "I didn't get a memo. I don't even know what's going on."

Sterling: "Well, we're getting a computer."

The computer, which is installed in the agency's metaphorical watering hole -- its lunchroom -- displaced a space where the staff would gather to chitter-chatter and spark big ideas. It's the background presence throughout the episode and is distinctly branded IBM, which was the mainframe state-of-the-art technology of that era.

Oddly, for many of the world's largest ad agencies -- the very same ones that have spent most of their recent years convincing clients to transition to newer, cloud-based transformation technologies -- it still is.

The technology most of the major agency holding companies use to process their media buys is still running on IBM 360 Assembler, a mainframe computer technology developed by the company in 1964, according to Ron Levy, president of ad agency technology consultancy Ron Levy Associates.

It was the same technology running on the IBM machine depicted in the offices of "Mad Men's" Sterling Cooper & Partners.

It was the same technology used by Michael Donovan when he founded Donovan Data Systems in 1967 and throughout the 1970s when Donovan became the service bureau processing most of the world's biggest ad agencies' data -- everything from their media buys to accounting to payments to the media, and even payroll to employees.

Over the years, big agencies have sourced parts of their processing to other technology suppliers, or built some of their own, but the core of their media-buying still runs on at least some descendant form of that IBM 360 Assembler processing, even more than a decade after Mediaocean was created by the merger of Donovan with a newer, state-of-the-art technology challenger -- MediaBank -- in 2012.

"I believe they've found some new hardware vendors, but some of that software is still running," Levy explains, adding: "They have put a pretty face on a lot of it."

To be fair, Levy points out that many other industries -- banking and the travel industry, for example -- still have that same legacy technology running in their background, for reasons that likely are similar to the ad industry's: because they are stable, secure, have long historical databases, and because the people who use them are reluctant to change.

"Despite [Mediaocean's] best intentions, they have never been able to migrate any major agencies over from one system to the other," notes Levy.

Levy should know.

He was part of the team at Datatech, one of the first systems challenging Donovan's dominance of the ad industry in the early 1980s.

Unlike Donovan's IBM mainframes, Levy recalls Datatech's technology was built on then relatively more state-of-the-art Unix system developed by AT&T's Bell Labs, which could be installed inside the agency so it could store and process their own data.

Datatech's first client, ironically, was the same agency that helped launch Apple Computer -- Chiat\Day, now part of Omnicom's TBWA\Chiat\Day -- but quickly began racking up other big shops, including Foote, Cone & Belding, which following a series of mergers, became the basis of what is the modern-day IPG Mediabrands, all of which are now predominantly Mediaocean clients running on some form of the old IBM mainframe technology.

Ultimately, Datatech was acquired by former Donovan rival MediaBank to form Mediaocean and ended up being run by MediaBank CEO Bill Wise, a former Yahoo exec, who was focused on transitioning the ad industry from Donovan's legacy technology to newer, state-of-the-art ones designed for the digital media age.

For reasons already noted in the preceding article in this series ("Hudson Hawks: How A Promising Agency Tech Startup Struck Out With Holding Cos., Stiffed Lots Of People"), it wasn't entirely because of Mediaocean's lack of trying.

Even as technology evolved, the culture inside big ad agencies hadn't changed much since the one depicted in Episode 4 of season seven of "Mad Men."

The reasons are multifaceted, according to a number of agency tech experts interviewed for these articles, but mainly because big ad agencies fear losing centralized control, as well as historical records, of financial data used to service clients and pay the media.

That fear of dislocation was a major reason why so many shops were loath to transition from Donovan's systems through the 1970s, 1980s, and 1990s, and even by 2008, it took a remarkable act of bravery for Publicis' Starcom MediaVest Group to back VC-backed startup MediaBank.

And that move wasn't without a great deal of technical and legal pain, which in retrospect, now seems for naught as a newer venture-backed startup -- Hudson MX -- promised once again to develop Madison Avenue's better technological mousetrap. Or what Hudson MX insiders touted as "the agency operating system of the future."

In hindsight, there are many things Hudson MX did wrong.

It chose the wrong big agency holding company -- Dentsu -- as its big beta client to develop a new enterprise system around, because in retrospect Dentsu didn't seem to know what it wanted. That ate up time, resources and precious capital for Hudson MX, contributing to its restructuring in February and raising questions about its long-term future viability.

It over-promised and under-delivered in pitches and joint development projects with other big agency holding companies, including Publicis, Omnicom and even pure-play independent Horizon Media.

During meetings with Horizon's top technology executives, Hudson MX guaranteed that at least two of the major holding companies were either in the process of transitioning or had iron-clad contracts to do so.

When the Horizon team brought an initial Hudson MX proposal based on those assurances to Horizon founder and CEO Bill Koenigsberg, they told them to hold off and proceeded to call around to his peers at those holding companies -- only to find out they were not actually switching to Hudson MX from Mediaocean.

Despite the misrepresentation, Horizon executives continue development talks with Hudson MX in the hope that it can fulfill on the vision it has laid out to them. The same is true of IPG Mediabrands, which has had several rounds of on-again, off-again development projects with Hudson MX, but which also remains hopeful that something can come from the startup, despite its burned capital and promises.

The reasons IPG and Horizon continue to talk and remain hopeful will be outlined more in the next article in this series, which is about the potential future of agency enterprise technology. If the lessons of the past are any indication, it likely will take a company like Horizon -- not wedded to the big holding company culture and processes of the past -- to break that log jam.

In defense of the assurances they gave about holding companies being on board with them, Hudson MX executives say they were told that and even given contracts in some instances, so they believed it.

The contracts and the promises, they say in hindsight, had "out clauses" and some chicken feet behind them.