We’re at the Rosewood Sand Hill, optimally close to Kleiner Perkins and the banks that whisper out our future like the oracles of air. No conspicuous consumption; in fact, we’ve moved beyond consumption into non-consumption, all electric and non-meat, so faithful are we to our own shared abundance.
And onto the stage comes a slightly seedy character, no longer young, ironic in the way of the 1990s, inhabiting a gestalt that’s gone. It’s me, in fact: about to argue yet again the importance of forgotten history to a room full of futurists. This is what you may have missed …
1. We Start with Netscape
The internet predates the browser, but the browser was the box on which we built the web. Before Mosaic and – more significantly – its virtual spin-out Netscape, launched in 1994, the internet consisted of academic cultists and gardens like AOL and CompuServe with walls so high they blocked out the sun; they weren’t the answer.
Mosaic was “achingly beautiful,” in the words of Vint Cerf, a project emerging from protocols developed by the British physicist Tim Berners-Lee (HTML, HTTP). Mosaic put images on the web, after a battle. It made hypertext easy. Developed by unpretentious hackers at the University of Illinois’ National Center for Supercomputing Applications (NCSA), it was profiled in the New York Times, which failed to mention its actual makers, Marc Andreessen and Eric Bina.
Thus stung by lab politics, Andreessen and then Bina and others agreed to join Silicon Graphics’ founder Jim Clark in the Bay Area to build a better Mosaic. It went public in 1995 and started the dot-com boom, through no fault of its own. Everybody was excited by Netscape; especially Microsoft, which promptly tried to bury it.
And while the academic internet was explicitly anti-commercial, Netscape was not. Nor were many of the publications that bravely stood up online, such as HotWired.com, the digital porting of the digiterati’s lifestyle manual: Wired magazine. Portals like Jerry Yang and David Filo’s link roster Yahoo! were initially anti-ad – as was Google at launch, five years later, — but ads got inevitable. From the beginning, nobody seemed to want to subscribe to bits on a screen or use a credit card online.
Ads were the only way to pay.
On the Netscape homepage was an early “live cam” (updated every few seconds) of a fishtank. It sat on the desk of a 24 year-old engineer named Lou Montulli. Now Lou was not part of the original NCSA cadre but was merited in due to his popular text browser Lynx, written while he worked the help desk part time in the Computer Lab at the University of Kansas, between racquetball games.
And Montulli of course – as everyone who has followed me since my Gartner days knows – invented the browser cookie. As he told me, it was to enable a shopping cart in the stateless-by-design internet, and immediate adoption by ad workers was not in the plan. Montulli decided to enable third-party cookies by default not because he wanted to invade anyone’s privacy but quite the opposite: he wanted the whole thing to be unobtrusive and transparent.
In other words, he made the engineer’s mistake of assuming that because a feature exists (we have always had the power to turn off our cookies), it actually exists. In fact, nobody bothered to look.
Lesson #1: Whoever owns the browser software, controls the Internet (i.e., Apple, Google).
2. The Dot-Com Ad Network Boom
In the 1990s, everything was an ad network. There was no real-time bidding on anything, and even big publishers didn’t have a big online ad business. Scaled large websites like NYTimes.com and WSJ.com weren’t scaled enough to attract TV money, at first, and smaller websites (i.e., everyone else) had neither staff nor knowledge, buyers nor technology, to extract rents from the web on their own.
So we have the ad networks, which do the work of signing up lots of publishers so they can go to ad buyers and offer them something they might actually want – millions of impressions across thousands of websites, never mind where. A natural extension of this middleperson model is the software that sits between publishers, ad buyers and agencies, deciding which ads to run, when, and counting what occurs.
This software is called an ad server, and many ad networks built their own ad servers, or tried, until the leaders emerged. This emergence happened quickly: by 1996, DoubleClick was dominant, and when it acquired on-premise rival NetGravity in 1999, it was super-dominant. People forget DoubleClick started as an ad network, and its ad server was a tool for the sellers. SaaS as a margin machine wasn’t real until the Y2K.
DoubleClick had the virtue of swagger; it was unapologetic in its quest for excess. Its hyper-charismatic, raven-haired co-founder Kevin O’Connor threw out business plans that were too cautious: “How does that lead to total domination?!” DoubleClickers loved him, and everyone in New York media wanted to work at DoubleClick in 1999. I tell you this from memory.
Silicon Alley – triangulated roughly around the Flatiron region, cheaper than midtown and convenient to the R and 6 – was a publicists’ pitch pushed by DoubleClick and the city of New York. The company’s logo was on just about every lamppost in lower Manhattan. Skywriters flew over beaches in the Hamptons telling ad buyers their campaigns were safe, the machines were on it.
And they were. Business was built – again, absolutely openly – on the compilation of a file against every DoubleClick cookie and I.P. address, appended with third-party data like location. With almost every premium publisher in its network, DoubleClick had a cookie on all our browsers, and so read much (not all) of our etheric adventures.
It was only when O’Connor acquired Abakus to insert real names and addresses into the pseudonymous cookie profile did USA Today, and then everyone else, raise a hand. But this was in 2000-01, when we all had other problems.
There was one outlier, a brave ad network that took a principled stand against dropping third-party cookies on browsers, for privacy reasons. It was a sortie that did not connect. In the 1990s, WebConnect was most definitely on the wrong side of history.
Lesson #2: Ad dollars flow to places with more information, not better ethics.
3. The Rise of Retargeting
Retargeting is the killer app for digital ads.
There is a very simple reason for this: it works.
In the 2000’s, ads targeted based on specific things we did in our browsers got at least 3X better engagement than non-retargeted ads, however we measured. Of course, many people claim to have invented this miracle-grow: Advertising.com, TACODA, MySpace, Criteo, BlueLithium/Yahoo. But it seems to have been invented quite naturally by DoubleClick, sometime around the dot-com crash.
Retargeting works, but it has a signal flaw. It is very easy to notice. In fact, we might say most civilians did not even suspect the persistence of cookies and targeted ads until retargeting gave them a clue.
Slowly at first, and then with more persistent vigor and rage, the ad-that-followed-me combined with a generalized uptick in paranoia to create a climate of conspiracy around the ad business. We went from being Mad Men to madmen with malodorous intent.
We have the spectacle, in the later 2000’s, of otherwise circumspect journalists reporting their businesses closer to the grave with richly-researched deep dives under ominous titles such as “The Privacy Project” (“I Visited 47 Sites. Hundreds of Trackers Followed Me” – New York Times) and “What They Know” (“They Know What You’re Shopping For” – Wall Street Journal).
Web surfing for the cultural elite assumed the soundtrack of a horror film. Yes, they probably do know what you did last summer.
Looking at Google Trends data over the past two decades, interest in ‘tracking’ has indeed swelled – even as interest in a randomly-selected fad (in this case: ‘Bieber,’ in red) has fallen off the cyber-cliff.
Lesson #3: In the long run, it is more profitable to be gentle with your superpowers.
4. The Epochal Moment Called 2018
Everything changed in 2018. Certainly Snowden had an impact, five years earlier, but his concerns were largely abstract, existential and political. He and his outraged adherents had much more important things on their minds than mere ads.
Which brings us to Mark Zuckerberg. He appeared in front of two Senate committees two years after the 2016 presidential election and some time after a German magazine revealed Facebook had given unredacted network data to a Cambridge researcher, who’d sold it to a conservative consulting firm called Cambridge Analytica. (Meta settled this case only last month, by the way.)
Nothing happened out of the senators’ naive posturing, but an estimated 80 million people saw part of the hearings – and everyone knew they were on. This was a yellow-card moment in digital; another ping at the populus. This time, for the first time, we started to look at our phones with suspicion. Our apps were watching us too.
By no coincidence, 2018 was also the year that digital ad spend overtook offline – that is, digital won – and the share of the triopoly (Google, Meta, Amazon) exceeded 80/20, including search. Digital became the dominant dimension in ads, and a handful of mastodons maneuvered for control.
Zuckerberg’s appearance in 2018 almost forced Apple to do what it did, two years later.
Intelligent Tracking Prevention (ITP) and the App Tracking Transparency framework (ATT) flexed the power of the browser and OS owner. We noted this with Netscape, and we see it again. Google was peer-pressured into blog posts and a Privacy Sandbox of professorial interest, while Apple filled a vacuum left by our Congress, appointing itself Neighborhood Watch for the World Wide Web.
Two things to mention here:
One, Apple’s specific tactics aren’t as important as its heuristic decision to make data collection contingent on an explicit opt-in by the user. We need to say “Yes.” Which makes sense until we ask the logical question: “Yes to what?” And we’re caught in the sepulchral eddy of the Privacy Paradox, which states that it’s almost impossible to make an intelligent decision about whether we do or do not want to “opt in” to collection of … what? … for what purpose? … and by whom?
Most of us don’t care. We click yes or no depending on our prejudice. (I always click yes.) Also, whether we’ve heard of the brand. And we’re grandfathered into the big apps like Amazon, Instagram, Facebook, TikTok. In fact, we have no idea what they know, and they don’t really have to tell us.
Two, nobody wants untargeted ads. Trust me on this. We might romanticize an experience of anonymity, but I challenge you to live it – as I did, — disable your trackers, obscure your I.P., use a VPN, and just look at the ads you’re served up: a late-night-cable-TV hodgepodge of personal pizza, car caddies, weight loss lies and class-action lawsuits.
Irrelevance has never been so loud.
Lesson #4: Apple is taking the place of a well-ordered Government today.
5. What’s Coming Next
As much as we crave good direction, howling in despair, the ad market still obeys its formulas. We need addressability and accountability; targeting and measurement. We target based on real information, inferred information, at a person- or a cohort-level, and then there is context (time, publisher, page, device). Measurement is an art form in itself and has never been exactly precise.
Building a campaign, we start closest to the point of decision – if we can – and build out from there; again, starting with those places where we have the most control, or think we have. So many times we start with paid search because it’s (1) close to the point of decision, and (2) in our control. We’ll overpay to meet these criteria. And so on, out into the wild terrain of unknown ad networks and disconnected TV.
Given the principles of proximity to a sale and greater information, where do we look?
Until it perfects its in-app storefront, Facebook/Instagram is hampered by its distance from the end zone. That’s why it needed the mobile ad ID (MAID). So-called retail media – which is living its own wonder years right now – can place ads close to the point of purchase and record a sale. It’s an ideal environment limited only by immature tech (in most cases) and the narrow data sets.
And it seems very safe to me to say that we are in desperate need of new ad formats.
Paid search is the most successful, and it was entirely unpredicted. Banner ads feel like a retreat to the mean. McLuhan tells us the content of any medium is always another medium. Digital is growing up and out of that, at last.
Look at the #Influencer phenomenon, reaching perhaps $13 billion this year. Those are blatant product pushes, in our pockets, from figures not too proud to beg – and they work. That’s digital for you. What about product placement? Imagine tools that paint products into sets (and eventually dialogue) without a seam. They’re coming soon. I call these “non-ad ad formats.”
Soon we’ll all just call them ads-as-usual.
So where does that leave the hard-working brand? You know that you need first-party data, but it doesn’t have to be your own. If you have it, scale it; get a Customer Data Platform (CDP). If you don’t, you’ll just need to ride along with someone else’s, once Montulli’s cookie’s gone. Don’t worry, though: those ‘someone else’s’ (Amazon, Disney, NBCU, Uber) – they all know you’re coming.
They’ll be ready. There will always be a way to advertise.
Happy New Year!