tl;dr: Modern advertising is powered by data-driven AI, and "Math Men" are becoming key to bridging where advertising was with where it needs to be as the primary monetizer of the modern internet (and making sure an agency doesn't bring a knife to a robot fight.) 


Introduction

The Internet of Ads. 

At 4:57 PM, February 1st 2016, Alphabet became the most valuable company in the world.

 

This came as little surprise to most people who know what the 'G' in Alphabet stands for.


If you asked a trader why Google was this valuable, you would probably get a medley of their hundreds of past, present, and future products, sorted by whichever is kicking this most ass at the time. 


Answers like:

But if you asked an accountant why Google was valuable, he'd give you a much shorter answer:

Ads.

On (financial) paper, Google is, first and foremost, an advertising company. 

We know this because during the Alphabet reclassification, Google moved any non-advertising revenue (like Google Play, Android, etc.) to something they like to call “Other Bets,” and in their Q4 2015 report, officially published the results of this new setup; advertising represented 90% of Google’s total revenue. 

In fact, Google makes so much money advertising it can afford to lose money on everything else.

This is not unique. Facebook, Twitter, Reddit, and many other places on the internet are betting on advertising to be their bridge to profitability.

Google was just the first (and best) at it:

 

 

 

 

 

 

 

 

This success can be primarily attributed to just two projects in the Google ecosystem, whose combination creates one the best ad platforms on the market today. 

 

Google-adsense-logo-2015-transparent.png
 

Step 1: Input Ads.

Google Adwords (and most other ad platforms) first allow advertisers to upload ad content in the form of images, videos, etc. alongside some definition of their target demographic (as defined by age, gender, location, interests, etc.) 

For $x, Google will give you y ad impressions.

 Google then processes these requirements and returns a price that's typically some derivative of a CPM (Cost Per Thousand Impressions; a modernized GRP) which varies based on how many other advertisers are bidding to reach that same audience. The advertiser then provides a budget for Google to bid with, and the robots get to work. 

One could think of it like a silent auction for every ad spot on the internet.

 

Step 2: Output Ads / Reward Users. 

For every impression you give Google, they’ll give you $z.

Google Adsense is the other end of that system; it focuses on ad delivery. It takes whatever ad content won the price bid, and serves it up wherever it thinks it can find that target demographic on the internet amongst its massive network of  'monetized' (ad-integrated) spots, which include most websites, apps, YouTube channels, and anywhere else on the GDN for starters, rewarding the owner of said spot with cold hard cash for every impression/video view/link click/<insert ad metric here> they give back to Google. 

 

 

This is the dual prize for the silent auction; your ad gets in front of your consumers, and whoever got you those consumers gets a slice of the money you bid. Everybody wins (in theory). 

Step 3: ???

Step 4: Profit. 

That's really all there is to it; by setting up a blind auction between these two 'consumers' and riding the spread between the bid and ask, Google has essentially built a self-serve advertising empire that may as well print money.

Not only does it empower any internet user to make money doing whatever they do so long as they share their traffic with a little ad box powered by the big G, but it also gives marketers the tools to sell their products to the end-user using the most powerful, widely distributed, and intelligent marketing tool ever designed by man. 

So why do ads suck?

The full answer to that question is much more complex than this post can go into. It is political, mechanical, technical, and sometimes even counter-intuitive.

However, after spending two years in the industry working alongside people at every level, I have a theory that can basically be boiled down to: 

 
Robots don’t understand advertising.

But before we get to any of that, let's briefly explore how we got here. 

 

Advertising's Past

From Madmen to machines. 

 

I love Madmen.

The AMC show based on the 'Golden Age' of advertising in the 60s set on Madison Avenue in the heart of New York City is easy to fall in love with, and hard to walk away from without romanticizing the industry a little bit.

Advertising in the 60s was glamour, money, power, trips, people, and adventures. It seemed like the Wild West with Italian-cut suits.

Back then, advertising was so well done it was actually looked forward to. Ads were infrequent, anticipated, and well-contained within 90 second chunks. And if you had creativity, tenacity, and a haircut like Don Drapers, you could build a legacy that could change the landscape of American culture with something as beautifully simple as 'It's toasted.

 
 

This is not the world we live in today.

Nowadays, we are used to ads more akin to the pre-roll that probably popped up in front of that Youtube video above. 

Used to an inconvenience that plants itself stubbornly between you and the content you're trying to consume, in what you could only assume is some villanous advertiser's bid to interrupt your day and steal your attention; advertising so poorly done and far removed from the experience that they often sell more ad blockers than any product they attempt to advertise

 

 

Some of these implementations deserve this sort of disdain, but it must be understood that these setups are not usually malicious, just misguided; a symptom of the deep misunderstanding of how ad tech affects its end-user. 

The honest truth is that most advertisers probably had little to nothing to do with where their specific ads show up, and just trusted Google to serve them up right after spending thousands of dollars creating and deploying all that stuff below the 'Skip Ad' button. Most will never know how you really felt about their product or ad.

In fact, the only thing they'll ever get to know is that his ad got viewed 1 more time, courtesy of Google.

160516.metrics.jpg

This is the new world; where neither the advertisers nor the advertised get to decide who gets served what when. 

Only the silent market. Only the quiet robots inbetween. 

Advertising has changed.

I'm not sure when exactly it happened; certainly well before I joined the industry 2 years ago.

When I moved to Toronto, I had no idea I would end up working in advertising. To me, advertising was  a problem of the internet. An infection I attacked viciously with adblockers, VPN services, and IP spoofing. It was the bane of the modern internet experience, and despite being behind the curtain, that part hasn't really changed; I can't stand most ads.  

To be perfectly honest though, I had no idea where I'd end up working. All I had was an undergraduate degree from UBC, a carry-on holding a crumpled suit, a 2-week lease of my buddy's couch, and a local network consisting of him and his dog. 

I was not exactly picky. 

After 2 months of odd jobs, apartment hunting, door knocking, and resume shotgunning, I got my first interview at an Advertising agency; Grip Limited. My first test was to analyze a Pizza Hut dataset to tell them how to improve their email marketing. I remember having to google what a 'KPI' was. 

I came back with a 12 page deck outlining the statistics of this campaigns performance, a go-to-market strategy for improving those results, and an ongoing KPI measurement framework that would keep the campaign healthy and on track.

Two days later, that report was in front of a client, who bought off on the entire project (and unbeknownst to him, me too.)

I was in.

I had never though my experience in finance, marketing, and economics would have landed me a job in such a 'creative' field, but it was a surprisingly perfect fit.

Agency life is fascinatingly strange, and far more varied than any other industry I've been a part of. Over the course of an average 9 to 5, I could be doing anything from building models to predict Honda car sales from website traffic, designing algorithms for the optimal budget distribution for $100K across hundreds of Facebook Ads, Twitter Cards, or Google Adwords, pitching strategies to transform huge brands at a fundamental level, or discussing the viability of brand new initiatives on cutting edge platforms over a neat Glenfiddich. 

I was solving unique problems for interesting people, and I loved it. 

 

Advertising's Present

The Data arms race. 

 

Over the next two years I continued to work in advertising, primarily under Mirum (a creative-oriented sibling of JWT under the WPP group) where I was eventually promoted to a Data Scientist role.

 
 

During that that time, I had crazy amounts of fun and many opportunities to polish my skill set, but I've also met all sorts of people who were...less than enthused by what I do. 

The majority of people I run into are extremely talented individuals that I love working alongside, but like any industry, there are always the few who make it harder than it has to be.

Whether they were ego-centric creatives who look at my spreadsheets with amused contempt, account managers begging me to find a way to make their big red numbers look like opportunities instead of failures, or fellow strategists who have stated to me plainly "I don't do numbers; I'm not a nerd", the core message was the same: 

Your numbers don’t belong in our Advertising.

'Math Men'; that's what we're affectionately known as the industry. Nerds who use the 'power' of numbers to support advertising efforts, whose once novel involvement as back office clerks is quickly turning into an arms race.

Why? Because, all joking aside, success in advertising has started to turn from a creative exercise into an algorithmic one

Advertising is still a fundamentally creative field, and the creation of beautiful, engaging ads has always been more art than science. However, lately the science part of that equation has started to evolve very quickly; faster than Advertising can keep up.

As an example, let us discuss one of the more comically tragic symptoms of this arms race, the hunt for the 'Millenial'. 

In today's hyper competitive market, almost every pitch and every agency I've run into  references the word 'Millenial' at least once. 

A 'Millennial' can be loosely defined as a generic 20-something North American who spends most of his or her time on the latest social network, has more disposable income than sense, and is constantly waiting for the next #Epic product to invade their iPhone app du jour so they can elbow Moms out of the grocery store line.

 

Speaking as a "Millenial", I can't help but find that thinking adorably hopeful (most of my friends are Millenials; we're a terrible marketing demographic), but I understand what it's stemming from; growing pains, and Fear Of Missing Out (#FOMO), not by consumers, but agencies. 

Why? Because Millennials don't need to watch TV. They don't have to listen to radio. They don't bother reading newspapers. They don't need to consume traditional media. 

Millennials are digital. 

They represent entire generation of consumers who basically rejected most popular platforms in favour of personalized ones, and in so doing jumped every old tool in the advertising playbook. This left modern advertisers scrambling for a means to tap into them, or be left serving hypothetical youth-hungry clients with an aging consumer base recruited manually via dying mediums. 

Left with obsolescence. 

So, driven by a need to tap into this new market, Mad Men came face to face with the internet.

I can only imagine the fascinated terror must've gripped those initial agencies. The terror of an entire economy being tilted by what would become the primary transformative technology of the 21st century.

For advertisers, this terror probably took the form of questions like:  

  • How do I transition from beautiful niche print ads, channel-specific commercials, and curated billboards to cater to an infinite, emotional, ever-changing void?  
  • How do I pivot to cater to web-based clients built by and for the internet when the skills of my craft lay in telling narratives in a whole other medium?
  • How do I advertise to people without ever knowing who those people even are
What the hell do we do now?

It was precisely that terror that the big G found a solution for. 

5Obhonwa.jpg

What if instead of evolving as fast the internet, advertisers could stay where they were already comfortable (creating pretty ads) and someone else worried about integrating it with the fastest evolving communications platform ever built? 

Someone who understood these new technologies at a fundamental level, whose integration with the modern internet could never be obsolete, and whose data on 'anonymous' internet users was as perfect as possible. 

Someone like Google

So Google et al. built the first digital ad exchanges; systems that converted the wild, unruly internet in into a level playing field  ripe for ad placement, while simultaneously inventing an easy way for any website to monetize their traffic, and turn visitors into value. 

The internet could be powered by ads, and anyone anywhere could make money doing whatever they wanted, courtesy of the silent backing of some brand somewhere who needed their visitors to see their latest product while they were visiting. 

It was absolute genius. 

 
Algorithmic Advertising.

Thus began the rise of algorithmic advertising; and the renaissance of the advertising machine.

Algorithmic advertising is simply shorthand for the class of bots powering this new market - bots often repurposed financial systems who could easily use navigate this ad bidding market to compete at a scale no human ever could; the scale of the real-time internet. 

Their task was simple: figure out where a target demographic hung out on the internet, bid for ad boxes in that spot, outbid competitors when necessary, and deliver the pretty ads we give you to the best possible users.

Buy low, aim right, repeat until out of money. Piece of cake. 

Truth be told, this advertising machine continues to work amazing well to this day, because it works with the literal persistence and diligences of a machine. The level of specificity behind ad targeting nowadays is nothing short of staggering.

Bots trading with bots.

However, the unintended consequence of  this system is that we created a blind market; a situation where advertisers never really know what context their ad will be seen in (since the bots make the calls), and consumers are being chased by weirdly similar ads wherever they go on the internet (just as we've programmed the bots to do). 

If you find that notion creepy or confusing, you're not alone. No human on either end of this platform thinks this is an ideal means of advertising. Advertisers consistently complain about the transparency of their media agencies, and that may as well be a quiet whisper compared to how consumers feel about advertising

The problem is that the platform in the middle isn't human

 

Advertising's Future

AI & You. 

So, what can this new ad robot doMore than it could do yesterday. 

Traditionally, advertising was a hopeful gamble of stylized creative attempts, specific media partner decisions, and a large amount of 'follow the leader'. A beautiful, but essentially shotgun-based approach.

If traditional advertising was a shotgun, then modern advertising is a sniper

fig12.png

Facebook is the second largest ad market supplier to build on top of what Google started.

They just took it up a notch. 

With the Facebook Business Manager, buying and placing ads became easier than ever, and it is now almost trivially easy to target cities, people, interests; damn near anyone to almost any level of specificity you can imagine.

As a social network, they know way more about you than just what you search (which is still a scary amount). They know who you are, what you like, who you talk to, and what you like to talk to them about.

What's more, Facebook ads have dropped the distinction between advertisers and consumers entirely; any Facebook user can advertise, and no Facebook user is free from advertising. 

They call it 'boosting content', and it's an insidiously brilliant because:  

 

Any Facebook user can create an ad for any amount, targeted at anyone.

Say you have something cool you'd like to show people, a rough idea who you'd like to show it too and where they are , and $7 in your Paypal account.

For $7 Facebook can make sure at least 1000 people see it.

To be fair, that's an expensive treat compared to Google Adwords, which can funnel hundreds of queries to your website for fractions of a dollar per click, but it has almost no learning curve whatsoever, and no barrier to entry. 

It is difficult to express just how transformative that is in the context of traditional advertising.

Even a few years ago, an agency with fewer than 20 people that made an ad with a production budget under $10K supported by a media budget under $50K would be hard pressed to get their ad anywhere near a consumer, and little to no idea who those consumers even were.

Now, one person can put up almost anything for $5, decide exactly who they need to be see it, have it in market before the end of the hour, and know exactly who saw it. 

In fact, I've done it. More than a few times:

No media agency. No go to market strategy. No primary objective. No consumer path. No persona generation. No universe analysis. No brand guideline. No channel-specific KPI framework. 

Just a creative friend, $5, and Facebook.  

The modern advertiser has more data, tools, and opportunities than they've ever had, and the sheer efficiency of these tools is quickly dethroning creativity as the only important factor in Advertising.

 

Creativity is a numbers game.

In the 20th Century, advertising ROI was a hopeful gamble based on historical expertise and a sprinkling of research.  

Creatives were the be-all and end-all of advertising, and there was no proof either way about what would and wouldn't stick to the wall, so they had the final say.

I do not resent that notion in any way; most old creative is genius, especially given the fact that they got there with just instinctGame recognize game

 

However, in the 21st century, advertising ROI can be quantified to the individual, so if your creative vision isn't engaging your audience, it won't get delivered, and a better ad will be swapped into it's place in the nanoseconds between a page is reloaded. 

Algorithms are not ambiguous. 

I am not saying this to offend old-school creatives. It does not bring me any particular joy to attempt to slice up your beautiful vision into descriptive numbers and comparable metrics, so forgive me for stating this next bit rather bluntly: 

Data doesn’t care what you made, only that it worked.

If it won the bid, hit the right people & delivered results, it has succeeded. If it hasn't, it's failed. 

This not my opinion; just a simple rule of the game as we play it today

Data is the currency by which any idea, creative or otherwise, will be be quantified and compared until only the best is left standing. And that winner will be served instead of the content the bots deem unworthy (often at great expense to the advertiser and his client.) 

Machines now decide what is and isn't good advertising, and if you can't quantify why your ad is good, neither can they

 

Conclusion

The future is friendly. Ish.

So why do ads still suck? 

  1. Advertisers are still trying & failing to use this absurdly powerful machine correctly.

  2. Bots march on, becoming smarter (and creepier) the more they learn about us. 

  3. Consumers are still uncomfortable with how well they chase us from site to site.

I created the following presentation as part of an internal series of lunch-and-learns called 'Mirum Next', in order to help my co-workers understand what I do a little bit better.

It provides an (over)simplified review of the fields powering Data Science, the applications of Data Science at Mirum, and a brief glance into the future of Data Science and AI:

Naturally, it is impossible to explain the value of Machine Learning, the role of Strong vs Weak AI, the usefulness of Tensorflow and cloud computing, the applications of Bayes theorem, the transformations of matrix algebra, or pretty much anything in 15 minutes, so please take this quick presentation with a grain of salt.

If you are another practicing data scientist, you will find the articles on KDnuggets or my Github learning repo far more valuable than this presentation, but for everyone else, consider this the shortest possible intro Data Science.

I presented it to show a few things:

  • This algorithmic dominance shows no signs of stopping, and has already transformed advertising from the inside.
  • These bots understand only data, so understanding your own data is imperative to communicating with them correctly. 
  • ROI is fully quantifiable, and its measurement is simply longer optional; it's expected.
  • The notion of using data to measure business performance is not new; only the tools

Data has shaken up much of traditional advertising (and indeed traditional business) and will probably continue to make more than a few jobs obsolete by the time it's done. Though it may make some more too; the need for high-level data analysis, predictive modeling, and superior ROI attribution are only going to go up, and all the data in the world won't make up for poor creative or inefficient management. 

GLaDOS_Based_On.png

I find this particularly ironic because these 'scary' technologies currently terraforming advertising are actually being pioneered by the very same companies funded by ads.

Google leads the world in AI, Machine Learning, and Data Science.

Their tools are powered by tech so cutting edge they may well as be from the future, but are still wielded by people who are just beginning to grasp how powerful they could be.

Therefore, given the investment in AI and it's adoption by some these behemoths of industry, there is no doubt we are heading towards this data-centric future

Regarding AI, we do not yet know whether we will run into GlaDOS or WALL-E first, but that's a still a long ways off.

21c59968ace29c8ee56f063e005b3375.jpg

For now all you need to know is that AI is the inhuman bridge between advertising's past and it's possible futures as the primary monetization vehicle of the modern internet, or a relic of a simpler time before we all subscribed to our software.

Understanding data is the key to navigating this quiet AI revolution, and deciding which of theses future we will have.

I mentioned earlier that robots are in charge of advertising. This was an exaggeration; they aren't really. 

Robots are no more in charge of our advertising than the humans in charge of those robots.

This post was just to shed some light on how this mechanism actually works at a high level, and to illustrate the need for agencies to realize data literacy is no longer a 'nice-to-have'; those who coast on the assumption that advertising is always going to be more creative than technical are essentially bringing a knife to a robot fight. 

However, It is worth nothing that if these tools are applied correctly, there has never been an easier time to be in advertising, and making great ads.

The silent robots powering our systems are built by some of the smartest people in the world, designed for maximum usability, and are waiting eagerly for a human that speaks their base-2 language to tell them what to do.

Understanding what these robots are capable of is the key to reviving the glory days of advertising, and the Math Men who calibrate them must never forget that advertising will always be as much an art and a science, and it is the union of these two fields that leads to truly game-changing work

Great work is not created merely by having great tools, but by great application.

So do not be afraid of the bots, for they are not more self-aware than your calculator, and all they want to do is help you do whatever it is your little human heart wants to do.

At least for now. 

</conspiracy>