Making Ads That Don't Suck

    The following is a concise (read: as brief as I get) summary on how modern online advertising works, as of October 26, 2016. Keep in mind that this post will probably be live long after the following model has changed.

    There’s a fear among people who don’t actually pay for advertising online that companies like Google store personal information in order to exploit that information for monetary gain. They think that Google wants them specifically in order to target them specifically, so an advertiser can know exactly who they are and give them an advertisement that will make them buy something.
    That fear is the result of an oversimplification of how online advertising works… to a fault. My brother owns an advertising agency and so I get to watch first-hand exactly how this system works.
    Yes, Google owns a company that tracks you online, for the purpose of gathering data about your habits, what you buy, when you buy it, and why. With that data they build a model of you, a predictive mathematical image enabling them to predict what you will probably do. Some say that these models know you better than you know yourself (that may be true, based on the levels of self-delusion many of us practice in order to avoid pain).

    Here’s the thing.
    Google isn’t just doing this to you. They’re doing it to everyone, everywhere, all the time. That means a universe of data. They don’t care about you as an individual because you as an individual don’t represent a significant enough monetary gain as to warrant their attention. Instead, they’re interested in types. 
    An advertiser is interested in commerce.
    They may make as little as 1% or as high as 70% (often considered Apple’s profit margin) on any given product they sell. The only way that method of sales works is by ignoring individuals and instead focusing on groups of people. What kind of person might buy their product? It’s not about individual sales. Focusing on individual sales on low-margin items doesn’t make any sense.
    Of course this changes when the price of an individual item is extremely expensive such that the 1% margin becomes sizable. See: mansions, yachts, skyscrapers, institutional investing, etc.
    But, for almost anything most people buy, it’s not about the individual because what they buy isn’t expensive enough.

    Let’s say I’m crowd funding for a movie I want you to know about. Let’s say my movie is a realistic portrayal of computer hackers. Let’s say that movie is called ALGORITHM.
    What kind of person would be interested in a realistic hacker movie? How many people could tell the difference between a realistic and a fake computer hacker movie? Well, computer science majors. In other words, college graduates who make between $80k - $5b a year. Mostly male, between the ages of 22 - 60.
    I don’t care about the specific hacker, unless they’re an influencer, like Steve Wozniak, John Draper, Kevin Mitnick, or industry specific journalists, such as Leo Leporte, Steve Gibson, and Tom Merritt, and of course hacker conferences like HOPE, DefCon, or 44Con. For all of the above, I would not use online ads. I would try to meet them specifically and let them know about the project.
    When I try to buy my online ad via Google AdSense, I tell Google I want to target college graduates, males, between the ages of 22 - 60, who like sci-fi. I get get as specific as I want. I could even target it by region and by time. I also set specific search terms so when people look up “Edward Snowden” they might see an ad for ALGORITHM.
    When all those things have been set, I then tell Google how fast I want to spend money. I might say $100/day for 30 days. Google will then do all the buying and placing of ads until the money runs out. Of course I get a pretty detailed map of which ads worked, how well they worked, which search terms worked, and a general map of what group clicked most. This is so I can refine my ad campaign to get more clicks for my money.
    Again, it’s not about the individual. All that’s pretty straightforward.

    But, there’s a huge systemic flaw in the business model and it involves several factors, all of which I’ll get into.
1. There’s no vetting for quality.
2. There’s no vetting for viruses.
3. Most website now looks like a pile of garbage through which users must sift to find the data.
4. The targeting doesn’t always work and people end up seeing ads that aren’t right for them.
5. The nature of ad payments lends itself to a degradation in quality over time.

    There’s just too much to accurately curate. Quality unfortunately rests on the shoulders of the ad creator, many of whom have no taste. Thus, this issue is unlikely to go away, ever, as long as capitalism or any other form of trade or information propagation exists that involves human decisions.

    (I use the term “viruses” to refer to any kind of malicious code. Sometimes it’s a worm, other times a trojan, other times an actual virus. So, for the sake of the pedantic readers, it’s a general term.)
    Not all ads have viruses. Most don’t, in fact. But that’s not the point. Some do and if any do, the potential for damage is huge because of the connected nature of the modern Internet.
    Most users aren’t sophisticated enough to have an ad blocker plugin installed on their browser. The ad will appear to them and infect their computer. They may not even be aware that their computer is infected and infecting others.
    The auction model doesn’t really allow for protecting against this, simply by the volume of ads that get sold. Thus, viruses continue until the O.S. manufacturer patches the software, like Apple did with the .jpg hack on October 10, 2016.

    People have now been taught everything online should be free. Since making good, cool stuff costs money, content producers are left with quite a problem: paywall or ads.
    A paywall is when content is restricted unless the end-user has a subscription. See: Foreign Policy Magazine or Netflix as examples.
    Movies are huge and without a subscription, they’re only available on file-sharing sites, which are usually only link-brokers rather than actually storing the huge files. Also, movie studios and production companies police the distribution of movies like the Stasi Secret Police. Thus, the issue news sites face doesn’t really transfer to movies/shows.
    However, when a news site has a paywall, usually some other lesser non-pay-walled site will summarize the pay-walled article. The non-pay-walled sites fund their thievery or occasional (usually substandard) original journalism through ads, thus, their pages will be littered and end up looking like the aforementioned piles of trash. They will often have so many ads that the bulk of the delay in downloading the site will be the distribution of ads.
    Some of those ads will be still pictures. But, advertisers have realized that over time people have learned to ignore the still-image ads. And advertisers havce adapted. Currently, when the ad is just an image, it will move slightly and slowly, but just enough to grab your attention.
    Some text sites even allow full audio and video ads that play automatically. These are the worst form of bad design.
    I’m ADD. This means that whatever’s happening in front of me, whatever is moving, flashy, shining, changing, or blinking is going to take my attention away from what I’m doing. That’s why I don’t write near other people, including the now cliche coffee shop. I brew instant mocha at home, and drink and write at home.
    Because I’m ADD, when I see an ad designed to get my attention, on a site I’m visiting to get a bit of information, I usually have to zoom into the text. If I can’t do that on whatever device I’m on, I don’t visit modern news sites.

    The idea of targeting a specific individual with exactly the kind of ad they actually want to see is, at this point, a distant dream. But, it is a dream worth aspiring to because it solves a great many problems.
    Most people find most ads annoying because they’re invasive and selling something we’re not currently interested in buying. Because of this failure, the site hosting the ad suffers, the user suffers, and the advertiser suffers.
    In the ideal future we will only see ads for stuff we want. Everyone involved will be happy. We’re not there yet. And there’s no indication that this will happen anytime soon.

    Aside from the points mentioned above, there’s another key issue that actually leads to the degeneration in the quality of ads, and it has to do with the way advertisers are charged. Essentially, the better an ad is, the more it costs the advertiser.
    Let me explain.
    The least expensive ad to buy is the one that the user clicks away in 4 seconds, immediately after the “Skip” option appears. 
    If the end-user watches the whole ad, the advertiser is charged more.
    If the end-user clicks the ad and goes to the advertiser’s site, the advertiser is charged even more. 
    If the unlikely event happens, and a user watches the whole ad then clicks to the site, and actually buys the product, the advertiser pays the most.
    This model seems to make sense because it charges the advertiser more the more their investment pays off, culminating in a sale.
    But, from an end-user standpoint, it’s actually the same as mutation slime optimizing for the most offensive and useless ads possible. If the advertiser is smart, they will say the name of their product and how to buy it in those first few unskippable seconds. 
    And that’s really just tasteless.
    It would be like someone walking up to you and as they pass, handing you something you don’t want. But, from a money-spent/dollars-earned ideal, it’s the best investment. Of course, the results aren’t technically trackable, as with the click-through model. But, if brand/product recognition is the goal…?
    If someone watches through an entire ad, it means either the ad was amazing, or the ad was for something the person wanted. Either way, this is a radical improvement on drive-by ads. People want to watch good content, and a well-made ad is good content. See: Super Bowl ads.
    The pinnacle of all advertising is when the end-user sees something they want and then goes to a website and buys it. Even though this is the best possible scenario for everyone involved, it’s also the most rare.
    In fact, the less the advertiser has to pay for the ad to get the sale they want, the better it is for them, which means they’re going to usually gravitate toward the drive-by.

    The solution to all the above actually lands not on the advertiser (which will always choose the cheapest fastest method to peddle their wares) and not on the end-user because they have no say how information gets presented (short of not visiting most sites, which is tantamount to staying off the Internet).
    The solution must be implemented by the ad auction houses like Google AdSense. And the solution is simple: The less effective an ad, the more it gets charged. Drive-by ads would be the most expensive, decreasing in price until the ad experience that leads to a sale is the least expensive.
    This will instantly weed out crappy ads because they’ll be too expensive to buy based on the broadcast-to-every-possible-person model. 
    This will also eliminate the trash pile news sites because one or two well-designed ads will pay for the whole page.
    And that ultimately makes the Internet a better place to hang out, which will make everyone involved happier.