Is ethical advertising possible? We all certainly know unethical advertising is possible, we live with it every day. The ad-driven Internet has created an insatiable hunger for personal data and as a result most of what the average person does in their web browser, or on their phone, and in real life is being measured, tracked, and sold to some degree. Yet if a company actually cared about your privacy and wanted to advertise its products, could it do so ethically? Can you track what a visitor does on your website without violating their privacy? We have been thinking about these issues heavily at Purism as we consider how to expand our marketing and in this article I’m going to explore where we currently are in our thinking.
Before metrics-based Internet marketing, companies would pay for a traditional ad in a magazine, on a billboard, or on radio or television targeted to the demographic groups they presumed would see it. The ability to track the performance of these traditional ads was limited. You might know approximately how many people might see the ad (through subscriber information the magazine or other media would share with you), but you wouldn’t know whether any of those people actually saw or read the ad. Maybe they changed the station when ads were playing or went to the restroom, and maybe they ignored the ads in the magazine.
You also wouldn’t have a direct link between someone seeing your ad and buying your product, just an indirect link between the money you spent on advertising (“ad spend”) and (hopefully) increased sales. If you wanted to figure out whether a sale came from a particular ad, you’d add some sort of custom coupon to it. That’s why radio ads to this day will say things like “Use the K105ROCKS coupon code to get 10% off your next order.”
Early in my career I worked as a sysadmin at an Internet direct marketing firm. They touted themselves as doing ethical direct marketing, which to them meant they were focused on ads and lead generation but were against spam and adware. This was during the early days of metrics-based advertising and marketing firms had discovered that the Internet provided them the ability to track an individual from seeing an ad, all the way to signing up for a service or buying a product (“the funnel”). They could collect and then sell the “qualified lead” (the contact information from the person who said they want more information about a product or service) to companies selling that product or service. The company would forward the leads to their sales team to turn those leads into a sale (a “conversion”). Companies would pay good money for leads to high value products or services because it’s much easier for a salesperson to sell something to someone who said they were interested, than “cold calling”–trying to convince someone, unsolicited, who hasn’t expressed an interest in the product.
Direct marketing firms built vast lead generation infrastructure to host websites with information about these products or services, always with some button on the bottom to pop up a form where potential customers could provide contact information. In addition, marketing firms would purchase ads on websites and search engines they didn’t control to point people to the websites they did control. To a direct marketer, the entire search engine results page is an ad.
When you search for information about a product, say new windows for your house, there are the ads at the top and side and bottom of the results for windows that point to a marketer’s lead generation platform. These results are labeled as ads and the marketing firm has to pay for them each time they are clicked (Pay Per Click). Then below the paid ads are the “organic ads” or what you think are actual search results that generate “organic traffic” to websites, but marketers view them as another form of ad. Those results also point to articles the marketing firm wrote about windows on their lead generation platforms, but the marketing firm doesn’t have to pay the search engine when you click those. Direct marketing firms are experts at SEO (“Search Engine Optimization” or gaming search engine algorithms to get a webpage top ranking) and will build their own websites with unique content specifically tailored to show up in search results. Because it can take months for a new website to get good ranking in a search engine, they will also partner with or outright acquire websites that already have good ranking for certain search queries, and inject their lead generation platform into the existing content.
If you ever wondered why it search results have gotten worse over time, this is why. It’s also nothing new. In the early days of the Internet, websites would add meta tags to their pages that would list the topics the web page covered, similar to modern hashtags in social media. Search engines would parse and trust those meta tags and use them to categorize those web pages in search results. It didn’t take long for people to game meta tags by injecting all sorts of topics unrelated to the page itself, so their site would get top ranking across categories. The first wave of search engines that used meta tags provided fewer and fewer useful results and ultimately folks like myself resorted to software that would query Altavista, Yahoo!, Ask Jeeves, and all the rest and then deduplicate and combine the results to give you a hope at finding a relevant web page.
Google became the top search engine because it didn’t rely on meta tags. It created its own PageRank algorithm based initially on which sites linked to other sites for topics, along with a few other metrics outside of meta tags that weren’t so easily gamed. When you strip away all of the gamed results, the results that remain are surprisingly relevant. This algorithm is constantly tuned to make it more difficult to game, yet if you look at the results of an average web page search today, it’s clear the algorithm is losing.
The lie marketers told us in the late 1990s and early 2000s (and that many in tech still believe as an article of faith) was that people don’t dislike ads, they only dislike irrelevant ones. The thinking is that if you could show someone an ad for something they actually wanted while they were trying to look at something else, they would like seeing ads. The problem is, to make ads ever more relevant, Big Tech firms have argued they must collect vast amounts of information about people to build dossiers on them that can predict their wants and needs. This dossier includes their search history, their friends, the websites they visit, the things they buy, their location, and other metrics combined with their general demographic data and compared against other people like them.
When people complained that they don’t like ads and don’t think relevant ads are worth the privacy trade-off, the ad tech industry countered that it’s just because the ads weren’t relevant enough. They just needed to collect more data. Now that ad tech has built vast databases on everyone, we are discovering the “uncanny valley” effect from relevant ads. The more relevant the ad, the creepier it is. Many people are convinced their phone is listening to them because they will have a conversation with their friends about something and then shortly afterwards see ads about it. That’s creepy but the truth is creepier. Modern ad tech doesn’t have to listen to your conversations to know what you want. It knew you wanted that thing before you told your friend about it, based on all of your other behavior online. You simply noticed this time because the topic was top of mind.
Are ads necessary? An incredible amount of what makes up the Internet today is funded directly or indirectly by ads. The companies and individuals who earn their living from ads of course view them as necessary (necessary for their survival even), but I mean outside of that ecosystem. At its most fundamental, an ad serves to inform someone of something they were unaware of, typically something they can buy. There are other ways people can become aware of things, such as word of mouth, so by that metric you could argue ads aren’t strictly necessary. Yet if you want to get the word out about something, they can often spread the word wider and faster than word of mouth alone.
While Purism has dabbled in ads and sponsorships a small amount in the past, it’s the exception. For the most part our marketing budget goes toward articles and videos on our site, and we have relied on promotion on our own social media accounts, word of mouth, an opt-in newsletter, press releases and the occasional review to get the word out about our products. The reasons for this are a combination our concerns about the ethics around modern marketing in general, and the challenge of trying to figure out whether our ads were effective (worth the investment) without violating people’s privacy. Spending $10,000 to get $1000 in extra sales is bad business, yet if you can spend $1000 to get $10,000 in extra sales, you might want to try spending $2000 next time.
We have long said that we sit on a three-legged stool of freedom, privacy and security. We have a diverse set of customers that value some combination of those three values, but our core customer base, especially initially, has been firmly in the freedom camp. A lot of our initial marketing efforts targeted this group the most, and it’s safe to say at this point that many if not most folks in the FOSS world are aware of Purism to some degree.
Yet there are many people who prioritize privacy or security who don’t operate in the free software world. Often these folks tell us this is the first time they have used Linux on a computer, and they found out about us because they just happened upon some third party article that talks about privacy- or security-focused products. Many of these folks would never know we exist strictly on word of mouth. We think everyone, not just free software geeks, deserve freedom, privacy and security, and that leads us to wonder whether we should invest in other ways to make ourselves known to people who have never heard of free software or Purism before. If it could be done ethically, ads could be an effective way to get the word out, especially to folks firmly entrenched in Big Tech with no knowledge there are alternatives.
The fact is that most of the approaches to modern marketing start with the assumption that you want to violate a person’s privacy. Most marketing tools make it all but impossible to use them without violating someone’s privacy and handing all their data over to Big Tech. When trying to come up with some rules of the road we could use as guidelines when trying to expand our own marketing, I realized pretty quickly there wasn’t a lot of prior art. Because there is so much focus on the worst privacy abuses Big Tech has caused, the conversation tends to be all-or-nothing. The main company I was aware of that was trying to use ads ethically was Linux Journal in the final months before it closed down, and the main related guidelines are in the GDPR, which is well-meaning but doesn’t go far enough and is already gamed by many websites.
We are currently in the process of creating guidelines for Purism so we can reach more people and make our site easier to use without violating our ethics and our Social Purpose. The guidelines aren’t finished, and even when they are, they will be something we review regularly for unintended consequences to make sure we are doing the right thing. Here’s a rough draft of where we are at the moment.
As a general rule we think data is toxic and avoid collecting data unless absolutely necessary, and when we do we delete it once we no longer need it. Yet some analysis of how people use our website can be enlightening, especially when it reveals areas where our web design has confused people or made our site hard to use. For instance, if we posted a link to the Librem 14 shop page on social media, and noticed that most people who started at that page immediately went to the product page before going back to buy a laptop, we might conclude people want more information than is on the shop page before deciding to buy. We might then use that analysis to conclude we should start people on the product page first. It’s also useful to know when a blog post resonates particularly well with the public. If we see a big increase in traffic, it’s nice to know why.
Traditionally we’ve accomplished this by doing basic analysis of our web traffic logs we retain for a short time for troubleshooting purposes. Most sites who want something more than that resort to Google Analytics or other plugins that send site traffic data to a third party. We obviously don’t want to do that, so we’ve decided to self-host the web analytics software Matomo. By hosting the analytics platform ourselves, we can ensure no customer data is shared outside of Purism, and no data is shared with Purism beyond what a visitor might already expect. Yet even in that case, the default Matomo settings do more tracking than we are comfortable with. To play it safe, we’ve started by turning off most of the bells and whistles so we can review the implications of each setting before we turn it on. Even then, we might make mistakes so we intend to review the settings we use routinely, because turning on a setting could have unintended consequences.
The modern version of a coupon code in a radio ad for the web are UTM (Urchin Tracking Module) codes. If you have ever seen a long string of text after the question mark in a URL that looks like “utm_source=foo&utm_campaign=bar” then you have seen a UTM code. Fundamentally, UTM codes function much like a “K105ROCKS” coupon code. They allow a website to know which platform incoming traffic came from. This is useful because if you are going to pay for ads on multiple platforms, you’d like to know which platforms were most effective.
UTM links can function similarly to the referral field in a web log that tells you when someone was referred to your site from another site. Yet modern ad tech has extended these codes (unfortunately) to do so much more. Instead of something like “utm_source=K105ROCKS” you might see something like “utm_source=9feq89235kzljf820”. Many modern marketing platforms allow you to generate custom codes that change per visitor. This allows you not just to track what website traffic came from, or which ad you showed, but which individual clicked on it along with other metrics about them.
When people buy something from us, obviously they share some personal information so we can ship a product to them, but beyond that we want to do our best to protect people’s anonymity when visiting us. The amount that individuals are tracked as they move about the web is a major privacy issue. So our current thinking is that if we use UTM links in an ad, it will only be to identify which platform the traffic came from, and what ad it was (so we can figure out later if it was worth the expense), but won’t use anything that could identify an individual. It goes without saying that things like tracking cookies are way past our red line.
Some people are surprised to know that in addition to our Librem Social and LBRY accounts, we have Facebook, Twitter, LinkedIn, YouTube and other social media accounts on Big Tech platforms (“dystopia”). Our stance on this is that we will have a hard time reaching people who live in dystopia and convincing them to move with us to utopia if we only promote ourselves in utopia. As I discovered myself, it can be incredibly challenging to remove yourself completely from a Big Tech platform and we can’t expect most people to do it all at once.
We maintain accounts on these platforms so we can share blog posts and videos explaining the privacy problems with Big Tech and inform people that there are alternatives. Similarly, when we start testing the effectiveness of ads, we will post them on ad platforms like Google and Facebook in addition to others. People who don’t yet know about Big Tech alternatives use those platforms, and those platforms are a prime way we can reach them.
Ultimately, we don’t want to be creepy. We value people’s privacy and want to protect it not just with our products, but with how we market our products. As we figure out the policies we will put in place around our marketing, we plan to perform constant “creepiness checks” to make sure we aren’t doing anything that crosses a line.
We also plan to share our policies with you, which will be inspired by a combination of our Social Purpose and our Digital Bill of Rights. Are there particular marketing approaches you find creepy and want us to avoid? Let us know, we are counting on your feedback as a sanity check to everything we do. Is ethical advertising possible? We intend to find out.