The Enshittification of Digital Marketing
Companies are often founded because the founders have a mission, see an unresolved problem, spot and solve a market inefficiency, or think ahead and create something disruptive and revolutionary in comparison to existing products. While in startup mode the focus is on creating innovative products and/or services. The first few clients will happily provide feedback which is used to further innovate the products and/or services. Quality matters! The company grows, a few more clients have been onboarded, the products have matured and the company is ready to scale. The startup becomes a scaleup and before you know you’ll have 20 FTE on the payroll.
Fast forward a few years: The company went public and the company now is the market leader in its segment. Each quarter the executive team will have a shareholder’s call to present the results and numbers accompanied with the achievements, press release, etc. At this moment in the maturity cycle of the company who or what influences the success of the company? You might think: “the good product?”. But, will a better product get more market share? Nope, the current market share is mainly the result of marketing and sales and the financial success is based on the legal/ accounting department having optimized profit and profit per share leveraging international tax structures. A better product wouldn’t make a big difference, simply because the company already is market leader.
Hence, the best marketing sales, accounting and/or legal employees are promoted, and over time the company will be run by these very same people. And the original great product people? They’ll get frustrated and eventually leave. As technology is going forward at an ever increasing pace their knowledge and experience is welcomed at new startups and scaleups. How do the public companies grow and innovate? By making innovations that aren’t in the best interest of the advertisers, only for the shareholders (and the bonuses). For example, platforms that by default enable audience networks, promote showing your advertisements on long tail websites, goal based campaigns aka performance max campaigns.
This described lifecycle (startup and product creation → scaleup → going public → decline) is typically what happens when shareholder value > product.
Remember how great Facebook was (a long long time ago), and then your parents and aunt Judy sent a friend request. How your feed only contained posts made by friends and wasn’t flooded with advertisements or promoted content?
Remember how ebay had great deals where non-professional people would sell from their attic on the cheap? Then the pros came in, first with auctions with a reserve, later with a fixed price, then promoted with preferred spots. The same happened with the Dutch version: marktplaats.nl
Remember how Google search once showed great results? Just take a look at the results the engine throws out: First results are promoted, and the other results showed are not in your interest but to maximize profit, ie. clicks on advertisements. These results contain MFS (made for searchresults) websites: keyword-optimized sites and blogs; showing programmatic ads hoping to monetize your clicks.
This decline of service doesn’t improve the user experience. Cory Doctorow came up with the word enshittification to describe this lifecycle [3] from the end user’s perspective.
enshittification (ɛnʃɪttifaɪˈeyeɪʃən) ● uncountable, The phenomenon of online platforms gradually degrading the quality of their services, often by promoting advertisements and sponsored content, in order to increase profits.
Platforms (both social media and large media sites and even webshops) sell advertisement space on their platforms as a stream of income. To monetize this they’ll need real human users to consume those advertisements. When their service was great to the humans using their service, the humans would gladly stay on the platform and thus consume the ads automatically. As time progresses, the companies running these platforms need to keep showing growth to their shareholders. With double digit growth sooner or later they’ll run out of humans. What to do? Supply and demand would dictate increase the price of advertisement slots. But, not in online marketing, because of the fear that advertisers will run away. So, the alternative is to bend the rules (Youtube showing targeted / personalized ads to children’s channels)[8][9] and dilute quality by allowing 3rd party audiences to be part of the platform, eg. FAN (Facebook), Pangle (tiktok), MSAN (Microsoft), TrueView (Youtube)[10], etc. just to keep hitting targets.
Audience networks and programmatic make money through a so called attribution model, ie. paid per impressions, clicks, etc. To survive financially they need to attract lots of humans, which is expensive and near impossible at scale at these low fees. This is where the demand for artificial traffic and inflation of numbers (bots, ad stacking, transparent ads, continuous reloading carousel, MFA, loading the video-ad completion pixel, browser extensions injecting ads, mobile apps loading apps in the background, etc, etc) comes from. To prevent paying attribution to bots and fraud these platforms have their own fraud verification tools and/or partner with a fraud detection vendor.
Fraud verification vendors
The two most prominent fraud verification vendors in the ad-space originally started as innovative companies. Their solution matched with the back then technology required to detect and flag bots, click fraud, etc. Again, these companies grew, became public listed, and a market leader duopoly in the ad verification segment. As described above quarterly numbers became more important than innovative technology: Shareholders value > quality of product (the product is: ad fraud detection).
And the fraudsters? They continuously improve their bots. Once detected by the large or any of the small innovative companies, they’ll adapt, learn and update immediately. They don’t work for VC money, or shareholders demanding quarterly growth, they generally work for themselves renting out their service. Quality is key to their success.
The lack of quality detection by the duopoly also helps the platforms hiring or partnering with the verification service in exaggarating their numbers. If they would report the truth (as Oxford Biochronometrics sees it), which is way more than the reported 1%, both (platform and ad verification company) lose. The relation between ad verification companies and platforms is more symbiotic than principal--agent [5]. This relation causes brands to overpay for the given service (buying advertisement space), of course the additional ad verifcation price is included in the price.
What can be done about this?
When you have for example 20% fraud in your campaigns it absolutely doesn’t mean that 20% of your budget directly goes to fraudsters. As the ANBA PWC shows[6] in programmatic half the money the advertiser spends ends up in the supply chain (the middlemen), the other half goes to the publisher. The ~50% going to the publisher is again broken down, run an infrastrucure, create attractive content, although when renting subprime traffic (bots), they’ll just hit your site without quality content. To break it down only a subset of this (max 10%) is attribution for clicks, or generated leads. That’s the part ending up in the fraudster’s bank account, IF they are not detected within the invoicing term. If they are, they’ll get nothing, except a monthly invoice from their cloud provider running the botnet and the residential proxies they have been using, etc.
So, how big of a problem is this? I’m sure if ad fraud would mean $100B less collected taxes/ annum, the problem would have been addressed. But, as a lot of the fraud sticks at middlemen and they do pay their taxes; it’s just a market inefficiency. And the fraudsters? They either operate as a legitimate business and thus pay their taxes, depending on the country they operate. In the end, if you generate a lot of bot traffic out of thin air: besides some cloud costs: it is almost all pure profit! And at these volumes and dollar amounts it is well enough to fund large troll farms, disinformation tactics and be able to influence elections, and probably even help evil regimes.
The only way to break this cycle is when companies treat marketing as an investment in future customers and sales, and monitor the investment as such. Clicks, views, how many marketing qualified leads (MQLs) are not important. It’s about quality and: Sales (closed deals), Customer Acquisition Cost (CAC), Customer Lifetime Value (CLV) the CAC:CLV ratio (minimal > 3, preferably > 5), the churn rate. This can only be achieved by continuously measuring the output, remove the poor performing parts and start (re)allocating and optimizing your budget to where it works. Stop blindly spending huge amounts of money, stop using low quality audience networks, use a whitelist of domains where your ads are shown, use a whitelist of mobile apps, realize that the lowest CPM is probably the lowest quality (bots). The most basic first step is to start measuring fraud at your landing pages (click fraud arrives at your landing page) and use the output to know which sources and campaigns are clean and which stink [7]. Also, when selecting a fraud verification company: where in the life cycle are they? Still at a great quality product? Or pushing for deals preferably with a 3-year locking to hit quarterly results.
Once you’ll have a great fraud detection you’ll see the cumulative effects over time: Compound returns on the continuously reoptimized campaigns, clean data, know who your real audience is (and not bots ,or click farms with fake profiles), etc.
Want to know more? Comment, connect, DM
#adfraud #digitalmarketing #frauddetection #enshittification
[1] https://en.wikipedia.org/wiki/List_of_mergers_and_acquisitions_by_Alphabet
[3] https://en.wiktionary.org/wiki/enshittification
[4] https://en.wikipedia.org/wiki/List_of_mergers_and_acquisitions_by_Microsoft
[5] https://en.wikipedia.org/wiki/Principal%E2%80%93agent_problem
[7] https://www.linkedin.com/pulse/how-create-b2c-data-driven-lead-generation-pipe-part-3-kouwenhoven/
[8] https://adalytics.io/blog/is-there-evidence-of-personalized-ads-on-made-for-kids-youtube-videos
[9] https://adalytics.io/blog/are-youtube-ads-coppa-compliant
[10] https://adalytics.io/blog/invalid-google-video-partner-trueview-ads