You have been to a made-for-advertising site, even if you have never heard the term. Say you reach the bottom of an article you were actually reading, and a widget offers you a row of chum: twenty child stars who aged badly, one weird trick, what she looks like now. You click — the machine has spent two decades learning precisely what you will click — and the page that loads holds a paragraph of text wedged between an autoplaying video, two sticky banners and a slideshow demanding forty taps of “next” to deliver no information at all. Somewhere around slide seven you close the tab, feeling slightly worse about the internet and about yourself.
That page was not journalism that failed. It was not journalism at all. It was a machine for converting one cheap click into thirty or forty ad impressions, and it worked exactly as designed.
The arbitrage
The business model is arbitrage, and there is a grim elegance to it: buy attention cheap, sell it dear. An operator pays a content-recommendation network a penny or two to deliver your click. The page you land on then serves a wall of programmatic ads — a dozen at a time, refreshing on a timer, stacked around a slideshow built so that every “next” counts as a fresh page view and reloads the entire wall. Each slot is sold in the same hundred-millisecond auction that sells every legitimate ad on the web. If your visit earns more in ad revenue than your click cost to buy, the operator profits. Everything else on the page — the words, the stock photos, the fifteen-part structure — exists to keep the wall legally clothed in “content”.
Hence the name the industry eventually settled on: made for advertising. Not supported by advertising, like a newspaper. Not funded by advertising, like this blog. Made for it — sites where the content is the pretext and the ads are the point.
Why the machine loves junk
MFA is not an abuse of programmatic advertising so much as its logical conclusion. Every metric the buying machine checks, a purpose-built junk site can beat an honest publisher on. Viewability? The ads are the page; of course they are viewable. Brand safety? Nothing is safer than celebrity-nostalgia filler with the flavour of wet cardboard — no politics, no tragedy, nothing to trip a classifier. Cost? Impressions by the tonne, at prices an investigative newsroom cannot match. A campaign that tells the machine to maximise viewable, brand-safe impressions at the lowest cost has, without anyone quite deciding it, told the machine to buy junk.
MFA is not a hack of the machine. It is the machine, optimised to a mirror finish.
And nearly everyone in the chain gets paid along the way. The agency’s fee scales with spend. The demand-side platform takes its percentage whether the impression lands on a broadsheet or a slideshow farm. The exchanges clear their fees per transaction. The only party unambiguously losing is the advertiser — along with every real publisher whose rates are dragged down by an effectively infinite supply of counterfeit inventory. Junk does not just waste the money spent on it; it cheapens attention everywhere else.
What the auditors found
For years MFA was a known smell without a number attached. Then the Association of National Advertisers — the trade body for the people actually paying for all of this — audited its members’ programmatic buying at the log level, tracing $123 million of spend and 35.5 billion impressions through the pipes. The first findings landed in June 2023, the complete report that December, and the numbers were worse than the pessimists had guessed: made-for-advertising sites took roughly 15 per cent of the spend in the study and served 21 per cent of the impressions.1
15%
The wider accounting was bleaker still. Of every dollar a marketer put into a demand-side platform, the ANA reckoned only about 36 cents ended up productively in front of an actual person; scaled across the roughly $88 billion open-web programmatic ecosystem, it put the recoverable waste at $20 billion or more a year.2 The average campaign in the study ran across 44,000 websites. No human being has looked at 44,000 websites. The entire point of programmatic is that no human has to — which is exactly how that 15 per cent walked out of the back door with nobody watching.
What happened next is, by ad-industry standards, almost heartening. In September 2023 four trade bodies — the ANA and the 4A’s in the US, ISBA in the UK and the WFA globally — published a joint definition of MFA, turning a vague smell into a formal category that could be measured, excluded and embarrassed.3 IAB Tech Lab added signals to the OpenRTB protocol so that an auto-refreshing ad slot can be identified inside the bid itself.4 Verification vendors began selling MFA blocklists. And when the ANA re-ran its benchmark, participating brands had cut MFA’s share of their spend from 15 per cent to about 4, and slimmed the average campaign from 44,000 sites to roughly 23,000 — a change the ANA credits in part to the press coverage the original report attracted.5 Embarrassment, it turns out, works on arbitrage.
It would be tidy to end there. But the category was defined, not defeated. The arbitrage still works wherever nobody is checking — the ANA’s Q2 2025 benchmark still counted $26.8 billion in wasted programmatic spend overall6 — and generative AI has cut the cost of manufacturing plausible filler to approximately nothing. The slideshow farm now writes itself. Keeping junk out of a media plan has become a subscription, not a one-off purchase.
The mirror test
Which brings me to the part where I examine my own shoes. This site is funded by advertising. There are four Google ads on the landing page right now, and if you arrived through it, they helped pay for the essay you are reading. So what, structurally, separates this blog from a slideshow farm? It is a fair question, and “taste” is not an answer.
The honest answer is the direction of the pretext. On an MFA site, the content exists so that ads can be served against it. Here the arrangement is inverted as deliberately as I could manage: the writing exists to be read, and the ads exist to be looked at — four of them, on the landing page only, labelled as the Exhibit, shown so that you can see what the machine currently makes of you. There are no ads on this page. Nothing refreshes itself. There is no slideshow, and no chum box waiting below this paragraph. The full accounting, including what those four ads actually earn, is in how this site makes money.
The writing exists to be read; the ads exist to be looked at.
A simple test for any ad-funded page, this one included: take the ads away and ask what is left. On a made-for-advertising site the answer is nothing, because the ads were the building and the content was the paint. Here, I hope, the answer is an essay. The day that stops being true, this site will have become the thing it writes about — and you should do exactly what you did last time: close the tab, somewhere around slide seven.