In her latest column, Shirley Marschall looks at the ad tech supply chain, MFA sites, and cockroaches in the wine…

Drop a cockroach in a barrel of wine, and the whole thing is ruined. Sterilise the cockroach first and most people still won't drink it. And yet, pour fine wine onto a cockroach, and it does nothing to improve the insect.

Psychologists Paul Rozin and Edward Royzman studied this asymmetry between good and bad things, including what they called contagion: the tendency for negative properties to spread through contact.

Ad tech knows this logic well. Much of the industry's effort to assess inventory quality is essentially an exercise in tracing contamination. Which domain touched which supply path. Which SSP sold what. Which inventory showed up where. Which website meets the definition of made-for-advertising.

Trace the contamination all the way back through the supply chain, and realise that it doesn't always enter in places the industry would expect.

Adalytics' research into the ANA's Programmatic Media Supply Chain Transparency study found MFA inventory appearing through private marketplace deals, despite PMPs being positioned as more controlled, premium environments. In January 2024, ads from 16 of the 17 brands that had publicly disclosed their participation in the ANA study were observed on sites meeting relevant MFA definitions.

In this case, bad actors got into the supply chain and the damage traveled with it. But what happens when nobody put the cockroach in?

That's a question Noam Gissis, Sr. Monetisation Initiatives Lead at T-Mobile, is trying to answer. Gissis described trying to read an article on a news site when Chrome crashed. He tried Incognito. That crashed too. He switched to a privacy-first browser and the article loaded normally. 

One browser crash proves very little, as Gissis himself points out. But it does raise questions: What if some of the aggressive monetisation behaviours associated with MFA aren't necessarily spreading from MFA sites at all? What if they are emerging independently elsewhere?

A premium publisher can have legitimate traffic, original journalism, reputable advertisers and a squeaky clean supply path, and still produce a page that consumes an extraordinary amount of bandwidth or processing power, or exhibits other MFA red flags. And no, the publisher doesn't need to have anything to do with a made-for-advertising operation. Financial pressure is reason enough; original journalism costs money, audiences are fragmented, advertising revenue is under pressure. Nothing new.

Readers can’t tell the difference between an MFA or MFA-ish website. They just watch their browser struggle to load a page and wonder why. Or close the tab. Or switch browsers.

But for advertisers, the distinction matters exactly because a premium environment alone doesn't neutralise the problem. Pour fine wine onto a cockroach and it remains a cockroach... A brand-safe, premium-labelled placement on a page that takes forever to load and triggers one-too-many autoplay videos still produces the same awful reader experience as inventory nobody would knowingly buy.

So if the label on the bottle doesn't change what ends up in the glass, how did the wine go bad in the first place?

One way is contamination: a bad actor makes contact, through a compromised supply path or a deal that wasn't as clean as advertised, and the damage spreads outward.

The other is convergence: nobody makes contact at all, instead, economic pressure works on different publishers until some of them independently arrive at the same (rock bottom) outcome.

Supply path optimisation has done substantial work tackling these problems by defining MFA, identifying aggressive practices, and helping buyers decide which supply to purchase. The question it primarily answers is: Should I buy this supply?

Very useful, and yet, it's not necessarily the same question as: What does this page do to the person trying to read it? 

This is where the MFA classification apparatus reaches its limit, and curation becomes ad tech's answer du jour to the quality problem. Instead of focusing on the supply path alone, it curates inventory against signals like ad density, page weight, and behavioural patterns to catch convergence-style degradation regardless of whether a domain looks legitimate on paper. Which gets closer to the pedigree-blind question about the actual user experience.

Now Google is giving the market even more data, with the browser itself becoming an inventory-quality measurement layer. The Chrome User Experience Report (CrUX) has added four experimental ad metrics: Ad Count, Ad Density, Ad Weight by CPU and Ad Weight by Network, all based on field data from real browsing sessions. 

And that brings the ad tech industry almost full circle. Inventory quality has largely been assessed by looking backwards through the supply chain: who sold it, where it came from and how it got there. Now, Chrome is adding another vantage point, looking forward from the user's browser at what actually happened once the impression arrived.

That's progress. But the wine metaphor has one problem: there are plenty of additional ways to spoil it all.