Agents and ad blockers on the Deeper Diligence podcast
previously: there are many paradoxes but this one is mine
I was on the Deeper Diligence podcast by Daypart, with Bogdan Patynski and Jeromy Sonne. (Deeper Diligence podcast RSS feed, this episode on YouTube. Available “where you get your podcasts.”)
A big issue that we covered is the relationship between an area that marketers are super fired up about discussing—“AI” shopping agents—and an area that sometimes people want to pretend doesn’t exist—advertising’s security problems.
Yes, they are related. Early adopter customers are disappearing in some ways and appearing extra times in others.
The ad blocking part is not surprising, at least from outside the marketing filter bubble. The FBI has been warning about crime in the search ads since the 2022 Christmas shopping season, and instead of cutting back on the problem, Google has actually made it worse. The 2023 Google Ads Trademark Policy ended “certain steps to restrict unauthorized trademark use proactively,” and now requires the trademark holder to report each infringing ad.
Since, under the new policy, brand owners will have to report violations each time they learn of one, it will be incumbent on companies to monitor sponsored advertising in Google search results and to act quickly to minimize the resulting harm.
But, as Jeromy pointed out on the podcast, “It feels like a little bit of a catch-22 there.” Google ad AI “learns” to target people who are likely to click, not people who are likely to report. On one hand, Google says, “You have to spot that ad in order to report it,” and on the other hand “We make it so you don’t spot that ad.” And what’s even worse is that the ads are progressively redesigned to be easier and easier to click by mistake. Kaiser Permanente employees who clicked on ads got a fake version of the employee portal site. So it’s not just a matter of “if you’re seeing ads you have a security problem,” it’s more like “if anyone in your organization is seeing ads you have a security problem.”
The advice in CSO Online isn’t for corporate IT to make an ad blocker available, it’s
Push a vetted ad-blocking extension to managed browsers through policy rather than leaving it to users to install it.
Advertisers trying to reach tech early adopters and security-sensitive audiences have been dealing with this problem, more or less well, for quite a while. We talked about the WiseGuyGaming ad blocking study. Gamers Love Ad Blockers. So This Site Found Another Way To Show Them Ads
But ad blocking is increasingly a thing for mainstream business audiences. I suggest on the podcast that “a lot of marketing decision-makers could really benefit by taking an IT manager to lunch.”
Yes, the subject of ad blocking as security came up at Texas Linux Fest.
Agents as extra traffic
As trackable customers are fewer, the number of trackable AI agents will go up. Agents will “learn” to try various profiles, to try to get classified as someone who gets a better deal. So the optimistic view is that agents will lower the ROI on surveillance pricing and other forms of adversarial personalization, and drive investment into other areas such as quality and service. As Patrick Collison put it, they’ll be a structural subsidy for product quality.
Mark Ritson says that AI agents don’t need their own marketing strategy, because the good agents will learn to copy customers.
AI is replicating human decision-makers faster and more faithfully than the systems being sold to you that target the machines.
The agents won’t copy the average customer, though. People will choose agents to get more of the behavior of a high-information shopper, or a bargain hunter. I don’t want an agent that shops for food as me, I want one that shops like [family member redacted] who would split the list across three different supermarkets and remember exact prices. So brands will be better off not doing a separate agent project, but instead doing a web a11y project and a knowledge base project—both of which benefit human shoppers, too.
Anyway, check out the whole podcast from the links above or on the Daypart home page.
Bonus links
Attackers
Used Optimizely to A/B Test Malware on Top Websites by Brian Krebs.
Websites can minimize the chances that a breach at a third-party
content provider spirals into a supply chain attack by taking full
advantage of
Content Security Policies
or CSPs, which specify the
domain names that are allowed to serve third-party content. But Edwards
said most websites — apart from perhaps the largest e-commerce, social
media and financial platforms — don’t even bother with CSPs, or else
tend to leave them overly permissive for marketing reasons.
Ad
Fraud Is Stealing Your Budget & Your Dashboard Calls It
Performance The advertiser pays for activity that will never turn
into a customer, and in doing so helps fund one of the world’s largest
criminal industries.
Why Every Website Should Have an RSS Feed by Ronalds Vilciņš. (fwiw, Google also uses RSS feeds as a sitemap, to get updated pages to crawl. I have two RSS feeds on here, the people feed (new content) and the bot feed (all changes).
DON’T REGRET WHAT YOU DIDN’T DO by Dave Trott. How one agency pitched a product name and brand assets to a new client.
Instagram’s Algorithm Is Promoting Antisemitic Content, Report Finds by Maggie Harrison Dupré. (YOUR AD HERE?)
Court
denies motion to dismiss DRAM price fixing case by LitNews Newsroom.
At the hearing, the companies quoted the late Judge Posner’s line
that competitors in concentrated markets
watch each other like
hawks.
Judge Wise wrote: But Plaintiffs’ allegations go beyond
hawks watching each other; their contentions are more akin to hawks
flying in formation – an activity that hawks ordinarily do not
do.