Beating the dead horses

This week we’re returning to a bunch of recurring horses…er, topics…including a Colossus update, open web troubles, and what Google’s terrible, no good week tells us about the company and its future.

IAB Tech Lab’s Ads.txt and Sellers.json

For the tenth anniversary of the IAB Tech Lab we have a special series of interviews with the people responsible for the most impactful standards to come out of that organization.

The man, the legend, the Doctor, Neal Richter of Amazon joins to talk about ads.txt and Sellers.json. These specs eliminated a whole category of ad fraud in one fell swoop.

Due to the generosity of the IAB Tech Lab, these interviews will remain outside the paywall for all users.

Podcast: Joshua Palau on pharma and in-housing

Josh Palau is the poster boy for programmatic in-housing, having led groups at Bayer, LendingTree, and now Pfizer. He talks to the pod about the state of pharma programmatic and his experiences building in-house teams.

Bonus: If you’d like to help Josh fight children’s cancer, here’s a link.

Listen to the pod now:

• Spotify

• Apple

• Anywhere you listen

Beating the dead horses

Colossus update

I know, dear readers, that you crave more news about the Colossus SSP mess. We just have two quick updates.

First, in last week’s newsletter we wrote that Colossus had indicated to us that they had a relationship with LiveIntent, an email and data company. Based on that information, we included LiveIntent as a possible cause of some of the identity problems.

We got a clarification from LiveIntent indicating that 1) they did sign a license about a month ago with the parent company, Direct Digital Holdings; 2) that agreement only covers the identity graph, not any “ID Bridging” technology; 3) they are no longer supplying the company with new files.

Given that the ID problems associated with Colossus have been ongoing for quite some time and the license from LiveIntent is relatively recent, this seems to eliminate LiveIntent or another bridging solution as the primary cause of the ID problems.

Second, and in less substantial news, Colossus removed their “explainer video” from the web. This video was largely unpersuasive so this isn’t exactly surprising.

Why don’t D2C brands use The Trade Desk?

All it takes is one good Tweet to send me down a rabbit hole, and this week I have user @oliviaakory to thank.

The usual cast of ad tech twitter jumped into the comments with explanations such as the minimums being too high, or that TTD caters to agencies, or that the CPMs are too high. We should ask ourselves what the question is really asking. The real question is “can open web programmatic be effective for selling products?”

I’ve previously opined about the many ways the closed web is better for advertising than the open web. I think we can all agree that the ROAS for an DTC brand is going to better on Meta than on TTD. But what are the specific elements of open web that are going to hinder the ROAS on the open web?

Complexity: You can’t just login to a DSP and start spending. There’s a significant learning curve and you either need to invest in people or pay a pretty hefty fee to your agency to be hands-on-keyboard.

Minimums: You can buy ads on Meta/Google for $50 on your credit card. There are some self-serve DSPs, but the big ones have contracts, onboarding, and minimums. Higher minimums = higher risk.

Targeting: Navigating through inventory and third-party data sets to find what works for your brand is hugely complicated. The walled gardens just make it work.

Optimization: DSPs generally have very weak optimization systems that require a lot of tweaking and hand-holding. That’s why independent companies like Scibids and Chalice have thrived. In the walled gardens optimization is the just the way the ad serving systems work.

Clickability: The open web ecosystem has been disparaging and fighting against click-through metrics for 10+ years, saying clicks are worthless. Unfortunately, if your sales motion involves users clicking through to your website…

Attribution: The other side of clickability, is that attribution without clicks on the open web is basically voodoo. All the walled gardens have invested in post-cookie attribution systems and —whether correct or not— they give you an answer to the ROAS you achieved.

With all that said, I have to believe that if you could achieve positive ROAS on open web you would find a way to make it work. But outside of retargeting and some very high margin products, I don’t think there is. Let me know if you have counter examples.

Google and the terrible, horrible, no good, very bad week

First, a caveat. There’s nothing I’ve been wrong about more often in my life than predicting the trajectory of Google. With regard to the stock, I make Jim Cramer look like Warren Buffet. So take this whole thing with a giant grain of salt.

Since the dawn of the AI era (ChatGPT launch on 11/22/2022) the bear case on Google search has been as follows:

1. AI is a technology wave as important as the PC or the iPhone. In every previous AI wave the dominant players have been usurped (think Microsoft beating IBM, or Apple beating Nokia).

2. Google search has 90%+ market share, it is not unreasonable to believe that this share will decline with new technologies and new user experiences. If you imagine a “Star Trek communicator” user experience why would you believe that Google would have the same share in that experience as they do in search?

3. Google search ads are already overwhelmingly cluttering the search experience. One has to believe that an AI answer experience will have fewer ad opportunities than the current “blue links” paradigm.

4. AI answers cost more per query than the highly-optimized search results so even if all else was the same, Google margins would decline.

5. If another company, like Perplexity.ai, were to nail the AI future, Google would be prohibited from acquiring them due to the political environment.

6. Google’s management has not shown the ability to innovate.

Let’s focus on this last one, the management challenge.

Google management is keenly aware of the challenges they face and they have been trying to get ahead of them by calling a “code red”, by announcing many new products, and now, by inserting AI directly into the search results. That should work, right?

By now you’ve certainly seen some of the many examples of Google’s new AI search results being embarrassingly, dangerously, hysterically wrong. What to make of one of the world’s most powerful and successful companies making their flagship product fail and not seeming to really care all that much?

This is a crisis for Google, and I would be shocked if there are not major changes on the horizon. If you’re interested in some long reads, there have been a number of good ones about the management and innovation rot at the company:

• Ed Zitron’s The Man Who Killed Google Search

• Forbes on Sundar as the “peacetime CEO”

• Om Malik asks “Does Google need a new CEO?”

As a former start-up founder let me say that there’s a big difference for what someone like Larry or Sergei can accomplish at Google versus the current team. The founders can make hard choices, they understand the customers better than anyone, and they just care more about certain things than non-founders do. And as a founder, if I saw the number one product for my multi-trillion dollar company getting fucked up this bad I would be at the office with a flamethrower the next day.

Developing…

Reading list

• TrustX is now Simitri, with Akamai’s PET tech to allow identity matching in the cloud (Ari and Eric invested).

• OpenAI does a deal with News Corp worth over $250 million over three years. Can we expect migrant caravans in our AI results?

• Eric Seufert noticed that Apple is running ads on Meta for apps and app bundles that drive to their own app store (Linkedin post).

• Semafor reports that Apple News is starting to be a significant revenue driver for some publishers with Daily Beast making $3-4 million/year in subscription revenue.

• Ari’s was a guest on Eric Seufert’s excellent MobileDevMemo pod where we talked more about the challenges of the “open web.”