Kirk Williams

Google: "Our PMax AI is smarter than you could possibly imagine." PMax performance if you change your ROAS target from a 450% to a 445%:

@Franklin_Graham @realDonaldTrump

I had received some questions about President @realDonaldTrump's recent posts, and here are my thoughts: I do not believe President Trump would knowingly depict himself as Jesus Christ—that would certainly be inappropriate. I’m thankful the President has made it very clear that this was not at all what he thought the AI-generated image was representing—he thought it was a doctor helping someone, and when he learned of the concerns, he immediately removed the post. When I looked at the illustration, I didn’t jump to the same conclusion as some. There were no spiritual references—no halo, there were no crosses, no angels. It was a flag, soldiers, a nurse, fighter planes, eagles, the Statue of Liberty, and I think this is a lot to do about nothing. There is so much ill-intended speculation. I think his enemies are always foaming at the mouth at any possible opportunity to make him look bad. And the illustration from someone else he reposted on Truth Social yesterday, I must say that I like the fact that this is a picture of Jesus whispering in his ear, or at least His hand on his shoulder, guiding him. We all need that—we all need to be listening to Jesus. Again, I think there is an attempt to spin this into something that it isn’t. Remember, President Trump didn’t draw this, he didn’t create it, he reposted it on his social media because he thought it was nice—I would have to agree. I’m not a Catholic, I’m an evangelical, but I appreciate how President Trump has defended religious freedom for people of all faiths, including millions of evangelicals and Catholics in the U.S. and around the world. He is the most pro-Christian, pro-life president in my lifetime, and he doesn’t shy away from it. I would hope that the President and Pope Leo can meet at some point, and that the Pope would have the opportunity to thank the President for his efforts to protect religious liberty for Catholics and people of all faiths.

@tomfgoodwin @purnavirji 👀

I would pay so much money to sit down with the head of LinkedIn and just show them my feed, and ask them to explain what I see An infinite numher of monkeys on infinite typewriters, would produce largely more valuable content.

@oliviaakory Hmmmm, I think it’s foolish not to identify the role macro conditions have on performance if they do play a role… but then as equally foolish to not begin immediate plans to solve for those macro conditions. Perhaps that’s what you mean by “track but don’t blame”?

Unwritten rule for Heads of Growth: Don’t blame performance on “macro conditions” to your founders. Track macro but don’t blame macro. They don’t want to hear it.

@KurtHenninger yes, agreed. It definitely helps. My main point in this article, is to help communicate that: you can't just drop a prompt "tell me how my Google Ads account can be improved" and expect it to be excellent without a necessary contextualized, communicated, systematic approach.

@PPCKirk What are you using for Claude? Claude Code, CoWork, or Claude web? Definitely helps a lot when you give any AI more context.

Using Claude to analyze your Google Ads account (let's call that a Claudit®) has two crucial issues you need to ponder and solve before just running AI loose in your Google Ads account, read those 2 concerns here: zatomarketing.com/blog/stop-runn…

@daniellgoldstei @TaylorHoliday then it sound like "agencies" isn't the problem here, but "bad agencies" which wasn't reflected in your "all agencies" OP 🤷🏻‍♂️

@TaylorHoliday Transparently I’m a CTC client and this is not about you. You do it well & my team there is great. But there are other agencies delivering absolute crap output as they move towards trying to expand their capabilities versus leaning into better executing their domain.

Hot take as a CMO 🚨 I want my team to be using AI as much as possible and my agencies to be using it as little as possible. I hire agencies for their ability to execute and their experience. I want them to tap into that. I want my internal team to compete with them using AI

@Seanfrank Congrats Sean, welcome to the dad club!! It’s pretty awesome 😎

New character alert. Happy, healthy, ready to start a supplement brand.

@philvischer please tell me you’ve seen this Dark Rock version of the vegetables theme song… facebook.com/share/v/17Lz2q…

@oliviaakory Mine is old so that means I’m more of a “classic car” guy than annoying… right??? 😂

@PPCKirk you don’t

G Wagons. Do they create annoying drivers or just attract annoying drivers?

@TaylorHoliday This seems like it's aiming everyone at the wrong objective. Token utilization isn't the goal, the quality of what they're building is.

What percent of salary should token utilization be per employee?

@TaylorHoliday @codyplof I think what you’re saying is possible in an ideal world (and vacuum… if one could theoretically identify and separate out emotions from all other behaviors/effects), but I do not think it is possible in *our* world 😁

@TaylorHoliday @codyplof But if you can’t actually measure them (only their effects), and if you’re measuring 1000 other data points that are easy(ier) to measure, how will you ever actually identify what is a direct effect from (hidden) emotions and not just improperly correlate all the time?

Rare non Claude tweet. Say you’re running upper funnel and you do an awareness lift or consideration lift test. What do you consider a good outcome? How do you know what is a good cost per lifted awareness or consideration outcome?

@TaylorHoliday @codyplof How do you define internal thoughts? And emotions? Those both play into purchase decisions but, how do we define them? I’m not convinced you can assign a numerical value to things like this, but I could be wrong?

@codyplof So what’s the value? If it exists you should be able to define it. And assign a numerical value to it.

In ~75 days, every agency running AI-generated video ads targeting New York is going to have a very expensive problem they don't know about yet. On Dec 11, 2025, Gov. Hochul signed S.8420-A. This is the first state law requiring advertisers to "conspicuously disclose" when an ad features a synthetic performer, defined as any AI-generated or algorithmically created figure designed to appear as a real human. This is a wildly expansive definition - it doesn't have to be a deepfake of a celebrity or digital replica of an existing person - but ANY AI-generated human figure in any visual medium (audio-only is excluded). The law takes effect June 9, 2026. The penalty structure is eye-watering for digital-heavy brands. 1st violation: $1,000. Each subsequent violation: $5,000. Each non-compliant ad is a separate violation. There is no prescribed cure period. There is no notice-and-correct safe harbor. There is no requirement that the state notify you before assessing penalties. The language is clear: every violation of the statute "shall result in" the above civil penalty. If your agency is running 50 active video ads with AI-generated talent across Meta, YouTube & CTV without disclosure, that's not a $5,000 problem. It's $246k problem. Scale to 200-300 creatives and the exposure looks less like an annoyance and more like a fleet of brand-new Porsche 911 Turbos. That's eye-opening on its own BEFORE you factor in the state's situation: NYS is staring down a cumulative 3-year budget gap of $34.3B. Federal Medicaid cuts are threatening to make it worse. The state has every incentive to treat enforcement as revenue-positive rather than a cost center. Discovery is trivially easy, even for bureaucrats: pull the public ad library, identify AI performers, check for disclosure, send the notice + invoice. The tools you used to generate the creative are the tools that prove the violation. There's no "we didn't know" defense when the prompt and the invoice both specified what was happening. And the best part for NY? Most of these companies are out-of-state - meaning relatively few upset voters. Most agencies have no idea this law exists. They're not advising clients, have zero disclosure plans in place and they're certainly not auditing existing creative libraries to ensure winners can be updated in time. They're doing what they always do: running ads until someone tells them to stop. By the time someone does, the bill could exceed a starter home in most states. The compliance obligation falls on whoever "produces or creates" the advertisement. That's the agency, production company, and/or brand. The law does include exemptions for audio-only ads, as well as cases where AI is used exclusively for translation purposes and "expressive" works (i.e. film, art) where the content of the advertisement is consistent with the content of the works. One potential reprieve: the same day Hochul signed this, the White House issued an Executive Order aimed at preempting state AI regulation, including a DOJ Task Force to challenge state laws. But an EO cannot overturn a signed statute - only Congress or the courts can. Anyone betting on federal preemption materializing before June 9 is making a bet I wouldn't take. If you need further proof, just look at TSA lines. The agencies paying attention will build disclosure into their creative process now. The ones that aren't will find out about this law the way their clients find out about most things their agency should have told them: after it costs them money. The clock is ticking.

@jamesonhaslam I don’t know, but I can see a world where no coordination happens other than the fact that someone sees a specific piece of content starting to gain traction, and so they quickly write their own to try to jump on the topic interest of the day.

I have a question about Twitter When you see 5+ voices (gurley, a16z etc) come out on the same day saying the same thing (paid customer acquisition is for losers?) is someone paying them to say it? Was there some event I missed? Help me understand

@PPCGreg This actually used to be a lot more common years and years ago by the bigger agencies. I think it was for their own reporting stuff, but ewww privacy concerns imo

why would an agency pass their client's customer id through with UTMs? Is there something you could do with that?

@TaylorHoliday incredible

This was 100% AI. Wild times…

@jphebdon @andrewjfaris Even the AI Super Bowl ads were incredibly misaligned. They were “joking” about the very thing everyone is terrified about, like dying randomly by Alexa or Ring “watching your neighborhood for lost dogs”. Was so interesting to see how much those brands hurt themselves

@PPCKirk @andrewjfaris I think it's hard for people to get too excited about AI when the news is filled with reports of super-profitable companies like Block, Microsoft & Amazon high fiving each other, laying off double digit percentages of their work force, citing AI efficiency gains as the reason.

This is a big part of why I am increasingly concerned about Gen AI for creative as a strategy. It's not that you can't use AI in the process at all; it's that if customers hate AI, they may be deeply turned off by lots of your uses of it in ads.