←
Collin Slattery24,768 posts
Collin Slattery avatar

Collin Slattery

@CJSlattery

Founder at @taikundigital. Building the best performance marketing agency in the world. Sharing lessons from spending $100mm+ per year.

◷ Joined May 2010▧ 1,175 Photos and videos
Collin Slattery avatar
Collin Slattery@CJSlattery·Sep 21

I've been so AI pilled, that I've done a complete 180. AI is a money incinerator. It's a collective hallucination. Everyone sees everyone else shoveling money into the incinerator and starts doing the same. Everyone just keeps shoveling money hoping that one day the engine starts running. And maybe it will. But it isn’t. AI is expensive. Especially when you have to actually pay for tokens. And the real systemic work, the stuff that replaces people because it is autonomous, runs on tokens. The question isn’t whether your AI systems are cheaper than Americans. It’s whether they’re cheaper than people in Latam, Southeast Asia, or Africa. They aren’t. AI also suffers from broken clock syndrome. When we task AI with doing a specific task, it can often give a quality or correct output. But the problems arise when we ask an AI to produce that type of output with different inputs. If you don't know the process it used to give you the initial correct output, there's no way to know if it can take your input type and generate a quality output repeatedly or if it is just a broken clock that happened to be right. So in order to build durable results with AI, you need to build the process the AI is going to use to generate the output. You need to build in oversight. And the more varied the inputs, the more complex the output, the more granularly you need to control the functions within the black box. Otherwise, your fail rate is going to be high. And this failure rate increases whenever there is nested processing and thinking. Then there is the cost of building these systems in the first place. If you actually want something that works, you need to invest considerable resources into the build. Senior talent to develop it, token costs during calibration, ongoing maintenance when it breaks, etc. The common argument against this is to “let AI do the building on its own.” The problem with this is AI doesn’t think. It doesn’t reason. It guesses the next most likely thing in a series. So it will build something for you that will look like a house, but you go in the basement and the foundation is upside down. It’s got wires running through water pipes. The hot and cold water taps are reversed. It looks right from a distance, but any deeper level of scrutiny the illusion falls apart. Nobody is making money from the “work” that AI is doing. If they were, we would have such a rapid consolidation of wealth nobody would know what is happening. Every industry would be consolidated overnight into unassailable monopolies. I hear from lots of people who say they are making money from using AI. And my question to them is how much did it cost to build and what’s the payback period. Large companies are pulling back on AI investment. These are the companies that should be most able to realize a positive ROI on AI. So the question is whether they are uniquely incompetent at development and deployment or are smaller businesses bad at calculating cost and ROI? This results in our current environment: The only people making money in AI are the people who are selling it. It remains to be seen when using AI is profitable and how much work it can profitably replace.

Collin Slattery avatar
Collin Slattery@CJSlattery·6h

The Google passkey rollout continues to be a nightmare. New issues cropping up for team members all the time. Stuff breaking. Just an absolute shitshow.

Collin Slattery avatar
Collin Slattery@CJSlattery·8h

2019 way of testing Meta ad creatives: → creative strategy wasn’t really a concept yet, you just made ads → launch the same ad 5 times with tiny tweaks, see which one Facebook favored, dump spend into that one → no UGC, no creator seeding, no Trybe, whatever you or your designer shot that week was the entire pipeline → obsess over the 20% text rule because Facebook would reject your ad if you went over 2026 way we test creative at Taikun: → launch high creative volume every cohort, not one or two variants and a guess → require a fair test gate on spend and impressions before any ad counts in the data → track hit rate against the account’s own CPA bar, not gut feel about how the ad looks → let hook rate diverge from ROAS on purpose, since a strong hook and a strong sale are different jobs → compare cohort to cohort at the same age, never lifetime numbers against fresh launches In one of our accounts, qualifying hit rate at day 14 went from 9.2% in June to 42.2% in October on the same CPA bar and the same test gate, just more volume and more discipline behind it. The same account also had an ad that ran 65% below average hook rate and still landed 51.5% above average ROAS. Hook rate told us almost nothing about which one would sell. Creative is still the biggest lever in Meta ads. What’s different now is everywhere that lever gets pulled from, partnership content, UGC, more formats, more concepts, and the testing discipline to know which pull worked.

Collin Slattery avatar
Collin Slattery@CJSlattery·10h

One of the big challenges I've been having as we grow aggressively, is where the edges of roles are. We started the year at 14 FTE, we'll end the year at ~50 FTE. As roles get more specialized there is more and more work that could reasonably assigned to a few different people. It can be hard to determine who it should actually be assigned to. It's a sticky problem that I had not thought about when we were smaller.

Collin Slattery avatar
Collin Slattery@CJSlattery·Nov 18

Effective communication is concise. The burden is on the content creator to get their point across in the fewest words possible. Unnecessarily long output shifts the burden from creator to consumer. Base AI output is verbose. AI has taken a real human communication issue (verbosity) and made it 10x worse.

A new faux pas I’m seeing: dismissing a teammate’s well-structured, probably decent agentic output as “AI slop” because you don’t want to read it. If a teammate (or agent) has skillfully prompted and carefully edited something to solve a problem or communicate with the business, it isn’t automatic slop just because AI generated it. There’s often good stuff in there... if it’s long or you're multi-tasking, feed it to your agent. Don’t overuse the term "AI slop" for any AI output. Save it for the actual sloppity slop, ya know?

Collin Slattery avatar
Collin Slattery@CJSlattery·Nov 17

📁 Taikun Hiring Process ┃ ┣ 📁 The Mindset ┃ ┣ 📁 Hiring One Person, Not Filling a Seat ┃ ┣ 📁 Disqualify Aggressively, Not Generously ┃ ┗ 📁 Grit Decides It, Not the Resume ┃ ┣ 📁 The Questionnaire ┃ ┣ 📁 95% Rejected Right Here ┃ ┣ 📁 Hobbies, Motivation and Fit Beat Resume ┃ ┗ 📁 Low Effort or AI Slop, Disqualified ┃ ┣ 📁 Round One: The Vibe Check ┃ ┣ 📁 Testing for Culture Fit ┃ ┣ 📁 15 to 20 Minutes ┃ ┣ 📁 Picture Them on the Team ┃ ┗ 📁 No Fit, Disqualified ┃ ┣ 📁 The Skills Assessment ┃ ┣ 📁 Paid at Market Rate ┃ ┣ 📁 Harder Than Expected ┃ ┣ 📁 Tests Knowledge, Presentation, and Effort ┃ ┗ 📁 AI-Written Submission, Disqualified ┃ ┣ 📁 The Presentation ┃ ┣ 📁 Defend the Work, Live ┃ ┣ 📁 Are They BSing or Do They Know It ┃ ┣ 📁 Disagree With Me and Explain Why ┃ ┗ 📁 Misunderstands the Work, Disqualified ┃ ┗ 📁 The Final Rule ┃ ┗ 📁 If It’s Not a Hell Yes, It’s a No

Collin Slattery avatar
Collin Slattery@CJSlattery·Nov 17

Just a reminder that lots of AI tools and resources still get basics wrong. Just because the frontier models are great and getting better doesn't mean that everything is running on frontier models. So you still get things like Bing getting the date wrong by 3 days.

Post attachment
Collin Slattery avatar
Collin Slattery@CJSlattery·Nov 16

“Google Ads is so hard in Q4” CPCs are up because every brand in your category is bidding into the same Black Friday keywords at once. Smart Bidding resets into a learning phase on top of it. Neither one means the account is broken. Most of the time it comes down to one of six things, each fixable this week: 1. CPC up, conversion rate flat  We check CPA against the account’s trailing average in Data Druid before we look at CPC at all. If CPA is holding, we leave the campaign alone. A higher CPC that still converts is doing its job. 2. Impression share lost to budget  We check lost IS (budget) in Data Druid before anyone touches a bid. A campaign capping out before noon needs more daily budget, not a higher bid. A bid increase does not fix share you lost to budget. 3. Smart Bidding relearning from your own promo pricing  Data Druid flags the relearning window the moment a promo price change hits conversion value. That flag is what stops us from touching tCPA or tROAS for 3 to 5 days. Touch it mid-relearn and you restart the clock and make the volatility worse. 4. Branded traffic eating the PMax budget  PMax spends against your own brand name first because it is the cheapest conversion available. We split branded out in Data Druid and judge every PMax campaign on the nonbrand number alone. Blended ROAS looks fine while prospecting campaigns lose budget underneath it. 5. Conversions still landing, just not where you’re looking yet  We don’t call a day down until the attribution window closes. Data Druid is what tells us when that is. A sale that closes today can still attribute back to a click from 4 days ago once DDA settles it. 6. Merchant Center flagging products mid-promo  We run a feed diagnostics check every morning this month, not just when spend drops. Price changes and stock swings during Black Friday trip GTIN mismatches and price disapprovals inside the feed. A disapproved product drops out of Shopping and PMax instantly, with no warning in the campaign itself. Bookmark this one and go through it before Black Friday.

Collin Slattery avatar
Collin Slattery@CJSlattery·Nov 16

The dollar values that top talent will be able to command in every industry is going to be absolutely insane. We hire the top 5% at every position, and in the last 12 months there has been heavy upward pressure on our salaries. Many more roles in the labor market are turning into winner take all structures.

I’m starting to hear about less jobs AI amplifies capability and competence It’s creating a $250k job and a $10k job (PH for example) vs 5 $50k jobs TLDR entry level jobs are toast

Collin Slattery avatar
Collin Slattery@CJSlattery·Nov 13

Sometimes you just have to get in front of a bad story. I would like to start by saying that I always endeavor to be a good boss. I try to be observant and pay attention to my team. This scandal does not reflect who I am as a teammate, employer, and founder. I sincerely apologize to my team for my failures. I have let each and every one of them down. I can be better. I must be better. I will be better.

0:42
Collin Slattery avatar
Collin Slattery@CJSlattery·Nov 13

People in the replies said this can’t work at scale. It’s running on one of our biggest accounts right now, doing 7 figures a month with every new ad set starting at $30. Phil’s been doing this longer than most of the replies have been running ads. For a founder, this decides how much of your new creative gets a fair test. Small budgets mean more of it gets tested before you kill it. Less of your money goes on ads that were never going to work. It’s how we test across the agency. This is what it looks like from the inside: → Every new ad starts small → A buyer checks the rules every morning → Winners get stepped up → Losers come down → Nobody guesses

Hot take: you don’t need 3x your CPA as a daily budget to test. Ad set budget campaign 30% of your target CPA per day If it converts, it goes up. If it doesn’t, it comes down. Small budgets find winners. 7 figures per month.

Collin Slattery avatar
Collin Slattery@CJSlattery·Nov 12

We’re actively hiring right now. The trait that decides who gets an offer has very little, if nothing, to do with the resume sitting in front of me. Grit decides it. Are you hungry to become better? Watch how someone responds when a take home assignment is harder than expected, when the brief suddenly changes halfway through, when the answer requires three more hours they did not budget for. Great hires are willing to get to the bottom of something, put in the extra three hours and turn in a finished answer, while weaker ones turn in something half finished just to make the deadline. They’re constantly digging. I’ve had great hires dig into a client account that already closed, just to understand why a campaign underperformed or superseded a client’s expectations. This is the single strongest predictor of who still cares about the work three years in. Talent sets the ceiling and effort closes the distance between where someone starts and how high that ceiling sits. Most people stop climbing long before they get anywhere near the top, usually right around the point where the work stops being fun. Taikun is hiring across several roles right now. If grit is something you have demonstrated rather than something you would just claim in an interview, head over to the company LI page (Taikun Digital) and apply.

Collin Slattery avatar
Collin Slattery@CJSlattery·Nov 11

There are roughly 30 billion possible combinations of settings inside a Google Ads account. About 1% of those combinations are optimal for any given brand, another 15% are good, 10% are fine, and the rest will waste budget. So there’s a tiny point among of all these combinations where you’re set up optimally for success. In that grid of possibilities, the circle of optimal structure moves based on the brand. A campaign structure that’s optimal for a $50M supplement brand running 70% gross margin can wreck a $3M skincare brand running half that. This is why cookie-cutter playbooks constantly fail. If you’re running your BFCM on a cookie-cutter playbook, expect this outcome: Volume hits, the account scales past what the structure was built to handle, and CPAs climb for four days straight while you’re trying to figure out why. A structure built around one brand’s economics gets applied to a different account with a different margin profile. The result is a ROAS drop, because the structure was built for the first brand’s economics, not the second’s. Meta runs on the same principle, just with fewer variables to navigate. The optimal setup still shifts by brand, but it takes fewer moves to find it. We manage $500M+ in spend across hundreds of accounts, and no two of them run an identical structure. Every account has its own version of that 1%. If your media buyer can’t tell you why your current structure fits your specific unit economics, you’re sitting in the 10% that’s fine, not the 1% that’s optimal. Audit your account against your own numbers, not someone else’s framework.

Collin Slattery avatar
Collin Slattery@CJSlattery·Nov 11

Software has never been more alive. The value from AI will accrue to software firms. AI knows how to write code. Software engineers know how to scope it and get the most out of it. Maintenance, management, security, etc are annoying. The amazing part is our software costs will go down. A $1000/mo SaaS will become $30/mo instead.

Software is dead. We’re a bunch of portrait painters and the camera just came out. Sure, there’s some residual value in zombie subscriptions and old people who don’t bother to switch. Or corporations who pay a fortune for things that should be free and bizarre main frame software that nobody has an incentive to update. But RIP 98% of the software industry. Fable 5.2 can built Microsoft Excel in ~50 hours (source: @MatthewBerman ). At the current rate of improvement: 2027: 7 hours 2028: 57 min 2029: 8 min 2030: 1 min 2031: 8 seconds ChatGPT just launched generative interfaces for chat results. Ask a question and it finds the most elegant way to display it (a map, data, etc). Within 3-5 years it will generate the perfect interface + software for your task in less time that it takes to microwave a burrito. People always love to answer this with: “There’s still record shops!” The difference is, the guy who used to own big record store owned a private jet. Now he drives a beat up Ford Taurus.

Collin Slattery avatar
Collin Slattery@CJSlattery·Nov 11

FINAL REMINDER: Go check your Meta credit limit NOW. And if you're not sure if it's enough, we built a free calculator you can use: lnkd.in/eaVNQvFH

0:45