Practice

Analytics

Reading the numbers without lying to yourself about what they mean.

• The Forty Percent

  Nearly 40% of GA4 properties have misconfigured events compromising their data. Everyone is layering AI analytics on top of GA4 right now. Gemini Insights. Looker summaries. LLM anomaly detection. All of it reads the same broken foundation. AI on corrupted tracking doesn't give you bad answers. It gives you confident bad answers. The most common error: enhanced measurement and GTM both tracking the same events. One checkbox, one tag, every pageview counted twice. GA4's setup assistant shows

• The Measure Trap

  Google's PageRank was supposed to measure website quality. Instead, it created a $80 billion SEO industry. Larry Page designed PageRank in 1998 to count links like academic citations. More links meant better content. Simple. Within three years, people were buying links, farming links, trading links. The metric became the game. Thing is... We don't measure what matters. We matter what we measure. Klout scored social influence until people started tweeting for points. Hospitals tracked hand-w

• Data Driven vs Data Inspired [Part 5 of 5]

  Breaking Free Your team can escape. But not how you think. The answer isn't better data. It's better questions. Data driven asks, "What should we do?" Data inspired asks, "What's happening here?" That second question changes everything. Here's what data-inspired questions look like. When team members fight, don't ask "How do we stop this?" Instead, ask "What structure creates this conflict?" When campaigns fail, don't ask "What worked before?" Ask "What changed in our audience's world?"

• Data Driven vs Data Inspired [Part 4 of 5]

  Why Smart Teams Stay Stuck Smart teams choose data-driven thinking. Even when they know better. It's not because they're dumb. It's because of how we reward teams. Here's what I mean. Two team members keep fighting about everything. Creative direction. Deadlines. Who gets the good projects. The team lead steps in, talks it out, sets boundaries, and solves the problem. For a week. Then they fight again. The same pattern, the same solution, and the same temporary peace. The team lead gets

• Data Driven vs Data Inspired [Part 3 of 5]

  Being Data Inspired Data inspired is different. Let me show you. Data driven sees: "Sales dropped last month. Run Campaign X again." Data inspired sees: "Sales drop every time we launch without testing messaging first." Notice the difference? Data-driven copy solutions. Data-inspired find principles. You look at the same numbers but ask different questions. Not "what did we do before?" But "what is this telling us?" Not "how do we fix this?" But "why does this keep happening?" Here's t

SEO

Earning organic visibility, not gaming it.

• Three Parts or Nothing

  Here's what AI Overviews actually do with your page: scan for entity, attribute, value, then move on if they don't find one. Most SEO content is written to persuade, not be extracted. An AI extraction layer needs a named subject, a specific property, and a definitive value before it can cite you. "May improve engagement" doesn't qualify. An EAV triple: "Google's AI Mode reduces organic click-through rates on navigational queries." Named entity, specific attribute, definitive value. That's cita

• The Citation Flywheel

  Most businesses will lose 18% to 64% of organic traffic in 2026. That's Kaushik's projection. The range depends on your query mix. Informational sites are at the high end. Answer Engines don't send traffic to your site. They synthesize the answer and deliver it. One or two sources get the click. Most get nothing. And as AEs go fully agentic, even that disappears. What most people miss: losing the click isn't the same as losing the customer. The businesses that survive AEO aren't capturing m

• Someone Else's Map

  Your topical map is probably someone else's. Most people build them the same way: open Google, find competitors, copy the gaps. It feels like strategy. It's actually copying. You can start from rankings or from reality. One gives you competitors' map. The other gives you yours. SERP-first asks: what are they doing that I'm not? Topic-first asks: what does full coverage require? The first catches up. The second pulls ahead. Most topical maps are built to catch up. That pull-ahead dynamic i

• Clarity Beats Position

  Source: Margarita Loktionova, Semrush Blog — https://www.semrush.com/blog/ai-tools-the-modern-buyer-journey-study/ Half of U.S. consumers who have used AI tools have already bought something after using AI to research it. Not "considered buying." Not "added to cart." Bought. That number comes from a Semrush survey of 1,030 Americans conducted in December 2025. Anyone still treating AI visibility as a future problem is already behind. I found the study while scanning SEO research feeds, and on

• The Citation Game

  The teams running the most sophisticated SEO programs in their category are the most exposed. They did everything right: cut brand spend, scaled informational content, measured everything. Their model was correct for a world that's ending. These teams built content machines. Hundreds of informational articles targeting high-volume queries. Strong domain authority. Consistent publishing cadence. They outcompeted on the metrics that mattered. The metrics that mattered were clicks, rankings, and

Paid Media

Spend that survives contact with an actual budget review.

• Split Signal Problem

  Stacking Smart Bidding, PMax, and AI Max on the same audience doesn't multiply learning. It splits it. Each Google algorithm learns from conversion signals. Run three overlapping campaigns and each gets a third of the data it needs. The same users, the same clicks, the same purchases are being attributed across all three. Google's threshold for exiting learning mode is roughly 30-50 conversions per campaign per month. Divide a 60-conversion account three ways and none of them ever graduate.

AI

Where the tools actually change the work, and where they don't.

• The Nine Percent

  Only 9% of organizations can intervene before an AI agent completes a harmful action. 57% have agents in production. 89% have observability tooling. Fewer than one in ten can actually stop one mid-task. Observability tells you what happened. Evals tell you if it should have happened. Neither tells you how to stop it mid-flight. If you've assumed your observability stack gives you control over your agents, this is your correction. The friction that keeps showing up: teams treat "kill the proc

• Rules Eat Context

  Somewhere in your project there's a plain text file that your AI reads before it reads anything else. Before your code. Before your conversation. Before the error message you're asking it to fix. A small document that sits at the mouth of a finite pipe and decides how much of everything else can get through. A team's version hit 600 lines. Their agent started hallucinating function names from three prompts ago. They deleted half the file. The hallucinations stopped. Not because the rules wer

• Why We Need Receipts

  Remember when you could open a car hood and point to what was broken? Now it's all computers. And we're doing the same thing to thinking. We want to believe bigger AI models are smarter. More context means better answers. Like assuming the kid with the biggest backpack has the best grades. But here's what actually happens. You feed AI your entire company knowledge base. It gives you an answer. You ask "why?" It can't tell you. Not really. Just processed... everything. Thing is, humans d

• The Cognitive Loan

  Every time you ask an AI to write for you, you're taking out a cognitive loan. The interest compounds silently. Here's what's happening: When we outsource thinking to LLMs, we're not just saving time. We're skipping the neural pathways that encode lasting memory. Think about it. You write a draft with AI help. The words look perfect. But a week later? The ideas feel distant, borrowed, like reading someone else's notes. This isn't about AI being good or bad. It's about cognitive debt. Ever

• Part 3 of 3: Expensive Cheap

  Amazon spent billions building robots for their warehouses. These robots can lift heavy things. They can move fast. They can work all day without getting tired. But Amazon discovered something interesting. The robots get confused by simple things. A crumpled shipping box stops them completely. They cannot find one item hidden behind another item. Tasks that any person could do without thinking. A billion-dollar robot defeated by wrinkled cardboard. This tells us something important about th

Coding

Building the pipelines and tools the strategy runs on.

First coding note is in progress.