Andrew Garson: How Claude Code Is Giving Marketers the Analytics They Used to Beg Engineering For

Last Tuesday, a marketing director at a consumer brand in Hoboken cancelled her company’s $1,200 per month Google Search Console analytics subscription. Not because she stopped caring about search data. Because she replaced the entire thing over a weekend.

She opened Claude Code, the AI coding assistant from Anthropic, plugged in two APIs she found with a quick search, DataForSEO at $0.002 per keyword lookup and SerpAPI at a penny per search, and told it what she wanted. Track 200 branded keywords across Google. Cross-reference rankings with Shopify sales data by keyword. Flag anything that drops more than five positions. Email her a summary every Monday morning.

Claude Code wrote the Python. She ran it. It worked.

Total build time: about four hours across Saturday and Sunday. Total monthly cost: roughly $40 in API calls, down from $1,200. The dashboard does more than the vendor tool did because she designed it around the exact questions she actually asks, not the questions some product manager at an analytics company decided were important three years ago.

She estimates the switch will save her team north of $15,000 this year. She is not an engineer. She has never written a line of Python in her life.

That gap, between what marketing people know they need and what they can actually get built, has defined the industry for as long as Andrew Garson has been in it. And Garson has been in it a while. Eight years at Cohn & Wolfe running integrated campaigns. Three years as SVP of Brand Marketing at WME, back when it was still called WME before the Endeavor rebrand. Two decades watching smart marketers wait in line behind product roadmaps for basic reporting tools.

“Every marketing department I’ve ever worked in had the same problem,” Garson said. “You know exactly what data you need. You know what the dashboard should look like. You know which metrics matter. And then you spend four months trying to explain that to someone who doesn’t understand your business so they can build you something that’s almost right.”

Claude Code, the AI coding assistant from Anthropic, is one of several tools blowing that bottleneck apart. Cursor, GitHub Copilot, Replit Agent. They all work roughly the same way. You describe what you want in plain English. The AI writes the code. You run it.

The difference is who gets to do the describing.

The Jira Ticket Is Dead

Here is what this looks like in practice. A marketing director wants to track how 200 branded keywords are performing across Google, compare that to competitor visibility, and flag anything that drops more than five positions week over week. Old way: vendor contract, $2,000 a month, dashboard you can’t customize. New way: describe it to Claude Code, get a Python script that runs on a schedule, outputs exactly what you asked for, costs nothing after the subscription.

Marketing teams are building CRM enrichment pipelines that score leads automatically. Automated reporting that pulls from Google Analytics, social platforms, and email tools into one document, formatted the way the CEO actually wants to see it. Competitive intelligence systems that track what rivals are publishing, what they’re ranking for, and where they’re spending.

“Three years ago this stuff cost six figures if you wanted it custom,” Garson said. “Now a marketing ops person with good instincts and Claude Code can build it in an afternoon.”

He is not exaggerating. The tools are that capable for this category of work. We are not talking about building machine learning models or running complex statistical analysis. We are talking about the 80% of marketing analytics that is really just: pull data from here, combine it with data from there, make it look like this, send it to me on Mondays.

What This Does to Agencies

Garson founded Albany CT Creative in 2018 after his agency career, and he is blunt about what this means for the model he came up in.

“A lot of agency retainers are propped up by reporting and analytics infrastructure,” he said. “When the client can build their own dashboards overnight, you better be selling something more interesting than a monthly PDF.”

That does not mean agencies disappear. It means the ones that stick around need to justify their fees with strategy, creative thinking, and category expertise. The execution layer, the part where you pay someone to pull together your numbers, is getting automated out from underneath them.

In-house teams win here. A single marketing operations manager armed with AI coding tools can replace what used to take a three-person analytics team or an agency retainer. The budget that frees up can go toward the work that actually moves the needle. Brand building. Content. Research.

Small Brands, Big Data

This matters most for companies that never had the budget for real analytics in the first place.

Procter & Gamble has entire data science departments modeling everything from demand curves to attribution. A 15-person wellness startup selling adaptogenic beverages out of Jersey City does not have that. Has never had that. Cannot afford that.

Except now the math is different. That same startup’s marketing lead can build a pipeline that connects point-of-sale data with influencer campaign metrics and social listening. Describe the logic. Claude Code writes the code. Run it.

Garson knows this space. He is a partner in Ahh Gave’ Spirits and 4GuysBeverage, both in the wellness beverage category. Brands where every marketing dollar has to justify itself.

“The playing field just got a lot more level,” he said. “You still need someone who understands the business deeply enough to know what questions to ask. But the technical barrier between the question and the answer is basically gone.”

Who Wins From Here

The marketers who benefit are not the ones who learn to code. They are the ones who always knew what to build. The ones who sat in meetings sketching dashboards on whiteboards that never got built because engineering had other priorities. The ones who understood attribution and funnel mechanics and competitive dynamics but could never get their hands on the raw data.

Those people just got very dangerous.

The ones who have been coasting on relationships and intuition without real analytical chops have fewer places to hide. When anyone can build a dashboard, the value shifts entirely to knowing which dashboard to build and what the numbers mean once you see them.

Garson, who PRWeek put on their “40 Under 40” list back during his agency days, frames it simply. The best marketing has always been a data discipline disguised as storytelling. The tools just caught up to the people who understood that all along.