I'm Running a DTC Brand's Entire Marketing with Claude Code

Two months ago I took over as fractional CMO at Incred Golf. Small founding team. Passionate technical founder with patented tech. A great product priced three to four times the market median.

Sales came entirely from the founders' personal network. That was the whole marketing function.

This is the exact kind of founder I like working with. Nakul has a clear point of view. Deep understanding of his product and his category. A personal story that ties directly into the innovation. The product has enough value to justify the premium, and more.

The job was to build marketing from scratch. I've done this before for DTC brands, always with a team, over one to two years. This time I did it with my macOS terminal and Claude Code in a week.

My method has always been the same. Growth marketing is a feedback loop, and the loop lives in a dashboard. You form a hypothesis. You ship the work. You watch the numbers. You revise based on what moved. That's how I carved out multi-million dollar revenues for passionate, founder-led DTC brands before.

Step one of that method has always been the same too: the audit. Map how the brand grows, name the single biggest lever holding revenue back, rank the highest-impact work, then ship. Traditionally the audit takes a team two to four weeks and produces a deck that gets filed and forgotten. I wanted the same diagnostic quality in a day, delivered as a single working report I could execute against the next morning.

Wiring the tools

I started with the plumbing. Watched a few YouTube videos and began wiring APIs into Claude Code for every tool the audit needs:

• Shopify Admin
• Google Ads
• Meta Ads (and Meta Ad Library)
• Google Analytics 4
• Google Search Console
• Google Merchant Center
• Klaviyo
• Ahrefs
• DataForSEO
• Screaming Frog
• PageSpeed Insights
• Serper

With those connected, Claude could crawl the site, pull performance data, mine customer reviews, score competitors, and write diagnostic artefacts straight into the client's GitHub repo.

The audit

I pointed Claude at Incred Golf and asked for one consolidated report with ten high-value deliverables, written as a single Word doc plus the markdown source straight into the client repo.

1. Executive summary with cross-cutting themes. Three to four sentences naming the single biggest lever holding revenue back, the top two quick wins, and the biggest access gap. Plus three to five strategic insights that span multiple sections, produced by a dedicated synthesis pass that reads the whole draft end-to-end. The story of the business emerges here as a strategic narrative threading through every finding.

2. Key findings and top five action items at the top. Five to seven one-sentence bullets across the report, plus the top five actions to ship this week pulled to the front with a one-line "why this first" tied back to the cross-cutting themes. A ninety-second skim for the founder who wants the take-away before the evidence.

3. Audience one-pager. Fifteen verbatim quotes mined across Reddit, X, Quora, YouTube comments, and niche golf forums. Avatar, core tension, recurring language, top objections, positioning angle, belief shift. Every claim traceable to a quote in the appendix.

4. Founder vision capture. Thirty-minute conversation with Nakul. Twelve-month revenue goal, positioning hypothesis, product origin, biggest-lever belief. Transcribed verbatim where his phrasing was distinctive, because the exact words become positioning raw material.

5. Competitor teardown. Five competitors. Full catalogue crawl, top 100 ranking keywords, content-gap analysis, 30-day Meta Ad Library scan, paid-SERP intel. Whitespace map of what nobody was doing. Attribute-tuple gap analysis on the catalogue.

6. SEO opportunity map. Technical site health, keyword footprint, content gaps, backlink profile. Opportunity-pattern table: striking-distance queries, low-CTR pages, cannibalisation. Page-level fixes ranked by estimated revenue impact.

7. AEO presence audit. Answer-engine citations across ChatGPT, Claude, Perplexity, and Google AI Overview on twenty commercial queries the audience actually runs. Extractability checklist per priority page. Princeton GEO visibility-boost plays applied to the content backlog.

8. Paid ad recommendations. Creative fatigue scoring on every active ad. Competitor creative intelligence: hooks, angles, formats, offers, rotation cadence. Three starter campaigns with objective, audience, budget band, and creative brief.

9. Email, CRO, and Analytics diagnostics. Flow coverage (welcome, abandoned cart, browse, post-purchase, win-back), revenue per flow, session-recording review, rage-click heatmaps, funnel dropoff versus DTC benchmarks, event coverage, and attribution-model review. Populated where Klaviyo, Clarity, and GA4 access was shared.

10. Prioritised action list and conclusion. The single biggest lever holding revenue back right now, named at the top with evidence tagged public-data or internal-data, a quantified counter-move, and confidence level. Then every finding ICE-scored and ranked. Top five to ship this week. Next ten queued. Backlog behind them. Concluding with next steps and an optional paid-engagement path.

Delivered as one Word doc Nakul could share with his team, plus the markdown source version-controlled in our shared GitHub repo. I reviewed, edited, approved. That report became the shared source of truth with Nakul.

Letting Claude do the work

Every piece of marketing work since has pulled from the audit.

The Google Ads campaigns ran on the audience language from Section 1. The homepage rebuild ran on the positioning gap Section 3 surfaced. The Klaviyo welcome flow ran on the Voice-of-Customer mining appended to the audience section. The SEO backlog came straight from Section 10's ranked action list. Every brief, every prompt, every edit was traceable back to a finding in the audit.

I pointed Claude at each channel and told it what to do. Redesign the homepage to this brief. Build three new landing pages. Generate 50 Meta ad variations from this product page and these customer reviews. Write an email welcome sequence and add it to Klaviyo. Find 40 UK golf influencers under 50K followers and draft the outreach. Audit SEO across every page and rank fixes by revenue impact. Set up answer-engine monitoring for ChatGPT, Perplexity, and Google AI Overviews.

Claude did it. I reviewed, edited, approved, deployed.

The domain knowledge still matters. Claude does not know why a welcome sequence needs a day-7 review request, or when to use Max Conversions over Max Clicks on Google Ads. You do. You point Claude to the work and Claude executes.

The learning curve is steep and fast. A week in, I was running ten tools I'd never touched before at a level that would have taken a year to reach with a human team.

What made it compound: every screw-up, I asked Claude to remember. Every aha moment, same. Wrong Google Ads bidding strategy, logged. Winning ad hook pulled from a review, logged. A Shopify schema quirk that broke a deploy, logged. A month in, Claude had a running memory of what works in this specific business. Every new task started from that accumulated context.

Two things get better at the same time. You get sharper with the AI tools because you use them every day. You get sharper at marketing because the feedback loop is so tight that every hypothesis gets tested within the day.

The biggest change: you stop being stuck on one channel at a time. I was running SEO, paid, email, influencer outreach, content, and analytics in parallel. When I hit a gap in my own expertise, I'd ask Claude to brief me on the basics and hand the execution back. As long as you know where you want to get to, the missing pieces stop being blockers.

From one-offs to playbooks

Every time I ran a task that worked, I wrote it down. The exact prompt. The inputs it needed. The output format. The tools it called.

Those writeups became playbooks. One per function:

SEO and content

• SEO Audit
• AI SEO (AEO / GEO for ChatGPT, Perplexity, AI Overviews)
• Programmatic SEO
• Schema Markup
• Site Architecture
• Content Strategy
• Blog Post Publishing
• Copywriting and Copy Editing
• Competitor Alternatives

Paid advertising

• Paid Ads strategy
• Ad Creative generation
• Hook and Script Writing (UGC and short-form video)
• Competitor Creative Audit

Email and lifecycle

• Welcome and post-purchase sequences
• Cold email outreach (influencers, journalists, partners)
• Churn prevention

Conversion optimisation

• Page CRO
• Product Page CRO
• Signup Flow CRO
• Form CRO
• Popup CRO
• Onboarding CRO
• A/B Test Setup

Social and distribution

• Social Content calendar
• Content Repurposing (one blog post into 30 social posts)

Analytics and strategy

• Analytics Tracking setup
• Weekly Marketing Review
• CRO Review (Clarity heatmaps plus GA4)
• Pricing Strategy
• Launch Strategy
• Review Mining (Voice of Customer)
• Marketing Ideas generation

Sales and revenue

• Sales Enablement (pitch decks, one-pagers, battle cards)
• RevOps (lead scoring, marketing-to-sales handoff)
• Referral Program
• Free Tool Strategy

Around 35 playbooks. Each one a self-contained skill Claude can run on demand. The audit is the first and most important one. It seeds every downstream workstream and feeds the client's actions.md with the week-one priorities.

The shift: Claude managing GTM

The playbooks were useful. They still needed me to decide what to run and when.

So I built one more layer. A master manager playbook. Every week it reviews what every other playbook produced. What moved the needle. What didn't. It scores output against traffic, conversion, and revenue data pulled from GA4, Shopify, and Klaviyo. Then it reprioritises what runs next week and queues the work.

I used to tell Claude what to do every day. Now Claude runs the weekly review, flags what needs my approval, and executes the rest.

Far enough along that my role has changed. My time now goes to strategy, positioning, and client conversations.

Run the audit on your brand

The audit is the playbook to start with. Step 1 of the operating system. The deliverable I ship on every new engagement. The artefact any founder can run on their own brand with a Claude Code terminal and a handful of API keys.

What you get when you run it:

• One consolidated report delivered two ways: a Word doc (shareable with your team) plus the markdown source version-controlled in your GitHub repo
• The ten deliverables above, front-loaded so the executive summary, cross-cutting themes, key findings, and top five action items are all readable in the first page
• A first-pass version that runs on external tools alone (DataForSEO, Serper, Keywords Everywhere, Meta Ad Library, PageSpeed Insights). No internal logins required on day one.
• An extended version that populates the email, CRO, and analytics diagnostics once you grant access to GA4, Shopify, Klaviyo, Clarity, and your ad accounts
• A ranked action list that merges straight into your actions.md: top five to ship this week, next ten queued, backlog behind them
• The single biggest lever holding revenue back right now, named at the top of the action list with evidence and a quantified counter-move
• A conclusion block with clear next steps and an optional path to having EntireCommerce run this for you

Grab the GTM Audit playbook here. Drop it into your Claude project. Full GTM diagnostic on your desk by end of day.

If you want this built for your brand, book a call.