How should DTC brands use AI for growth strategy? Rafael Gi and Duncan Ferguson of Pilothouse use Claude to diagnose why a brand stopped growing, then make the call themselves.
Raf is head of strategic growth at Pilothouse and runs the audits behind new client pitches, five to twelve of them every two weeks at peak. Duncan leads strategy on the delivery side and builds the systems the team operates. They walk through how a nine-figure brand's growth flatlined, why one client's AOV fell from $40 to $32, a regional market maturity model that took two weeks instead of six months, and what happened when Duncan ran Pilothouse's strategy process in Claude without his own inputs.
Get the DTC Newsletter: https://directtoconsumer.co
WHAT YOU WILL SOLVE
- Your growth flatlined and nothing changed. A brand that had just hit nine figures came to Pilothouse with no idea why growth stopped. Raf found 70% of its media going into three markets. Meta kept spending where the brand already had customers, and there were no new ones left to reach there.
- You want AI to find the problem for you. Raf pulls eight to fifteen reports across platforms, then forms a hunch from experience. A frequency of 140 in one area usually means overinvestment, so he asks his knowledge base to validate it or prove him wrong.
- Your AOV dropped and you can't see why. One client's AOV fell from about $40 to $32 after it launched a $20 product. Without budgets set by product, Meta spends toward the cheaper item because it moves more volume. Raf confirms it by pulling the spend split between the two products over time.
- Your AI setup breaks between teams. If sales and delivery each define a client's "goal" differently, the agent pulls the wrong section of the wrong document. Pilothouse aligns object definitions and the naming of files, folders, agents and skills across departments, and Raf and Duncan share skill files to keep context in sync.
- Your audit is too long for anyone to act on. Pilothouse's CEO calls it Claude food and human food. Claude gets the raw data, and the strategist's job is turning a hundred-page output into a ten, five or one-page version a client can absorb and act on.
- A data project needs a data scientist and six months. Raf built a regional market maturity model for a nine-figure brand in four workshops over two weeks, doing the processing in Claude on a 178-megabyte dataset. The numbers went through the brand's CFO.
- You're wondering whether AI can run strategy on its own. Duncan turned Pilothouse's scaling growth system into a set of Claude skills, then ran it with his own inputs removed. It didn't reach the same conclusions, and it made persuasive arguments that were technically wrong.
- Your strategy dies after the handoff. Duncan calls it seagull strategy: the plan ships, hits reality, and the strategist isn't there to see the results. Pilothouse strategists stay to operate the systems, down to finding a JavaScript snippet in a Shopify backend that was distorting how AOV registered.
- Customer research costs too much. Raf uses AI for semantic mapping, grouping the words people use when they react to a product, work that used to mean an expensive focus group and people coding transcripts by hand.
ABOUT THE GUESTS
Rafael Gi is head of strategic growth at Pilothouse. Duncan Ferguson leads strategy at Pilothouse. https://pilothouse.co
STAY CONNECTED
DTC Newsletter, daily ecommerce marketing and ecommerce growth tactics: https://directtoconsumer.co
YouTube: https://youtube.com/@dtcnewsletter
LinkedIn: https://linkedin.com/company/directtoconsumer
TIMESTAMPS
02:30 How AI is transforming marketing strategy
08:00 Building connected AI workflows across teams
13:00 Using AI to identify ecommerce growth bottlenecks
20:00 Why human judgment still matters in AI strategy
28:00 The challenges of scaling AI automation
.svg)
.svg)
.svg)
.png)
.svg){kind=icon}
.svg)
.png)
.svg)
