Finisterre is a British surf and outdoor brand, founded in 2003 and still run from a clifftop workshop in St Agnes, Cornwall. The first European surf brand to become B Corp certified, it makes technical and lifestyle wear for people who spend their time in and around the sea, selling online, through a growing portfolio of UK and North American stores, and through wholesale partners.

Most retail marketing teams have more customer data than ever. But getting meaningful insight from it, the kind you can actually trust and act on, is the bigger bottleneck. For most teams, getting any information beyond the static reports and dashboards served up by your platforms usually spells manual spreadsheet work, tickets to your data team, or expensive data projects.

That’s where Deep Insights Agent came in. Finisterre were among the first retailers to be given access to it, and the first useful answers arrived within a week.

What makes Deep Insights Agent different?

Any AI answer is only ever as good as the data going into it, and that’s where similar tools fall short. 

Unlike other MCP connectivity tools, which pass raw API data straight to the model, Deep Insights Agent automatically cleans, contextualizes, and enriches your data so the model works from an AI-ready foundation. 

It normalizes your source data, bakes in retail KPI definitions alongside your own, and loads in the business context that tells the model how you actually operate. That means you get answers grounded in reality, not an LLM's best guess.

Optimizing campaigns

Their first big question was a simple one: where could their abandoned browse campaign be improved?

Looking at how the flow was performing, the agent surfaced insight that would have otherwise stayed hidden: The frequency caps applied to the flow were limiting sends, which meant a large share of eligible customers weren't receiving the email at all. 

As a result, Finisterre removed the caps from both abandoned browse and abandoned basket, opening up the volume the flow had been missing.

The finding had been sitting in the data all along. What changed was that the question became easy enough to ask.

Identifying impactful new segments

Next they turned to segmentation, asking the agent how their existing segments were performing and where the untapped opportunities sat.

It came back with two high-potential groups worth testing, including customers who had browsed bags without ever buying one, and customers who had bought active outerwear but never a bag. That gives the team two new audiences to experiment and communicate with in more personalized ways.

Two days of analysis, done the same morning

The team asked which newsletters drove the most first-time orders in a given period, and which automations drove the most second purchases. Done by hand, that was a two-day job. It came back quickly enough to act on the same day.

Finisterre's CRM team can now get to key insights in minutes rather than days. They ask questions in plain language and get plain language answers back, so there's no ticket to a data analyst, no queue, and no waiting on a future project. That means they can move faster on the customer experience work that actually moves the needle.

What's next?

Finisterre has a growing list of things they want to point the agent at, including weekly trade reporting, seasonal acquisition and category insights, email performance by country, and more.

For a small team covering a lot of ground, that's a lot of new territory suddenly within reach.