From fragmented data to customer intelligence: Making marketing more human

Marketers have access to more customer information than ever, yet many still struggle to turn that information into something useful. Transactions live in one system, website behavior in another and customer feedback somewhere else entirely. The challenge is no longer simply gathering data. It is connecting the signals, making them accessible and helping people use them to make better decisions.

In this episode of the Marketer's Alchemy podcast, host Kathryn Turnoff sits down with the Chief Digital Officer at POLYWOOD to discuss how his team is bringing technology, marketing, customer service and creative together around a more complete understanding of the customer.

What’s inside this episode

For Benjamin, effective digital transformation is not simply a technology initiative. The infrastructure matters, but its real value emerges when employees can apply it to their work.

That philosophy shaped POLYWOOD’s effort to connect more than a decade of e-commerce, drop-ship, transaction and customer service information. By combining data from sources such as Google, Facebook, Klaviyo and Shopify in a unified data environment, the company could begin connecting the person who visited the website with the person who clicked an email, placed an order or later contacted customer service. The goal was not merely a larger database. It was a clearer, more actionable view of the customer.

Keeping the detail that makes data useful

When organizations bring information together, there can be a temptation to simplify it immediately into weekly averages, summary dashboards and high-level trends. Benjamin takes a different approach.

His team preserves the detailed, raw information that can serve as a leading indicator. For a company with more than 15,000 product SKUs across colors and variations, that granularity can reveal how preferences differ by region, season, product type or cushion choice. The richness, he explained, comes from detailed data rather than averages.

That same thinking extends to customer ratings and reviews. POLYWOOD initially used text analytics and natural language processing to identify recurring feedback, such as requests for different dimensions, thicker pillows or other product changes. The team later began using large language models to organize customers into persona types and explore what particular groups liked, disliked or searched for. Those insights could then help product and design teams evaluate potential colors, features and product improvements.

But the first version of the solution illustrated an important lesson: Access does not guarantee adoption.

The team initially created numerous dashboards that required users to navigate filters and sliders. Those dashboards did not fit naturally into the designers’ work. Adoption increased when the experience became conversational, allowing employees to ask direct questions of the data and quickly investigate an idea without submitting a request to an analyst.

Using AI to enhance people, not replace them

Benjamin’s approach to AI is grounded in augmentation. He is not looking to ask a model to independently design a new piece of furniture. Instead, he wants technology to help designers and creative teams explore possibilities, visualize concepts and test assumptions more easily.

That could mean giving designers the ability to see how a chair might look beside a lake or how a pergola could fit into a particular outdoor environment. Technology accelerates visualization, but people continue to supply creativity, judgment and brand understanding.

This distinction is critical for marketing organizations evaluating AI. The most valuable application may not be replacing work. It may be removing the technical and operational friction that prevents talented people from doing their best work.

Finding useful signals in open data

As privacy expectations have evolved and more customer information has moved into walled environments, marketers have had to reconsider where valuable signals can be found. Benjamin sees significant potential in publicly available information, particularly open government data.

One example is historical weather data. Because interest in outdoor furniture is closely connected to when people begin spending time outside, POLYWOOD examined years of hourly weather information alongside order data.

The initial hypothesis did not hold. The weather at the exact moment a customer completed checkout did not show a meaningful correlation. When the team expanded the window to examine the preceding seven days, however, a stronger pattern emerged. Several sunny days could prompt people to begin thinking about their outdoor spaces. The pattern also varied geographically, reinforcing the need to account for regional differences rather than relying on a single seasonal calendar.

POLYWOOD now uses what it learned from historical patterns alongside current weather information to inform when customer communication should begin. The insight affects not only digital advertising, but also catalogs, content timing and seasonal messaging.

Moving from demographic profiles to moments of intent

Traditional customer personas can describe someone’s interests, attitudes or aspirations. But for POLYWOOD, a more practical question is whether that person has an outdoor space to furnish.

Benjamin describes this as a shift from individual personas to “home personas.” Square footage, outdoor space, pools, patios and moving activity can provide a more relevant picture of whether someone may need outdoor furniture. These signals help the brand consider both the size of the opportunity and the type of inspiration that may be useful to a particular household.

Timing matters just as much as eligibility. A new homeowner may be a strong potential customer, but the first days after a move are crowded with messages from service providers and other businesses. POLYWOOD uses its historical customer data to understand when buyers with certain types of homes typically turn their attention from the interior to the exterior.

Weather is then layered onto the moving timeline. Someone moving into a home in New Hampshire during the winter is likely to behave differently from someone entering a comparable home in a warmer climate. Although the team considers additional micro-signals, Benjamin identified weather and moving timelines as the strongest indicators in its current approach.

The result is marketing built around relevance rather than volume: reaching the right household with useful inspiration at a moment when the need is more likely to exist.

Building capabilities around the people who use them

POLYWOOD has also brought its creative and marketing organization in-house. Benjamin pointed first to privacy, noting that the company handles significant customer information because it ships directly to buyers and does not want those customer files leaving the organization unnecessarily.

The in-house model also keeps teams close to the brand and product. Employees have assembled, used and even participated in the manufacturing of the furniture, creating a deeper understanding of what they are communicating to customers.

Just as importantly, the creative team can help shape the technology being built for it. Rather than receiving finished outputs from an external partner, POLYWOOD’s internal teams are developing capabilities intended to support how their creative professionals actually work.

Turning relevance into stronger customer relationships

A more connected view of the customer also allows POLYWOOD to communicate with existing buyers more thoughtfully. Instead of sending a general sale message, the company can recommend a piece that complements a customer’s current furniture, such as a side table or chaise lounge in a compatible collection or color.

Benjamin shared that POLYWOOD’s repeat purchase rate increased from 1.5 to 1.8. Because its products carry a high average order value, he described that change as meaningful for the business. Early indicators included site traffic and basket activity, particularly when customers added items, refined their selections and returned later as they considered a purchase with family members.

The example highlights a broader principle: Good personalization is not simply inserting a name into a message. It is using what the brand already knows to make the next interaction more useful.

From more data to better questions

When asked which source marketers most often overlook, Benjamin’s answer was open data. Government, state and city portals can contain information ranging from construction and tree-planting permits to pool permits. For marketers willing to explore it, that information can reveal timely and relevant indicators of customer needs.

He also rejected the idea that marketers must choose between AI and first-party data. The value comes from using AI to understand the data an organization already possesses. Raw information might remain inaccessible to most employees, but conversational tools can make it easier to investigate relationships, identify outliers and discover patterns.

The takeaway is not that every organization needs more dashboards, more models or more data. It needs connected information, accessible tools and people empowered to ask better questions.

That is where customer intelligence becomes customer understanding, and where technology helps make marketing more relevant, timely and human.

Subscribe to Marketer’s Alchemy wherever you listen to podcasts: Apple Podcasts, Spotify, Audible, and YouTube.