Personalization Isn't The Goal. A Better Customer Experience Is.

I recently moderated a panel at the Seattle eCommerce Summit on marketing personalization at scale. Between AI, customer data, and the growing number of ways brands can reach people, there is no shortage of technology promising a more personalized experience. The harder question is deciding which of those opportunities actually makes the customer experience better.

We had four very different perspectives on stage, from consumer subscriptions and outdoor apparel to industrial equipment and video analytics. That made for an interesting discussion, because what works for one brand would not necessarily make sense for another. We talked about what teams are putting into practice today, where AI is genuinely useful, and where more personalization can become too much.

Our panelists were Zach Solomon, Director of eCommerce & Digital at Filson; Ryan Fritzky, Co-Founder & Chief Marketer at Bean Box; Jackie Atkin, Director of Global Product Marketing, Brand and Creative Studio at Fluke Corporation; and Nik Makarov, Head of Partnerships at Hyperfocus.tech.

Knowing Who Not to Target

Zach Solomon at Filson brought up a use of personalization that is easy to overlook: knowing who to leave out of a campaign. Filson sells outdoor apparel, much of it built for colder conditions, so the weather matters. The team uses AI to help create weather-based audience segments and exclude customers from campaigns that are unlikely to be relevant. Living in Texas, I can appreciate not getting a wool jacket promotion in the middle of a heat wave.

The same approach applies to purchase history and engagement. Someone who just bought a jacket doesn’t need another promotion for that jacket. And if a customer has received several messages without engaging, sending more of the same probably will not help. Zach’s point was that protecting the customer relationship sometimes means sending less, not finding another reason to send.

Nik Makarov raised a related concern from his work analyzing video content and audience response. Brands can see a trend gaining traction and feel pressure to replicate it, even when the content is a poor fit for their audience or identity. The same goes for aggressive personalization. Knowing something about a customer does not mean the brand needs to bring it up. Relevance has to include good judgment.

What Customer Cancellations Can Tell You

Ryan Fritzky framed personalization differently for Bean Box, where it is closely tied to the product itself. When someone subscribes to coffee, they want something they will enjoy, delivered at a pace that works for them. Get the selection or quantity wrong often enough, and there is not much reason to keep subscribing.

Ryan explained how Bean Box uses AI to review cancellation feedback alongside subscription changes and customer preferences. Looking at those signals together helps the team understand what may have gone wrong and identify patterns across customers. The point is not to produce a more elaborate report on churn. It is to find something the team can improve before the next customer has the same experience.

He also talked about how much faster his team can act on what it learns. Work that once meant pulling data from different places and spending days evaluating an idea can now move into testing much sooner. That creates room for more experimentation, including smaller improvements that might never have made the priority list before. It is a practical use of AI: helping a team do more with the information it already has.

Your Data Does Not Have to Be Perfect

We spent a good part of the discussion on data, and I suspect most marketers could relate. Everyone wants a clearer picture of the customer journey, but that picture is rarely as complete as we would like. The interesting part was hearing how differently the panelists work around those limitations.

Jackie Atkin explained that at Fluke, a customer might research a product with the brand and ultimately buy through a distributor. That means the team does not always see every step between initial interest and purchase. They still have to understand what a customer is trying to accomplish and make the interactions they can influence more useful. Jackie also questioned the value of collecting data without a clear plan for using it.

Zach described a different version of the same challenge at Filson, where online and in-store customer activity does not always connect automatically. If someone buys in a store but their digital profile does not reflect that purchase, the next marketing message can be completely out of step with what they need.

Zach made a distinction I think is worth emphasizing: having access to data and knowing how to make a good decision with it are two different things. Better integrations certainly help, but teams can still make meaningful progress by focusing on the information that changes how they serve customers.

AI Supports the Work. People Protect the Brand.

Zach described Claude as his “omniscient intern,” which is a fairly accurate way to think about some of the best applications of AI in marketing. At Filson, it can take on technical work involved in building email flows. The creative team still develops the copy and imagery and makes sure the result sounds and feels like Filson. That division of work frees people to focus on the things the brand actually needs them to own.

For Jackie, that human involvement carries even more weight. Fluke makes technical equipment that people depend on to do their jobs safely, so accuracy is not negotiable. Her team is exploring AI-assisted experiences that help buyers narrow down complex product choices, including interactive product-selection tools. That is personalization with a clear purpose: helping someone make a more informed decision. It also makes knowledgeable review essential.

Nik echoed that point from the content side. Hyperfocus can analyze video patterns and audience reactions at scale, but the data does not decide what a brand should stand for. A creative approach may work extremely well for one company and feel entirely wrong for another. Someone who knows the brand still has to make that call.

At Bean Box, Ryan made the case for preserving the people and expertise behind the coffee. Customers are not just getting an algorithmic recommendation; they are buying into a carefully selected product and a relationship with the people who know it. AI can help the team work faster, but the brand still needs to feel like Bean Box.

Give Every Channel a Clear Purpose

Toward the end of the panel, an audience member asked how brands decide which messages belong in email, SMS, social and other channels. With so many ways to reach a customer, it is a fair question—and one that becomes even harder when technology makes it easy to personalize every touchpoint. Zach’s answer was to start with the role each channel plays, rather than with the message you want to send.

At Filson, for example, people sign up for SMS because it gives them priority access to limited releases or special opportunities. Email has more room for product information and brand storytelling. Social can build interest in the brand without asking every post to generate an immediate sale. Those are different jobs, and customers have different reasons for engaging with each one. Repeating the same promotion across all three would miss that distinction.

The idea extends beyond digital channels. Zach talked about Filson reaching out to customers with product-care information and inviting them to “Wax and Whiskey” events at its stores. I like that example because it gives people a reason to stay connected after a purchase without immediately trying to sell them something else. It also makes sense for a brand whose products are meant to last.

Ryan added an interesting counterpoint. Bean Box tested highly individualized SMS offers and found they did not necessarily perform better than a well-targeted broader message. There is a point where adding more complexity stops adding value. Give each channel a purpose, understand what customers expect from it, and test whether the personalization is actually helping.

Practical Takeaways for Marketing Teams

1. Start with relevance. Before a campaign goes out, consider who should not receive it based on purchases, engagement or immediate needs.
2. Use customer feedback to improve the experience. Cancellations, subscription changes and other behavior can show where expectations are not being met.
3. Work with the data you have. Identify the gaps that affect customers most and focus on decisions your team can make now.
4. Define the role of each channel based on what customers value. Then measure whether the messages and level of personalization are actually serving that purpose.
5. Use AI to help teams analyze, experiment and execute more efficiently, while keeping brand judgment, accuracy and customer trust with people.

The Customer Experience Comes First

The examples across the panel were quite different, but they all came back to fairly straightforward customer needs. Do not send me something irrelevant. Help me choose the right product. Pay attention when my experience is not working. Give me a reason to keep engaging with your brand after I buy.

Ryan made the point that customers are not buying personalization; they are buying great coffee. That is a useful reminder as marketing teams consider how far to take AI and automation. We have more tools and data available than ever, but those capabilities only matter when they help us deliver something customers actually want. The customer experience should be the starting point, not the afterthought.

Thanks to Zach, Ryan, Jackie and Nik for being candid about what they are trying, what is working, and where they are still figuring things out. That made the discussion far more useful than another conversation about what AI might eventually make possible.