The discussion around data in ecommerce really splits into two schools of thought. One side argues for the power of "big data," using sophisticated tools and complex models to achieve growth. The other side champions "small data," arguing that direct, qualitative customer insight is what truly moves the needle.
Camp A: Go Big on Data
This camp believes that success lies in aggregating and analyzing massive datasets. Proponents like Dane Atkinson, who I heard on Ecommerce Conversations, and Edward Upton on The eCom Ops Podcast, emphasize that a rigorous, quantitative approach is essential for scaling. The idea is to build a complete picture of the business by tracking key performance indicators (KPIs) like customer lifetime value (CLV), customer acquisition cost (CAC), and return on ad spend (ROAS).
By leveraging specialist analytics tools and sometimes even data warehouses, you can run cohort analyses, build out detailed customer segmentation, and use predictive analytics to forecast trends. Taylor Holiday and Andrew from Ecommerce Playbook often discuss the insights they glean from their portfolio of over 200 brands, showing how large-scale data can reveal patterns that are invisible to a single store operator. Their point is that with enough clean data, you can make better, faster decisions about everything from marketing spend to inventory management. It’s about building a system for Data-Driven Decision-Making that provides a real competitive advantage.
Camp B: Focus on Small Data
In the other camp, you have people like Kurt Elster, who on The Unofficial Shopify Podcast makes a compelling case for "small data." He argues that big data is often overwhelming and, for many businesses, not nearly as actionable as direct Customer Research. Big data can tell you what is happening, like a specific page having a high drop-off rate. But it can’t tell you why.
That "why" is where small data shines. It involves customer interviews, surveys, and analyzing user-submitted feedback. Instead of getting lost in spreadsheets, you talk to actual human beings to understand their frustrations, motivations, and desires. Let’s say your analytics show people are abandoning their carts. The "small data" approach is to find a few of those people and simply ask them what went wrong. The insights gained from just a handful of these conversations are often more direct and powerful than a complex attribution model. This approach is about finding high-impact opportunities by understanding the user experience on a deeper level.
I believe this is a false choice. You absolutely need both, and the best operators create a system where the two work together. Big data is your smoke alarm, and small data is your firefighter. Use your high-level analytics (the "big data") to identify where the smoke is. Is a specific campaign underperforming? Is your conversion rate for mobile traffic suddenly dropping? This quantitative insight tells you where to focus your attention. It points you to the problem.
Once you’ve identified the problem area, you deploy your "small data" tactics to figure out the cause. Watch session recordings of users struggling on that page. Run a poll. Email customers who dropped off and offer them a gift card for a 15-minute call. This qualitative research gives you the context, the story behind the numbers. It helps you form a hypothesis about what’s actually broken. From there, you can design a solution, run an A/B test, and use your big data analytics again to measure whether your fix worked. This creates a powerful feedback loop.
So, what should you do? If you’re an early-stage brand with limited resources, start with small data. It’s cheap, fast, and will give you the most impactful insights early on. Talk to your first 100 customers. As you grow, you’ll need to build more sophisticated big data capabilities to manage the increasing complexity of your business. The goal isn’t to replace your small data habits but to augment them. A scaled business that still talks to its customers is one that will continue to find opportunities others miss.
