How to Discover Which Features and Products Users Like Best

You have 200 products in your catalog. Your Shopify Analytics shows sales by product, revenue by category, and page views by item. You can sort by top sellers. But sorting by top sellers answers the wrong question.

The question is not “which products sold the most?” It is “which products do users actually prefer?” Those two things diverge constantly, and the gap between them tells you exactly where to focus your catalog, your promotions, and your sourcing decisions.

Why Sales Volume Misleads

A product can generate high sales volume for reasons that have nothing to do with genuine preference: it was featured on the homepage, it had a discount running, it was in a paid campaign, it faced less competition in that category slot. Strip those factors away and you might find that users almost never return to that product, abandon it in the cart at twice the rate of other items, and never search for it directly.

Meanwhile, a product sitting quietly at position 47 in your catalog, rarely promoted, might have a 58% repeat purchase rate among the small segment that discovers it. That product is building genuine preference. You just cannot see it in a sales report.

eCommerce teams often optimize for the wrong signal because the right signal requires behavioral data, not just transactional data. What eCommerce product analytics actually measures is specifically this distinction: what users do across your catalog, not just what they ultimately buy.

What Product Preference Analytics Actually Measures

Product preference analytics looks at behavioral intensity and frequency across your catalog. For any given product or product attribute (color variant, size, category, price tier), it asks:

• How often is this product viewed relative to others with similar traffic exposure?
• What percentage of viewers add it to cart vs. just looking?
• Among buyers, what fraction returns to that same product within 30, 60, or 90 days?
• How does engagement compare to transactional volume? Are users showing strong interest without converting, which suggests a pricing or friction issue rather than a preference problem?

The goal is to separate products that perform because of distribution advantage from products that perform because users genuinely want them. Once you can see that difference, you can make better decisions: where to invest in inventory, which products deserve more catalog real estate, which attributes should appear in more of your product line.

A concrete example: Product A sells 3,200 units per month and sits at the top of your sales report. Product B sells 700 units. But Product A has a 61% cart abandonment rate and a 5% repeat purchase rate. Product B has a 24% cart abandonment rate and a 44% repeat purchase rate. Product B is the preference signal. Most stores are running ad spend toward Product A.

You can read more about this dynamic in how to find your best-converting products using product analytics, where conversion rate by product, not overall store conversion rate, reveals which items are actually earning your customers’ trust.

The Feature Dimension: Preference Within Products

Within individual products, preference signals also live at the attribute level. In eCommerce, “features” means product attributes: color variants, materials, bundle configurations, price tiers, sizes, use cases.

If 71% of customers choose the matte black variant of a product and only 3% choose neon yellow, that’s a preference signal about an attribute, not just a product. If the starter bundle version of a product converts at 3.8x the rate of the individual-item version, that’s telling you something important about how users prefer to consume that product type.

Most native eCommerce analytics tools do not surface attribute-level preference accurately. Shopify Analytics shows sales by variant but does not normalize for exposure. A variant that appears first in a dropdown gets selected more often regardless of genuine preference. Stormly’s Product Features Analysis measures interaction intensity relative to exposure, not just raw volume. That normalization is what turns click data into preference insight.

A kitchenware brand using Stormly discovered that their 4-piece starter set had 3.1x the engagement intensity of their individual item listings, but because the starter set was buried two clicks deeper in the navigation, it accounted for only 9% of total sales. Moving it to a featured catalog position increased its revenue share to 28% within 45 days. The preference data was always there. The distribution disadvantage had been hiding it.

How Stormly’s Product Features Analysis Works

Stormly’s Product Features Analysis pulls behavioral event data across your catalog and surfaces preference signals normalized by exposure. For each product or product attribute, it shows:

• Interaction intensity: meaningful engagements per session that includes this product (views, add-to-cart events, repeat views, and for stores with wishlists, saves)
• Preference rank vs. sales rank: which products rank higher on preference signals than on pure volume, and vice versa
• Engagement-to-conversion gap: high engagement with low conversion typically means a pricing, trust, or friction issue, not a preference problem. It is an actionable signal, not just a data point.

When Yogile, a photo storage platform, ran this analysis on their product feature set, the team discovered that their photo editor feature, which was an outsourced paid service, showed the lowest engagement intensity of any product capability. Users were navigating around it entirely. That single data point justified cutting the feature and recovering its ongoing licensing cost. The decision took an afternoon. Without preference analytics, it would have been a month-long internal debate.

For an eCommerce store, the equivalent is finding that a specific product variant or sub-category generates substantial traffic but near-zero engagement (no cart adds, no repeat views, immediate session exits). That is not a marketing problem. That is a catalog problem, and product preference analytics is the tool that makes it visible.

Once you have identified which products carry genuine preference signals, segmenting eCommerce customers by product behavior shows which customer groups are driving the strongest preference for which products, so you can align acquisition and retention strategy around those combinations.

Removing Low-Preference Products Matters As Much As Promoting Winners

One underused output of product preference analytics is the list of products to retire.

Catalog bloat is a consistent problem in eCommerce. A store that started with 40 SKUs and grew to 350 over three years often has a tail of 150+ products that generate more noise than revenue: inventory investment, navigation friction, diluted search relevance, and analytics that become harder to read. Because those products occasionally sell, they rarely get reviewed for removal.

Product preference analytics provides an objective signal for catalog pruning. Products with consistently low intensity scores, high exit rates immediately after product page views, and no repeat engagement across 90 days are candidates for retirement or consolidation. Removing them often improves catalog clarity, navigation usability, and conversion rate for the products that remain.

eCommerce stickiness analytics reinforces this: the products that keep users coming back across sessions are your stickiest assets, and they are frequently not the ones generating the highest raw sales volume. Identifying them through behavioral intensity, not just transaction count, is the starting point for a catalog strategy that compounds over time.

Using Preference Data for Sourcing and Development Decisions

For stores that develop or source their own products, preference analytics at the attribute level directly informs what to build or stock next. If your highest-preference products consistently share a specific characteristic (material, form factor, price tier, intended use case), that is a signal about what your customer base actually wants more of.

This is different from running a customer survey or reading product reviews. Behavioral preference data reflects revealed preference, what users do, not stated preference, what users say they want when asked. The two diverge constantly. Repeat purchase analytics for eCommerce reinforces this by identifying which specific products create the behavioral loop of genuine loyalty, not just post-purchase satisfaction.

Teams that use product preference data to inform sourcing close the gap between “this product performed okay” and “this product type is what our customers genuinely want more of.” That is where catalog strategy becomes intentional rather than reactive.

A Starting Point If You Have Not Done This Before

If you have not run a product preference analysis before, the fastest starting point is to look at three behavioral signals side by side for your top 50 products: cart abandonment rate by product, repeat purchase rate by product, and views-to-add-to-cart ratio. Products where all three signals align (low abandonment, high repeat, strong view-to-cart conversion) are your genuine preference winners. Everything else is either a distribution artifact or a pricing friction problem.

Stormly’s Product Features Analysis surfaces exactly this comparison across your full catalog without requiring manual data exports or spreadsheet work. You can run it within minutes of connecting your store data and see immediately which products have earned user preference and which are riding placement or promotion advantages. For teams building a repeatable decision process around this data, an eCommerce analytics workflow your whole team will actually use turns preference analysis from a one-time audit into a weekly input.

Find out which features and products your users actually want. Try Stormly free and run your first product preference analysis today.