The $2M annual revenue mark is a genuine inflection point for DTC brands. At this stage, the founder is typically no longer managing every product listing personally. A first dedicated merchandising hire is either just in the door or being recruited. Shopify's out-of-box tools, which were perfectly adequate at $500K, are starting to show friction. And the catalog has usually grown from a focused initial range to something broader and harder to manage coherently.
This is the moment when merchandising mistakes get baked in. Not because the team is negligent, but because the patterns that worked at smaller scale don't map forward, and the new merchandising hire walks into a system that was never designed with them in mind.
Here are five merchandising problems we see consistently in DTC brands at this stage, and what to do about each one.
Mistake 1: Treating Bestseller Sort as Neutral
Sorting a collection by units sold seems like the most objective possible default. The products customers buy most go first. What could be wrong with that?
The problem is that bestseller rank is a historical aggregate. It reflects what sold to your past traffic mix, across all channels, over whatever window your platform uses for the calculation. That mix includes every visitor who ever landed on that collection, regardless of intent, traffic source, or device. The bestseller list from your last 90 days of traffic is an average of many different purchase intents, not a good predictor of what any individual visitor right now will convert on.
For brands with substantial paid social traffic, bestseller sort compounds this problem further. Social traffic tends to over-index on whichever product has strong creative or has been running a discount. That product accumulates conversions, rises to position 1, gets more traffic, accumulates more conversions. The bestseller list calcifies around a few social-amplified products while organically performing products stay buried. A merchandising manager inherits a collection order that was shaped by Facebook ad performance, not product-market fit.
The practical fix at this scale: segment your bestseller calculation by traffic source. The bestseller order for email traffic and organic search traffic is often substantially different from the order driven by paid social. Apply different collection ordering rules per traffic source entry point, even if those rules are manual and revisited monthly.
Mistake 2: No Differentiation Between Mobile and Desktop Collection Ordering
At $2M revenue, mobile likely accounts for 60-70% of your traffic and a lower share of conversions. The conversion rate gap between mobile and desktop is real and persistent for most DTC categories, and it is partly a merchandising problem, not just a UX problem.
Mobile visitors see fewer products in the first scroll than desktop visitors. A 4-column desktop grid shows 8-12 products above the fold. A 2-column mobile grid shows 4-6. The products in positions 5-8 on a mobile grid are effectively invisible to the majority of mobile visitors who don't scroll past the first two rows. If your collection ordering is designed for a desktop view where 8 products are visible immediately, you may be burying high-intent products below the mobile fold where most of your traffic never reaches them.
This is solvable without a full replatform. Most Shopify themes support metafield-based collection ordering, and a simple mobile-specific product ordering that surfaces your fastest-moving products in the top 4 positions is within reach for a merchandising hire with basic theme familiarity. Test it with your highest-traffic mobile collection first and measure RPV changes before expanding.
Mistake 3: New Product Launches Dying in the Grid
A new product gets launched. It has no sales history, so the bestseller sort pushes it to the bottom of the collection. The launch email sends traffic, and those recipients find the new product because they know to look for it. But organic traffic and paid traffic to the collection see the established bestsellers first and never scroll to where the new product sits.
After two weeks, the product has 40 orders. The founder asks the merchandising manager why the launch performed below expectations. The real answer is that 80% of collection traffic never saw the product above the fold. The launch email performance was fine. The storefront performance was invisible, because the ranking system is indifferent to newness.
At the $2M scale, this requires a deliberate new product promotion protocol. Not just an email. A collection placement rule that pins new launches to positions 1-3 for a defined window (2-4 weeks depending on catalog velocity), overriding the bestseller sort. This gives the new product a fair exposure window. After the window, the product earns its rank based on actual performance data, and the pin is lifted. Without a protocol like this, new products at DTC brands systematically underperform their actual demand because the discovery mechanism fails them from day one.
Mistake 4: Using Return Rate as a Negative Signal Without Context
When a DTC brand's merchandising hire starts pulling data, return rates by product are often one of the first things they look at. High-return products are candidates for demotion in the collection. That instinct is understandable but incomplete.
Return rate without context is a misleading signal in product ranking decisions. A product with a 22% return rate in a category where the category average is 24% is performing well. A product with an 18% return rate where the category average is 8% has a fit or expectation problem. The absolute return rate number is nearly meaningless without the category baseline.
More importantly, some high-return-rate products are high-margin products that attract high-value customers who return them occasionally but also make multiple annual purchases. Demoting those products in the collection because of return rate alone, without looking at customer lifetime value or repeat purchase behavior, can push your collection toward a lower-margin mix that is easier to return-manage but worse for the business overall.
We're not saying ignore return rates. We're saying return rate is a product quality diagnostic, not a collection ranking input, unless it is adjusted for category baseline and paired with LTV and margin data. Build that adjusted view before making ranking decisions based on returns.
Mistake 5: Collection Architecture That Maps to Your Org Chart Instead of Customer Intent
This is the subtlest mistake and often the most expensive one. Collection architecture at early-stage DTC brands usually maps to how the founding team thinks about the catalog: by product category (tops, bottoms, accessories), by brand season (fall 2024, spring 2025), or by product line (the core range, the premium range, the collab range).
These are internally coherent organizing principles that make sense to the team. They are often not how customers arrive with intent.
A visitor who clicks a paid search ad for "breathable workout shorts women" arrives with activity-specific intent. If your collection architecture routes them to a general "shorts" category that includes shorts for five different activity types, they're doing extra filtering work on a page that was designed around your catalog structure, not their purchase intent. Browse abandonment from these mismatched landing experiences is often much higher than from well-matched collections.
The fix requires auditing your top traffic entry points by search term and ad creative, and asking honestly whether your collection architecture serves those specific visitor intents. In many cases, this means creating intent-specific collections that duplicate products across multiple categories, which feels redundant internally but dramatically reduces friction for the visitor who arrives with a specific use case in mind.
At $2M revenue, you have enough data in your analytics to identify the top 10-15 search intent profiles driving traffic to your collections. Auditing collection architecture against those profiles once per quarter, and updating collection organization to match, tends to produce better RPV gains than any ranking algorithm change alone. The merchandising system can only help a visitor find what they want if the collection they land on was designed with their intent in mind.