Every collection page puts products in some order. On most stores nobody chose that order, which means the platform chose it, and the platform doesn't know what makes money.
Open any collection page on your store. The products are in some order.
Somebody decided that order. Statistically, it wasn't you.
It was whatever your theme defaults to, which is usually creation date, alphabetical, or a "best selling" sort that ranks by historical units and therefore recommends what already sold because it was already at the top. That last one is worth sitting with. Best-selling sort is a feedback loop. Products that got placed well sell well, which keeps them placed well.
Merchandising is the practice of making that decision on purpose. In a physical store it's the most obvious job in retail. Online it usually belongs to nobody.
What merchandising actually controls
Four things, roughly in order of how much they matter.
Which products appear, and where. Collection order, what sits above the fold, what gets a homepage slot, what surfaces in search results.
What gets grouped together. Collections themselves. Whether "gifts under $50" exists, and whether it's curated or a price filter with a nice name.
What appears alongside a product. Related items, complete-the-look, cross-sells on the cart. What you put next to the thing someone is already looking at.
What's emphasized on the item itself. Badges, stock messaging, the order of images, which variant loads first.
None of that is design. You can merchandise a plain store well and a beautiful store badly, and the beautiful one will lose.
The default sort is working against you
The platform's default doesn't know anything useful about your business.
It doesn't know your margins, so it happily puts your thinnest-margin product first. It doesn't know your stock position, so it promotes the item you're about to sell out of and buries the one sitting in a warehouse accruing storage fees. It doesn't know what converts, so it treats a product with a 4% conversion rate the same as one with 0.3%.
And it usually doesn't know what's out of stock. Plenty of stores leave sold-out items in the first row of a collection. Every shopper who clicks one has spent their attention on a dead end, and you paid for that session.
Most of your traffic is on mobile, where the first screen holds two products, maybe three. Median mobile session conversion on Shopify runs around 1.2% against 1.9% on desktop. The mobile shopper sees a fraction of the collection and decides from it. What's in the first two slots on a phone is close to the whole decision.
A sort order that isn't stupid
You don't need an algorithm. You need to stop using the default.
A reasonable starting rule, in priority order:
- In stock. Sold-out products go last or get hidden. This is the cheapest win on the list and a surprising number of stores skip it.
- Converts well. Products with a high view-to-purchase rate earn their position. You already have this data.
- Margin. Where two products convert similarly, promote the one you keep more of.
- Inventory pressure. Push what you're overstocked on, protect what's nearly gone.
- Newness, last. New products need exposure to earn data, so give them a window rather than the permanent top slot.
That's a rule, not a ranking model, and it beats creation date by a distance.
Curation beats filtering
A filter is a query. A collection is an opinion.
"Under $50" is a filter. "Gifts for someone who already owns everything" is merchandising, and it works because it does the thinking the shopper was about to do. The same eight products in both, and the second one sells more.
This is where most stores have room. Collections get built to mirror the catalog structure, which is a warehouse view of the world. Shoppers arrive with an occasion, a problem or a person in mind. Building three or four collections around those, and putting them where people land, is a day of work with a long tail.
The bit that gets skipped
Merchandising decays. That's the actual reason it goes undone.
Stock changes weekly. Seasons turn. The product that converted brilliantly in March is wrong in August. A sort order set once in a burst of enthusiasm is out of date within a quarter, and because nothing breaks visibly, nobody notices. The store just quietly gets worse at selling.
So the useful question isn't what the right order is today. It's who's changing it, how often, and how they know it worked.
That's the job Qosmic does. The audit finds the collections and placements worth changing, then we build the change and adjudicate it against real revenue, both halves, so a new sort order comes back as a number rather than a preference. It runs continuously, which matters more here than in almost any other part of a store, because this is the surface that goes stale fastest.
You can do it without us. A monthly calendar hold, an export of conversion by product, and somebody with authority to reorder collections will get you most of the way. The failure mode isn't capability. It's that the calendar hold gets moved.
Do this week
- Hide or demote out-of-stock products in every collection.
- Pull conversion rate by product. Move the top five up in their collections.
- Open your three biggest collections on a phone and look at the first four products. Ask whether those are the four you'd hand someone.
- Build one curated collection around an occasion rather than a category.
- Put a recurring hold in the calendar to do it again.
The phone check takes two minutes and usually settles the argument about whether this matters.
