- Published on
Product Intelligence: How High-SKU Shopify Stores Spot Hidden Gems Before Ads Burn Them
TL;DR
Shopify tells you what sold. Product Intelligence tells you what is a star, a hidden gem, at risk, or underexposed. On a catalog with thousands of SKUs, that difference is the difference between merchandising the whole store and merchandising the ten products you already know.
- Authors
- Name
- Isaac Lewin
- Shopify Architect
- @iliveoffgrid
Look.
You do not have a merchandising problem because you lack data. You have a merchandising problem because Shopify Analytics, Ads Manager, and your collection pages all shout about the same bestsellers.
The rest of the catalog is quiet. That quiet is expensive.
On a 5,000 SKU store the bestsellers are easy. The long tail is not. Filters disappear once a collection crosses Shopify's documented 5,000-product cap. Search stays literal. A shopper types the job they need done and gets zero results, even when the right SKU is in stock.
Merchants say this out loud in the Shopify Community and on Reddit. They split giant collections into smaller ones just to keep filters alive. They pay for traffic that lands on an All Products grid and then leaves. They keep buying ads for the SKUs that already convert because those are the only SKUs they can see clearly.
Product Intelligence exists to name the rest of the catalog.
What Product Intelligence actually is
It lives inside Path Explorer. It is not another revenue chart. It is a product list with four working labels:
- Stars: high sales and high popularity
- Hidden gems: they convert when people find them, but few people find them
- At risk: they used to move, then views or velocity dropped
- Underexposed: they barely show up in journeys at all
Those labels come from two axes you can see on the graph: sales score and popularity. Revenue sits next to the status so you spend time on the items that actually matter, not the €20 accessory that happens to look interesting.
You can export the table as CSV. On a 5,000 to 10,000 SKU catalog that export is how you compare this week to last week. The in-app table is the daily view. The spreadsheet is the memory.
That is the feature. One screen that sorts the catalog by how it actually behaves in sessions, not by what you featured last month.
Why the labels matter on a large catalog
A high-SKU merchant does not need another list of top sellers. They need a way to stop treating 8,000 SKUs like one pile.
Hidden gems are the point. These are products that earn the sale once a shopper reaches the page. They lose because discovery never gets them there. On a store with drop-ship plus second-chance inventory, that is most of the catalog most of the time.
At risk is the budget warning. A tray can sit high in Google Ads, burn spend, get turned off in Merchant Center, and then fall into At Risk because views collapse. Without a status, that looks like a random dip. With a status, you know the product lost the only channel that was feeding it.
Underexposed is the inventory shadow. The SKU is live. The collection exists. Nobody lands on it. The agent can recommend it if you train it to, but first you have to know it is invisible.
Stars are not the prize. They are the baseline. You already know those SKUs. Product Intelligence is useful when it tells you what is not a star.
Furniture operators with 5,000 to 10,000 SKUs already map these statuses onto the same mental model their ads specialist uses: heroes, dummies, ghosts. The labels become the input for metafields that decide which products get paid traffic. That is a concrete next action. Not a dashboard hobby.
How this compares to what you already have
| View | What it shows | What it hides |
|---|---|---|
| Shopify Analytics bestsellers | Units and revenue on products that already sell | SKUs that convert only when found |
| Google Ads / Merchant Center | What you chose to pay for | What would sell if anyone saw it |
| Collection merchandising | The order you set by hand | The 4,000 products you did not pin |
| Product Intelligence | Stars, hidden gems, at risk, underexposed | Anonymous sessions you cannot attach to a product |
You still need Shopify for orders. You still need Ads for spend. Product Intelligence is the missing middle: which products deserve a push, which ones the on-page agent should learn to surface, and which ones are leaking because nobody can find them.
The honest limits
Clicking a row today opens the Shopify product page. It does not yet spell out the why in one sentence (views dropped, velocity slowed, ad cost went up). Operators who live in this screen still have to read the graph and the revenue column themselves.
The full table on a 3,000-plus SKU store is a lot of rows. CSV export is how serious users compare weeks. Filters and pagination on that screen are still being tightened. Do not pretend the raw "All" list is a daily workflow.
Labels are only as good as the sessions behind them. Bot traffic can inflate popularity if you are not looking at the human count. Path Explorer has an exclude-bots toggle. Use it. Then trust the product statuses more than the raw session total.
Product Intelligence does not rewrite your catalog. Dirty titles, missing specs, and set-quantity products still confuse both shoppers and the agent. A hidden gem with a bad title stays hidden after you label it.
And a label that nobody acts on is just a badge. The value shows up when you move ad budget, pin a collection, or teach the agent which underexposed SKUs to offer when intent matches.
Where this sits in agentic commerce
Harley Finkelstein keeps repeating the same idea. Agents should surface what fits, not what paid for the slot.
The products that will be found will be merit-based as opposed to ad-based. I think that's democracy. That'll level the playing field.
That is the climax for a high-SKU store. Merit only works if the merchant can see which products already have merit and which ones never get a chance.
The on-page agent handles the shopper side. Product Intelligence handles the merchant side. One classifies the catalog. The other can put a hidden gem in front of the person who asked for it.
You still have to do the work after the label. Turn off the at-risk SKU that is burning ads. Feed the hidden gems to the agent. Stop buying more traffic for products the grid already over-serves.
That is Product Intelligence. Four statuses, a graph, a revenue column, and a CSV. Enough to stop guessing which 200 SKUs in a 8,000 SKU store deserve the next hour of your week.
Install ShopGuide and open Product Intelligence in Path Explorer
Frequently Asked Questions
What is Product Intelligence in ShopGuide?
It is the Path Explorer view that scores products on sales and popularity, then labels them as stars, hidden gems, at risk, or underexposed so you can act on the long tail instead of only the bestsellers.
How is a hidden gem different from a bestseller?
A bestseller already gets traffic and orders. A hidden gem converts when someone reaches it, but the journeys show almost nobody gets there. That is a discovery problem, not a product problem.
Can I export the list?
Yes. CSV export is how merchants with 5,000-plus SKUs compare statuses week to week. Use the in-app table for a scan. Use the file for history.
Does this replace Google Ads or Shopify Analytics?
No. Ads still spend the money. Shopify still records the order. Product Intelligence tells you which SKUs are earning their visibility and which ones are invisible or slipping.
Will the agent automatically push hidden gems?
Not by magic. The agent recommends from live catalog data and whatever training you give it. Product Intelligence is how you decide which SKUs deserve that training and that merchandising attention.
Is this useful under 1,000 SKUs?
You still get the labels. The pain is sharper once the catalog is large enough that bestsellers drown everything else and native filters start to fail.
Master Agentic Commerce
Join Shopify founders receiving weekly insights on AI agents and autonomous growth.
Trusted by top Shopify Plus brands