How to Analyze Product Performance in Ecommerce: A Step-by-Step Guide
by Trivas.ai
|
8 min read
Oct 20, 2026
Why Most Product Performance Reviews Miss the Point
Most teams run a product performance review by sorting a spreadsheet by units sold and calling it done. That tells you what moved. It doesn't tell you what made money.
A SKU can rank third in revenue and still be a loser once you factor in a 22% return rate and the ad spend it took to get that sale. Another SKU might sit near the bottom of the revenue list but convert at twice the site average with almost no paid support behind it. Revenue alone hides both stories.
Here's the thing: "performance" isn't one number. It changes depending on the decision in front of you. Deciding whether to reorder inventory needs a different lens than deciding whether to cut a SKU or raise its price. If you're learning how to analyze product performance ecommerce-wide, the goal isn't a prettier spreadsheet. It's a framework that matches the right metric to the right decision, every time you run the review. That's what this guide walks through, step by step.
Step 1: Define What "Good Performance" Means for Your Catalog
Before you pull a single number, sort your catalog into rough buckets: hero SKUs, growth candidates, long-tail products, and decliners. Each group gets judged differently. A hero SKU needs to hold margin and turn inventory fast. A growth candidate needs a rising conversion trend, even if revenue is still small.
This matters because a flat comparison across your whole catalog is misleading. A $20 impulse-buy product and a $200 considered purchase shouldn't be judged against the same conversion rate benchmark. The $20 item might convert at 4% and be healthy. The $200 item converting at 4% could be outstanding, or it could be masking a cart abandonment problem depending on your traffic source.
Set up a simple scorecard before you go further. Four columns, minimum:
Revenue contribution (share of total, not just raw dollars)
Gross margin %
Return rate
Inventory turn
That's your baseline. Everything else in this guide adds nuance on top of it, but if you skip this step you'll end up comparing apples to a completely different fruit.
Step 2: Pull the Core Metrics You Actually Need
With the scorecard structure set, you need the raw inputs. The full list looks like this:
Units sold
Net revenue (after discounts and refunds)
Gross margin
Conversion rate per product page
Return/refund rate
Repeat purchase rate
Days of inventory on hand
Each one lives somewhere different. Units sold, net revenue, and margin usually come out of Shopify order data. Conversion rate per product page lives in GA4, tied to on-page behavior rather than order data. If you sell on Amazon, Seller Central or Vendor Central holds its own version of units, returns, and even margin, calculated with a completely different fee structure than your own store.
This is the point where most teams hit a wall. None of these platforms talk to each other natively. So the "analysis" becomes an afternoon of exporting CSVs, renaming columns so they match, and praying the SKU IDs line up between Shopify and Amazon. It works, until you need to do it again next month and the manual process starts eating a real chunk of someone's week.
If you're running this across channels, it's worth seeing how a Shopify-specific data setup handles the export side before you build another manual pipeline by hand.
Step 3: Connect Sales Data to Ad Spend Per Product
Most DTC brands track ROAS at the campaign level and stop there. That's a mistake. Campaign-level ROAS tells you an ad set is profitable. It doesn't tell you which product inside that ad set is actually driving the return.
Product-level ROAS is the number most brands skip, mainly because it's annoying to calculate. You have to take spend from Meta or Google, broken out by campaign or ad set, and match it to the specific SKUs or collections that campaign was promoting. If one campaign pushes three products and only one of them converts well, blended ROAS will tell you the campaign is "fine" while quietly funding two underperformers.
The practical issue is structural: ad platforms report spend by campaign and ad set, not by product. There's no native field that says "$400 of this spend produced this SKU's sales." You have to join ad spend data with order-line data yourself, usually by mapping UTM parameters or landing pages back to product IDs.
Once that join exists, the picture changes fast. A product with strong revenue but a bloated ad spend behind it might be break-even or worse. A quiet, low-spend SKU might be your actual margin engine. You won't see either until spend and sales sit in the same table.
Step 4: Build a Repeatable Weekly or Monthly Review
A one-time deep dive is nice. It's also useless three months later when nobody's updated the numbers. Set a cadence: weekly check-ins on your hero SKUs, a full monthly pass across the whole catalog.
Keep the review format dead simple. Rank your top 10 and bottom 10 products by margin-adjusted revenue (revenue minus returns and ad spend, not just gross sales) and look at them side by side. Patterns show up fast once you're not scrolling through 400 rows.
From there, flag each product for one of three actions:
Reorder: high inventory turn, healthy margin, no return rate spike
Reprice: strong traffic, weak conversion, usually a pricing or positioning problem
Cut: high returns, thin or negative margin, no improvement over multiple cycles
This is also where a BI reporting setup earns its keep. A dashboard that already has last week's and last month's numbers sitting side by side turns this into a 20-minute review instead of a half-day spreadsheet rebuild.
Step 5: Automate the Parts That Don't Need a Human
Here's the honest reason most teams stop doing this analysis consistently: pulling Shopify, Amazon, and ad platform data into one spreadsheet by hand gets old fast. It's not that the framework is hard. It's that the manual labor behind it is tedious enough to get skipped the first busy week, and then it just stays skipped.
A centralized dashboard built on a data warehouse (Trivas runs on Amazon Redshift) solves the stitching problem by pulling Shopify, Amazon, and Meta or Google ad data into one product-level view automatically. No manual exports, no matching SKU IDs across three tabs by hand.
On top of that, an AI insight layer can do the scanning a human usually skips because it's boring: flagging a SKU whose return rate jumped 8 points in two weeks, or catching margin erosion on a product before it shows up as a quarter-end surprise. That's the difference between AI-driven insights and a static dashboard that only shows you what you already knew to look for.
If you sell on Amazon, this matters even more, since Amazon's reporting structure and fee logic don't map cleanly onto Shopify data without some translation layer in between.
Common Pitfalls When Analyzing Product Performance
A few mistakes show up constantly, even on teams that are otherwise disciplined about reporting.
The first is comparing products across wildly different price points using revenue alone. A $150 product generating $15,000 in monthly revenue looks better than a $25 product generating $8,000, until you check margin and discover the cheaper product is actually more profitable per dollar of revenue.
The second is ignoring return rate until it becomes a cash problem. Returns don't just cost the refund. They cost the original shipping, the restocking labor, and often the ad spend that acquired the customer in the first place. A SKU with a quietly climbing return rate is a slow leak, not a one-time hit.
The third is treating Amazon and Shopify performance as interchangeable. They're not. Amazon fee structures, return policies, and customer intent (someone searching Amazon is often price-comparing) are different from a Shopify storefront where traffic usually arrives already warmed up by your own marketing. A product that's a star on one channel can be mediocre on the other, and judging both with the same benchmark hides that completely.
Next Step: Turn This Into an Ongoing Process
The framework, in short: define what good looks like for each product segment, pull the real metrics instead of just units and revenue, connect ad spend to specific SKUs, review on a fixed cadence, and automate the data-stitching so the analysis actually happens every month instead of only when someone has a free afternoon.
Knowing how to analyze product performance ecommerce-wide isn't complicated once the data is in one place. The hard part is almost never the math. It's the manual work of gathering numbers from four different platforms before you can even start.
If that's the part slowing your team down, it's worth seeing what a connected view looks like instead of rebuilding the spreadsheet every month. You can check out more frameworks like this one in our guides and reports library, or just start a trial and see your own product data pulled together automatically.
Content author and contributor at Trivas.ai, sharing insights on e-commerce analytics, business intelligence, and data-driven strategies to help businesses grow.
Continue Reading
explore more insights
Understanding Customer Lifetime Value (CLV)
3 min read
Ecommerce Analytics for DTC Brands with 100+ SKUs: What Actually Works at Scale
3 min read
Triple Whale Missing Features That Trivas Has (2026 Comparison)
.d53b12e5.png&w=3840&q=75 "Trivas.ai")
