An approved product can receive impressions and clicks without producing a good business. A generic item may attract bargain shoppers who do not complete checkout. A sale may look successful in an ads dashboard but lose money after supplier shipping, support, and a return. A delayed order may produce a conversion today and a refund next month.
Measure Shopping visibility through the completed order, including fulfillment and exceptions. Use Merchant Center to understand whether products are shown and clicked, store analytics to see what shoppers do after arrival, and an order ledger to calculate what the sale actually contributed. Keep paid and free traffic distinguishable.
This is the final part of the Merchant Center Dropshipping Guide. The preceding articles establish eligibility, store trust and policies, product-data accuracy, and diagnostic resolution. Measurement should tell you whether those practices produce buyers you can serve profitably.
Define the funnel you can observe
Use a simple sequence: eligible offer → impression → click → product-page visit → add to cart → checkout → purchase → carrier acceptance → delivery → retained sale or return. Each stage answers a different question. No impressions may be a visibility or eligibility issue. Clicks without purchases may point to price, page clarity, shipping, trust, or mismatched search intent. Purchases followed by refunds point to product or fulfillment quality.
Google’s Merchant Center product performance guidance describes product-level impressions, clicks, and click-through rate, with ways to filter online store traffic and paid versus organic sources. Those metrics show discovery and interest, not the final order margin. Use them as the beginning of the analysis.
Choose a reporting period long enough for delivery and returns to mature. A seven-day campaign report may show sales but not the later cost of a slow cross-border shipment. Keep a cohort by order date or acquisition period and revisit it after the normal delivery and return window. A conversion is not the same as a retained, satisfied customer.
Separate paid and free visibility
Merchant Center reports free listings as organic traffic and can separate ads from free listings. Google’s product performance tracking guidance discusses using Google Ads for paid clicks and Merchant Center for unpaid clicks. Keep those streams separate because paid traffic has acquisition cost and may be optimized toward a selected conversion goal.
Do not compare raw conversion rates without context. Paid campaigns may reach a different audience or product mix than free listings. A new product may get many impressions from broad searches and a low click-through rate while a niche item gets fewer, more qualified clicks. Segment by product, variant, country, traffic type, and time before deciding that one channel is “better.”
Track the product ID consistently from Merchant Center to your storefront and order ledger. If the feed ID changes every sync, performance history becomes hard to reconcile. If a product has multiple variants, decide whether to analyze them separately or as a group and preserve the underlying IDs. A red variant with a high return rate should not disappear inside a good average for the whole product family.
Verify purchase measurement
A purchase event should fire when an order is actually completed, not when a shopper reaches checkout or reloads a thank-you page. Google’s Analytics ecommerce guidance describes purchase parameters such as transaction_id, value, tax, and an items array. Its transaction-ID guidance explains how IDs help deduplicate purchases and process refunds. Test your implementation with real or test transactions, including a return or cancellation.
Check what purchase value includes: item payment, shipping charge, tax, discounts, and refunds. Different tools can report different values by design. Record the convention and reconcile a small sample of transaction IDs to payment and order records. If one system counts a $40 customer payment and another counts a $36 item subtotal, that difference needs explanation before you compare return on ad spend.
Respect your privacy and consent setup when collecting analytics. Missing consent, ad blockers, attribution windows, and cross-device journeys can create differences between dashboards. Do not force the numbers to match by inventing sales. Use order records as the financial source and analytics as a directional view of how customers found and used the store.
Calculate conversion quality, not just count
Useful measures include product-page-to-purchase rate, checkout completion, retained-order rate, refund rate, on-time carrier acceptance, delivered-within-promise rate, buyer support contacts per order, and final contribution per order. Some require your own ledger; Merchant Center alone does not know all supplier charges or customer-service costs.
The AliExpress landed-cost guide gives a full contribution formula. For paid Shopping traffic, subtract ad cost attributable to the product or cohort. For free listings, there is no click fee for that traffic, but product work, site operation, and support still have costs. A “free” click is not a free fulfilled order.
Create a cohort table:
| Metric | Why it matters |
|---|---|
| Eligible products | Shows the share of the catalog that can actually appear |
| Impressions and clicks | Shows discovery and interest |
| Purchases | Shows checkout outcomes |
| Ad spend | Shows paid acquisition cost |
| Supplier and delivery cost | Shows cost of fulfillment |
| Refunds and chargebacks | Shows lost sales and exception exposure |
| Retained orders | Shows sales that survived the buyer experience |
| Final contribution | Shows whether the cohort supports the business |
Review by product and traffic source. A product with high click volume but repeated refunds may need a listing or supplier fix, not more bidding. A product with few clicks but strong retained contribution may deserve better images, titles, or more eligible coverage before budget is increased.
Interpret a simple fictional campaign
Imagine a product receives 10,000 impressions and 200 clicks in a month, so click-through rate is 2%. Twenty buyers purchase, a 10% click-to-purchase rate. Paid media for those clicks costs $120. If each order initially contributes $12 before advertising, the twenty orders contribute $240 before media and $120 after it. If four orders later require refunds that reduce cohort contribution by $80, final contribution falls to $40.
These amounts are illustrative, not benchmarks. The lesson is that the same campaign can look strong on clicks and purchase count while leaving little money after fulfillment. If the four refunds share a cause, fixing that cause may improve economics more than a higher bid. If refunds are due to inaccurate delivery promises, update the page and shipping settings as well as the supplier process.
Do not compare a two-week new-product test with a mature campaign as if the uncertainty were equal. Small samples are volatile. Use a defined test budget and stop rule. If a product cannot meet a reasonable contribution target under a conservative cost and return case, it may not be ready for scale even when the first few sales look promising.
Compare return on ad spend with contribution after ad spend. ROAS is attributed sales value divided by advertising cost under the reporting tool’s settings; it is not profit. A product with a 4:1 reported ROAS can still lose money if supplier, shipping, fee, and refund costs consume more than three quarters of the buyer payment. Conversely, a lower ROAS on a high-contribution product may be worthwhile. Calculate a break-even acquisition cost from the actual order margin before deciding what a campaign can afford to pay for a sale.
Be explicit about attribution limits. One shopper may see a free listing, return through an ad, and later purchase from an email. Different tools may credit different interactions, and late refunds can land outside the original campaign window. Keep the attribution setting visible in reports and use the order ledger to judge realized economics. Do not add the same purchase as separate revenue in organic and paid summaries simply because both influenced the journey.
Diagnose changes in the right order
When traffic or sales change, inspect eligibility first. Did products become disapproved or out of stock? Did shipping settings change? Is the product page reachable? Then inspect impressions and clicks by product and traffic source. Only after confirming that people arrive should you analyze page behavior and checkout. Finally, check fulfillment and return outcomes.
For paid campaigns, also review budget, bidding, audience or geographic settings, and conversion tracking. Google’s Shopping campaign fluctuation guidance notes that changes in conversion signals can affect campaign performance. A sudden drop in reported conversions could be a broken purchase tag rather than a sudden collapse in demand. Verify against the order ledger.
When click-through rate falls, examine image, title, price, competitive context, and whether Google is showing the offer for relevant searches. When clicks remain steady but purchases fall, inspect page speed, variant selection, checkout, delivery cost, and trust information. When purchases remain steady but profit falls, inspect supplier prices, ad cost, refund causes, and border or shipping charges.
Choose a decision, not a vanity metric
At the end of a test period, decide whether to keep, revise, or stop each product and campaign. Keep products with dependable supply, accurate offers, a satisfactory buyer experience, and contribution that survives normal exceptions. Revise products where a specific fix can be tested, such as a clearer size chart or a different verified shipping method. Stop products whose economics or claims cannot be supported.
Write the decision and expected effect before making a change. “Improve product page” is vague; “add measured dimensions to reduce size-related returns, then compare the next thirty delivered orders” is testable. The same discipline applies to a supplier switch, feed update, or budget increase. Change one major cause at a time when possible.
Use Merchant Center’s performance reports to find where buyers discover the offer, but let retained-order quality and final contribution determine whether that visibility deserves expansion. A small store wins when the promoted promise can be fulfilled repeatedly, not when a dashboard shows the largest number of clicks.