A simple and effective way to decide which products to advertise is to position each product against two questions: how much interest does the product get (GA4 product views) and does that interest turn into sales (purchases). When you compare these two axes with your store's own average (preferably the median), products fall into four groups: those viewed a lot and sold a lot, those viewed little but selling well, those viewed a lot but selling little, and those weak on both. The advertising decision differs for each group; the biggest opportunity is usually in products that are viewed little but convert well, because these products are likely to bring sales when they get more impressions.
Why isn't ad platform data enough?
Ad platforms only see the traffic that comes from their own ads. They do not know how a product performs from organic search, social media or email. GA4, on the other hand, measures all traffic on the site at product level:
- Product views (
view_item): How many times was the product page viewed? - Add-to-carts (
add_to_cart): How many times was it added to the cart? - Purchases (
purchase): How many units/times was it purchased?
Make sure these events are set up correctly; in particular, sending the product ID (item_id) consistently in all events is essential for the data to be matched to the product.
The product matrix: four groups
Split products on two axes against the store median:
| Group | Views | Purchases | Typical decision |
|---|---|---|---|
| Stars | Above the median | Above the median | Keep them in ads, protect the budget, watch the stock. |
| Hidden gems | Below the median | Above the median | Increase visibility; test with a separate campaign or a higher budget. |
| Opportunities | Above the median | Below the median | There is interest but no sales: review price, image, size, description and stock problems. |
| Underperformers | Below the median | Below the median | Reduce the ad budget or move them to a separate, limited campaign. |
The reason for using the median instead of the average is to prevent a few best-selling products from pulling the average up and making almost every product look "below average".
Choosing the period
A short period (7 days) shows the current situation but is noisy; a long period (90 days) is more stable but catches seasonal shifts late. In practice a 30-day period is a good starting point; you can choose a shorter one in fast-moving categories and a longer one in slow-moving ones.
Ad structure by group
- Stars are the backbone of your main campaign. When they run out of stock, the ads should stop too; apply the rules in the stock and broken size run guide to this group in particular.
- Create a separate campaign or listing group for hidden gems. In Performance Max, these products may not get a share of the budget while they sit in the same campaign as the stars.
- Opportunities are most often not an advertising problem but a product problem. If views are high while sales are low, check price competitiveness, product images, size availability and the product page.
- Before shutting off the underperformers group entirely, check whether the product is new; new products have not accumulated data yet.
The easiest way to use these groups in campaigns is to write the product's group into the feed as a custom label. Details are in the custom label strategy guide.
Signals to look at beyond the matrix
- Trend: Compare the short-term daily average with the long-term one. Products whose views are rising fast are the season's standouts; those falling are losing favor.
- Channel breakdown: If a product sells well from organic search but is weak in the paid shopping channel, there may be a problem with the ad setup.
- Add-to-cart rate: Products that are added to the cart but not purchased may have shipping, price or size problems.
- Market price: The price comparison data in Merchant Center shows whether your price is above or below the market.
Implementation steps
- Verify that GA4 e-commerce events arrive correctly with the product ID.
- Split products into four groups using 30 days of data.
- Write the groups into the feed as a custom label.
- Split the campaign structure by group; give hidden gems a separate budget.
- List product-side improvements for the products in the opportunities group.
- Check how the groups have changed every two weeks.
The HexFeed product matrix
HexFeed pulls product-level view, add-to-cart and purchase data through its GA4 connection. The Product Matrix screen compares this data with store medians and splits products into four: Stars, Opportunities, Hidden Gems and Underperformers; it has a 7–90 day period, segment and category filters, and can be exported to Excel. Product Trends compares short-term and long-term averages, and Product Source shows the last 30 days by GA4 channel group.
This data can be used directly in feed rules: 7, 15, 30, 45 or 90 day GA4 view, add-to-cart, purchase and conversion metrics, the matrix segment, purchases by GA4 channel group and price difference against the Merchant Center market price. You can also sort the products in the feed by a metric of your choice after the rules and limit the feed to the top N products. For details see the product matrix and feed filtering pages; for setup, see the Google Shopping with Shopify page.
Checklist
- The
view_item,add_to_cartandpurchaseevents in GA4 arrive with the product ID. - I group products by the median, not the average.
- I chose the period according to how fast my category sells.
- The product's group is written into the feed as a custom label.
- There is a separate campaign or listing group for hidden gems.
- I reviewed price, image, size and page problems in the opportunities group.
- I evaluate new products separately from the underperformers group.
- I check how the groups change at regular intervals.
Frequently asked questions
Why does GA4 data differ from ad platform data?
The two systems use different attribution models and different measurement methods; you should not expect an exact match. What matters for the product matrix is not the absolute values but how products rank relative to each other.
Is the matrix meaningful for small stores with little data?
If the number of sales is very low, the groups become noisy. In that case, choosing a longer period (e.g. 90 days) or also taking add-to-cart data into account can give more meaningful results.
Should I stop advertising products in the opportunities group?
Not right away. These products attract interest; the reason they do not convert is most often a price, size, image or page problem. Find and fix the cause first, then reassess the advertising decision.
How often should matrix groups change?
The groups can update on their own based on data, but on the campaign side you do not need to react to every small change. A review rhythm of every two weeks or once a month is enough for most stores.
