Table of contents :

Popularity Is a Recommendation Strategy Too

100-days-of-woocommerce-personalization-025

Table of contents :

Personalized recommendations do not always need an individual customer profile. In many stores, the most useful signal is collective behavior: what other customers are buying, viewing, or adding to their carts. A carefully designed popularity strategy can help shoppers discover products quickly, especially when they have no purchase history.

What popularity means in WooCommerce

Popularity is not one universal metric. A store can rank products by different types of demand, depending on the recommendation placement and the customer journey:

  • All-time sales: Useful for evergreen products, but it can permanently favor older items.
  • Recent sales: A rolling period such as the last 7, 30, or 90 days can reflect current demand.
  • Units sold: Useful when comparing products with similar prices.
  • Orders containing the product: Helps avoid allowing unusually large orders to dominate the ranking.
  • Views or add-to-cart events: Useful for new products that have not accumulated many completed orders.
  • Category popularity: More relevant than store-wide popularity when the shopper is browsing a specific product category.

WooCommerce product sales counts are useful for a simple best-sellers implementation, but agencies should not treat the built-in sales count as a complete personalization system. It is generally cumulative, so it does not automatically represent current demand, customer segment, inventory risk, or product relevance.

Match the popularity signal to the placement

The same ranking should not be used everywhere. A store-wide “popular products” module on the homepage can use recent sales across the catalog. A product-page module should usually combine popularity with the current product’s category, attributes, or price range.

  • Homepage: “Popular this month” can introduce high-demand products across several categories.
  • Category archive: Rank products popular within the current category or subcategory.
  • Product page: Show popular alternatives with a similar category, use case, or price point.
  • Cart page: Prefer popular complementary products rather than simply showing the store’s best sellers.
  • Search with no results: Use popular products from the searched category or department as recovery recommendations.

Use a time window to prevent stale recommendations

An all-time best-seller list can be stable, but it can also become self-reinforcing. Products that appear in the list receive more exposure, which can produce more sales and keep newer products from entering the list.

A rolling window is usually more useful for merchandising campaigns. For example, an agency might calculate a popularity score for completed and processing orders from the last 30 days, then refresh the result hourly or daily. Products with no recent activity can still receive a fallback score based on longer-term sales or editorial priority.

A practical scoring model might look like this:

popularity_score = (units_sold_30_days * 4)
                 + (orders_30_days * 2)
                 + (add_to_cart_events_30_days * 1)
                 - (returns_30_days * 3)

The exact weights should be tested rather than assumed. For a high-priced catalog, order count may be more representative than unit count. For products commonly purchased in bulk, units sold may be the stronger signal.

Filter before ranking

Popularity should rank eligible products, not override basic catalog rules. Before displaying recommendations, exclude products that are:

  • Out of stock, unless the business intentionally accepts backorders.
  • Not purchasable or not visible to the current customer.
  • Restricted by country, customer role, membership level, or catalog permissions.
  • Already present in the shopper’s cart when the recommendation is intended to add a new product.
  • Discontinued, hidden, or unavailable for the current shipping region.

For variable products, decide whether popularity belongs to the parent product or a specific variation. A clothing store may rank the parent product by total sales across all sizes and colors, while a store with materially different variation prices or stock levels may need variation-level data.

Combine popularity with relevance

Popularity works best as one component of a recommendation score. A simple product-page formula could give relevance more weight than aggregate demand:

recommendation_score = (category_match * 5)
                      + (attribute_match * 3)
                      + (popularity_score * 2)
                      + (margin_priority * 1)

This prevents a globally popular product from appearing beside every item, even when it is not a meaningful alternative. For example, a popular coffee grinder may be appropriate beside another grinder, but not necessarily beside a coffee mug merely because both products belong to the same broad store catalog.

Popularity can also be segmented. A B2B customer may need the products most frequently purchased by businesses, while a consumer may respond better to products popular among shoppers in the same category or price range.

Implementation options for agencies

For a small catalog, a curated list or WooCommerce best-seller query may be sufficient. For larger stores, agencies should avoid calculating expensive order aggregations during every page request. Store the calculated result in a transient, object cache, custom table, or indexed analytics system and refresh it asynchronously.

A lightweight implementation can use a scheduled task to calculate product IDs for each placement, category, and time window. The frontend then requests a short cached list rather than querying all order items while a customer is waiting for the page to render.

When building a custom query, use WooCommerce and WordPress APIs where practical, apply capability and visibility rules, and use prepared database statements for custom SQL. Do not expose raw order data to the browser. The browser needs product identifiers and presentation data, not customer, order, or revenue details.

Protect new and lower-volume products

Pure popularity creates a cold-start problem. A new product has no sales history, so it cannot compete fairly with established products. Agencies can address this with controlled exploration:

  • Reserve one position in a recommendation block for a new or strategically important product.
  • Use category-level popularity when product-level data is sparse.
  • Apply a minimum exposure period before evaluating a new product.
  • Use a Bayesian or smoothed score so a product with two sales does not automatically outrank one with hundreds of reliable sales.
  • Rotate products with similar scores instead of displaying the same items on every visit.

Any exploration rule should still respect stock, margin, fulfillment constraints, and product quality. Popularity is a demand signal, not a substitute for merchandising judgment.

Measure more than clicks

Track recommendation impressions, clicks, add-to-cart events, purchases, revenue per session, and margin where available. A popular product may receive many clicks but produce fewer completed orders because it is expensive, out of stock, poorly positioned, or not actually complementary.

Use a control group or an A/B test when the store has enough traffic. Compare a popularity-only module with a relevance-plus-popularity module, and measure performance separately for new and returning customers. Also monitor concentration: if a small number of products receive nearly all recommendation impressions, the strategy may be reducing discovery rather than improving it.

For WooCommerce agencies, the operational requirement is straightforward: define the popularity event, define the time window, filter products before ranking, cache the result, and validate the business outcome. A “best sellers” label is only useful when the products shown are current, available, relevant, and measurable.

Trending posts
You might also like