E-commerce: The move beyond page-based to product-based analytics
The limitations of page-based analysis
Perhaps the most striking development in the growing sophistication of e-commerce sites and shoppers has been the dramatic increase in the importance of product aisle and product search pages.
These pages now do most of the heavy lifting when it comes to merchandising and often contain a rich mix of products, merchandising drives, and customer offers.
Remarkably, almost nothing interesting from a merchandising perspective is captured in the view of the page itself. The effectiveness of the page is a function of many internal merchandising levers.
For instance the products actually shown on the page, the order of those products, the offers made on the page, the mix of prices, discounts, presence of ratings and reviews on the page, and the mix of call-outs and calls-to-action displayed by product.
None of these is typically captured by the Web analytics tools whose ostensible function it is to assist online marketers make informed decisions.
The trend toward product set marketing
The vast majority of heavy merchandising lifting on e-commerce sites is no longer concentrated on the product detail page. Indeed, almost all the important drivers of consumer choice now come before it on pages at least one level up.
In a search driven world, the product contents of a facet page may be beyond optimisation except for a small set of high-demand searches.
For fixed aisle and category pages, discount and offer pages, and for a host of other relatively static pages, the actual product mix on the page is critical to the page’s success.
The products themselves, however, are only a part of the merchandising equation. The price mixture on the page has its own specific impact as does the associated product ratings and reviews – both of which are a curse and a blessing to the online seller.
Merchandising analytics explained
These product list pages have an almost daunting array of possible levers to pull. But options are only valuable when you have good methods for choosing amongst them.
As a merchandiser, it is imperative to fully understand product-set pages and ask some of these questions:
What is the:
- optimal price spread (highest to lowest) for a product set page?
- optimal gap between the highest and the average price of products on the page?
- optimal density of merchandising call-outs (such as discounts, banners, highlights) on the page?
- optimal value spread between discounts offered on a product set page?
- optimal density of discounts offered on a product set page?
- optimal position for the largest discount on the product set page?
- relationship between largest absolute and largest relative discount on a product set page?
The answers can pave the way for a truly significant improvement in product set merchandising – but how do you answer them?
The first step is in the collection of the necessary data to measure and analyse differential performance. That means knowing exactly what a shopper saw when they viewed a product-list page.
To do this, you need to create a method of understanding the layout of a page.
You need to capture the areas on a page that contain products, the grid layout of the those areas, the products (SKUs) and their price, discount, offers, rating, number of reviews, type of merchandising call-outs and position within the area.
With this data feed, you can to begin answering those merchandising questions.
By focusing on a subset of the key merchandising levers (such as density of call-outs or discount increments), it’s possible to develop data-driven rules to optimised overall page merchandising performance.
Controlled testing on a fixed product set can also produce the data necessary to optimise across multiple merchandising variables.
In areas like search, however, controlled testing is generally impossible. Instead, you'll need to analyse large numbers of product list combinations to create merchandising rules that can help drive the search results logic.
There is no one answer to the right balance between analysis and testing. Your best strategy is to start with a comprehensive analysis that identifies the most important merchandising levers and likely initial testing strategies.
By following this up with controlled tests, you can validate the results and further refine your merchandising hypotheses.
For most e-commerce sites, answering even the basic questions enumerated above provide significant opportunities for site improvement and competitive advantage.
Through careful analysis, you may find the best mix of levers to drive optimal merchandising performance on your critical multi-product pages.
For most e-commerce sites, pages with multiple products displayed are the single most important and impactful merchandising pages on the site.
The temptation is to treat these pages as if they were product detail pages – and simply add more and more merchandising levers to each product.
This doesn’t work. Unlike product detail pages, adding merchandising levers to the products on a multi-product page are more likely to shift the distribution of product clicks than to drive superior overall performance. Indeed, you may easily be shifting visitors from more to less profitable products.
What’s more, overuse of merchandising levers can create pages with “wall-to-wall” discounts that can erode brand perception and diminish the effectiveness of your merchandising strategy.
Careful use of analytics can help you understand the optimal density and type of merchandising levers as well as the optimal mix of products on the page (by price, ratings, etc.). For most e-commerce sites, there is no bigger optimisation opportunity in analytics.
- » Look to the cloud as the silver lining for CX and fast fashion’s global growth
- » The importance of becoming a ‘unifier’ CMO – and building relationships in the C-suite
- » Starbucks flips the equation with Deep Brew – by using AI for a more humanised customer experience
- » Gartner’s latest report reveals the top three barriers to marketing innovation
- » Is data-driven marketing in rude health? Why CMO opinion may not tell the full story