For a Category Manager of an online shop with hundreds or thousands of products, it can get difficult to find those products that are performing poorly AND are worth optimizing.
You cannot simply work on all products that did not sell well last week, there are just too many. Focusing on only those products with high margins is not going to help much either.
Here are some helpful performance metrics you can create in Adobe Analytics (formerly SiteCatalyst) – and partially also in Google Analytics.
Product Optimization – What is That?
Product optimization means improving your current assortment. Let’s focus on what the Category or Merchandising Manager can do on-site here, e.g.:
- improving the Product Detail Page (PDP), e.g.:
- checking for missing or wrong pictures or pictures of bad quality
- improving text (correcting product description, features, title)
- adding relevant text so that the on- and offsite search can find the product better
- improving visibility/marketing, e.g.:
- adding lucrative products to a homepage or category page teaser or placing it higher in lists
- feature the product more prominently in marketing campaigns or increase bidding price for ad clicks or impressions
- prioritize the product in recommendation engines
- shutting down or improving those ads that bring inefficient traffic to this product or category
Product Performance in Terms of (Under-)Utilized Potential
Some people measure product performance simply by looking at the sales figures (Orders and Revenue) per product or category. This “bottom line” of course is important, but it does not say anything about the performance of a product in terms of “(under-)utilized potential”. And plain sales figures do not help you find the products that could get you quite some revenue if you invested some effort in optimizing them.
So how can you measure product performance with a Web Analytics tool and maybe a bit of help from Excel? Let’s focus on metrics that are most helpful when improving on-site performance, and let’s use metrics which are still basic enough so that you can create and use most of them without additional implementation effort. You do need to have a working implementation of either Google Analytics Enhanced Ecommerce or Adobe Analytics ecommerce tracking from at least the Product Detail Pages on to the order page though as the basis.
Now to the metrics:
1. PDP Views (better: Visits)
Nothing spectacular, but of course you need a base metric to know how much traffic (how many Visits) you are basing your later ratio metrics on, or in other words, how large the front end of your funnel is. A Cart-to-Detail Ratio of 50% does not say much if it is based on only 10 Visits (Sessions).
Google Analytics offers only Hit-based Product Detail Views (not Sessions) in its Enhanced E-Commerce Reports (Conversions -> Ecommerce -> Product Performance -> Tab “Shopping Behaviour”), so views and especially the ratio metrics (see below) that are based on the Views will be distorted if a product is viewed several times in a Visit (views upwards, ratios downwards).
2. Cart-to-PDP Ratio
Definition: Visits in which this product was added to the cart at least once divided by Visits in which the product detail page (PDP) for this product was viewed at least once. Closest thing to it in Google Analytics: Cart-to-Detail Ratio (but again, problematic because hit- and not session-based).
This is the key performance indicator (KPI) of the PDP because it tells me: How well does my PDP perform at persuading people to take the next step in the funnel, which is adding the product to their cart. For the Category Manager, if a product gets decent traffic (again: better: Visits, not so much Detail Views) and Cart-to-PDP Ratio still stinks, the product needs to be looked at. So the Category Manager could work with a daily list of all products in his category that had at least x PDP Views yesterday and order that by the worst Cart-to-Detail Ratio. In our screenshot we do have quite some products with a ratio of 0%. That way he gets those products that perform poorly, but which also receive enough traffic so it’s worth working on them (1pV means “counted once per Visit”, so “PDP Views (1pV)” is the same as “PDP Visits”):
3. Orders-to-(Cart-or-Checkout) Ratio
Definition: Visits in which the product was ordered divided by Visits in which the product was added to the cart or the checkout was started with the product in the cart at that time.
You may ask: Why not simply the Buy-to-PDP Ratio? Because that one does not help much with optimizing the product, as so much happens between viewing and buying a product.
People often add products to their cart to not having to find them again later (because they know that carts usually store products for some days) and then they start comparing on other sites. If they find a better-priced product or an easier checkout experience somewhere else, they may buy it there. That might occur to a greater or lesser degree depending on the competition for this product or the price (the more expensive, the lower this rate usually becomes).
So I want to focus on that slice of users for whom my PDP has done its job (= product is in cart), and I want to evaluate how the rest of the site performs, e.g. the Checkout or the Retargeting campaigns that catch cart abandoners.
That being said, the typical helper metrics should be there as well of course:
4. Orders-to-PDP Ratio
Definition: Orders (in which the product was a part) divided by PDP Visits. Sort of the Conversion Rate for Category Managers, however not very actionable. The closest thing to this in Google Analytics is the “Buy-to-Detail Rate”, again this one is not sessionized, so it is prone to distortion if a product is viewed a lot by some visitors.
Moreover, in GA, the Buy-to-Detail Rate mostly makes sense only on the product level (e.g. SKU or Product Name). If you use it on a category level, it will be based NOT on “Orders which included products from this category” (which is what everyone thinks) or on “Units sold” (that is the “Quantity” metric). Instead, the Buy-to-Detail Ratio is based on the “Unique Purchases“, which is a rather tricky GA E-Commerce Metric and means: “how many times was at least one unit of this product bought” (e.g. if you buy 2 shorts, 3 pants and 9 shirts from the “clothing” category, the category will get 3 Unique Purchases in GA, not one (!), because the metric deduplicates at the product, not the category level!). Very confusing and number 2 among misinterpreted Google Analytics metrics right after the evil Sessions which are actually Entrances.
5. Average Order Value
Shows you the potential in revenue gains if you managed to improve the Orders-to-PDP Ratio. So the metric helps you to focus on the categories which usually sell expensive products instead. If people order only cheap cables from your computer section instead of expensive laptops, this will not show in the Order-based metrics. So check your Buy-to-Detail ratio, but always have a counter metric like Average Order Value.
See the green bubble in the following screenshot for example for a category that would get you a huge revenue gain if you just managed to improve the Orders / PDP Ratio a tiny bit:
Average Order Value is not available in Product and Category Reports in Google Analytics, though the Average Price per Product is a good-enough to work with.
6. For Categories: Revenue per PDP Visit
Definition: Revenue for a Category divided by Visits with PDP Views of this Category
This is the type of Conversion Rate where not all orders are equal, but they are weighted depending on the revenue they brought in. Assuming you have more or less the same margin per category, this helps you focus on those categories where you are not only performing poor on orders but also on revenue (again think of cables vs. laptops).
Then, for deeper analysis, hop into Analysis Workspace (click the “try in Analysis Workspace” button above the report):
See the last line here for example. This category’s share of PDP Views is quite low (7.1% of all Visits see this Category), but the Cart/PDP ratio is decent, and if people have the product in their cart, they are not often held up on their way to the order (comparably good Orders/Cart Adds or Checkout). So the category’s products perform well and it might be worth some more ad spend to send more traffic there.
The first line instead shows the opposite: Lots of traffic share, but bad performance, especially on the way to the checkout (Orders / Cart Adds or Checkouts is dark red). This points to a high-competition category where people search elsewhere for the best prices and may need more intense conversion help than others.
As always in Analysis Workspace, you can of course easily drag and drop elements into the report, e.g. to compare different time frames (in this example: yesterday with the last week with the week before that and the last 4 weeks) or do as many dimensional breakdowns as you like, e.g. to narrow down the bad or good performance to different subcategories, marketing channels, devices, or even segments.
In the example below, two Level 2 Categories suffered in terms of Cart-to-PDP Ratio last week. By drilling down to the Campaign Sources and Campaign Names, it can be seen that one channel (“AdWords” is just a fake example here) had a campaign that had started to send a lot of traffic to this category last week (PDP Views), but that traffic did not produce a single order (Orders / Cart Adds = 0%) and performed poorly in Cart-to-Detail Ratio.
To finish this off, let me use this opportunity to tell Google and Adobe what I miss the most when analyzing product performance at the moment:
Adobe: Please allow metric filters (I heard it is on the roadmap, but do make it prio1!) so I do not need to export stuff into Excel to filter out only products that got a sensible share of the traffic when sorting for low ratios for example. Also please offer better Dimension filters (RegExp!) (not just “contains” and “does not contain”).