Just to refresh our minds, coverage, or recall, measures the quality of product tagging. It tells us how many tags were correctly added to products, out of all available and relevant tags. In contrast, precision measures how many of the tags that are applied to a certain product are correct. It is important to measure coverage because it tells us how well buyers can slice through thousands of products and narrow their focus down to a much more manageable number.
We wanted to measure the navigation coverage on Ebay in the same categories we measured precision, so we went back to Ebay.com, Ebay.co.uk and Ebay.de, and reviewed the same categories as before: Men’s T-Shirts, Men’s Sneakers, Men’s Watches, Digital Cameras and Fragrances. This time, however, we looked at how many products are there in each category, and then selected an attribute which values are unique and cannot be applied together. For example, in the Men’s Watches category, we selected the Case Size attribute, because no watch has multiple case sizes at the same time. However, a watch may have several colors, or fit many styles, and therefore can be tagged with multiple values from the Color and Style attributes.
After selecting all the values, we measured the number of results we were left with. In this situation, ideally, we should get back the same number of results as without checking the attribute, for a coverage level of 100%. However, as we will see, this is hardly the case.
On a side note, all three Ebay sites offer in each of their attributes the value “Not Specified”, which in some attributes, covers the majority of products. But since nobody ever selects this value, we didn’t either.
The Men’s T-Shirts category in Ebay.com had over 7 million products. After we went to the Sleeve Length attribute and selected all lengths possible (which, presumably should cover all the products in the category), we received back about 5.5 million results, for a coverage level of 78%. This means that in this category, 78% of the products were correctly tagged for the Sleeve Length attribute.
The Men’s Sneakers category produced almost 4 million results. Since in this category there aren’t many unique values, our choice was limited. We selected all the values in the Shoe Shaft Style, because a sneaker cannot be both high and low cut, and received a little less than 700,000 products back. This puts coverage level for this attribute at a low 17%.
The Men’s Watches category looked considerably better. Out of the total 390,000 products, 272,000 were left after we selected all types of displays (analog, digital and combined), for a coverage level of 69%.
123,000 products came up in the Digital Camera category in Ebay.com. Out of those, only less than 48,000 remained after selecting all values in the Battery Type attribute, for 38% coverage. Once more, the majority of products in this category are not tagged for this important attribute.
As mentioned in the previous post, the Fragrances category is a tricky one. While the precision levels in this category are surprisingly high, the coverage level is quite low. Initially, the category showed a total of almost 1.2 million products. But after selecting all the values from the Size Type attribute (regular, travel size, etc.), the coverage level reached only 10%.
In the Men’s T-Shirts category, selecting all the values in the Neckline attribute (because a t-shirt cannot have two types of necklines), brings us from a total of 16 million results to 11 million results, for 69% coverage. A little behind Ebay.com.
Ebay.co.uk’s Men’s Sneakers category contains more than 800,000 products in total. This number diminished to only 237,000, a coverage level of only 28%, after we selected all the values in the Shoe Shaft Style again. Ebay.co.uk displayed better results than Ebay.com.
The Men’s Watches category keeps up the same performance as Sneakers, for a coverage level of 34%. Out of the total 867,000 products in the category, only 294,000 were shown after we refined the results using all the values in the Case Size attribute (which covers every size possible). This means two thirds of the products in the category are not tagged for the Case Size attribute.
The results improve in the Digital Cameras category, which offers almost 34,000 products in total. Out of those, more than 20,000 remain after using the Maximum Resolution attribute to narrow the results. This puts this attribute at 61.5% coverage level.
As we said, Fragrances is an unexpected category. Compared to the Ebay.com, Ebay.co.uk puts up quite a performance. Out of 188,000 products, 159,000 come back after we selected all types of Volumes. At 84.5%, this is the highest coverage level in our review.
Ebay.de’s Men’s T-Shirts category came in last among the Ebay’s. Almost 12 million products were on sale in total, but only 3.2 million remained after using the Sleeve Length attribute to refine the results. This put this attribute at a 27.5% coverage.
The Men’s Sneakers also did not measure well, but it was not the worst across different sites. At a coverage level of 21%, 590,000 products in the category were reduced to just 125,000, when we narrowed down the results to show all types of Shaft Style.
The Men’s Watches category contained in total 154,000 products. We selected all the values the Case Shape had to offer, and got back 86,000 products, for a 56% coverage level.
Selecting all Maximum Resolution values in the Digital Camera category, brought the number of results down from a total of 26,500, to 7,045, which puts it at 26.5% coverage.
By now we know it is hard to predict the results for the Fragrances category. Of the total 95,000 results in this category, 68,000 remained after selecting all the values from the Volume attribute, for a coverage level of 71.8%. Not the best, but far from the worst.
We can see a great variance in navigation coverage between the different categories, and between the different platforms. In contrast to navigation precision, however, there is no correlation navigation coverage on Ebay and whether or not a category is structured. That is why the seller is responsible for tagging his products, and for better coverage - tag as many of them as you can, with as many relevant tags you can. By helping customers find what they are looking for easily and quickly, tagging will help you improve conversion rate and overall sales.