Sayu Labs

Google Shopping Rankings Strongly Correlated with Google Organic Rankings

Posted by Tony

by Dr Gerda Arts, Head Statistician, Sayu Limited

When searching on Google Shopping and results are shown by relevance (which is the default), what does relevance actually mean?

If you think the fields in the Google Merchant feed determine how relevant the product is to the search, you could be wrong!

Although the product/category and other fields in the Google Merchant feed could play a role, it seems there is more to it than that if you want to achieve high rankings in the Google Shopping results.   A google employee pointed out that title and description tags on the landing page could affect the ranking, with better ranking being achieved if those tags are also relevant to the product searched for.

If that can have an affect on the ranking, are there any other factors outside the Google Merchant feed that are affecting the ranking? In particular, could it be that any factors affecting Google Organic rankings are also a factor in the Google Shopping ranking?

We investigated the relationship between Google Organic and Shopping rankings, recording positions for around 2,500 product related searches on both Google Shopping and Google Search. We recorded the pages ranking in the top 20 on Google Shopping, and in Google Search we recorded all pages in the top 100. There were 7,242 pages showing in the top 20 on Google shopping as well as in the top 100 on Google Organic.

For the pages showing in both results, we found a highly significant correlation between the two rankings.  Spearmans correlation coefficient (see footnote) is 0.62.  This is quite a strong correlation, implying that there is a lot of overlap between rankings on Google Search and rankings on Google shopping.

In other words, pages that rank well on Google Organic are very likely to rank well on Google Shopping as well. Figure 1 illustrates the relation between the two rankings.

Shopping Results v Search RankingsFigure 1: Relation between ranking on Google Shopping and Google Organic Search.

Such a high correlation seems to suggest that a lot of the factors affecting Google Organic Search rankings also affect the Google Shopping rankings.  Even more reason to think about doing SEO work on your site, as you will not just benefit on Google Organic Search, but also on Google Shopping.

Footnote: Spearmans correlation can be between -1 and 1: 0 if there is no correlation between the measures, 1 if the ordering of the two measures is exactly the same, -1 if the orderings of the two measures are exact opposites.
Filed under: SEO Research No Comments

Brands Win Following More Panda Updates

Posted by Tony

The first Panda update in April has been followed up by several more updates during the last couple of months.

Are these really updates with further changes to the Google ranking algorithm, or are they perhaps more of the same?

We looked at what had changed after the first Panda update ( and as mentioned by many, one of the factors we found to have affected rankings is the use of the same anchor text in a large percentage of backlinks.

Having checked positions of the same sites on the same keywords, we have seen the effect has become even more significant. This is mainly due to the fact that more sites have been affected. The sites already affected haven't seen a further drop in position.

Of the 1081 pages that used extreme large percentage of backlinks with the same anchor text, 243 (22%) saw a drop in position by at least 10 on 30th April. By the 15th June 515 pages (48%) had seen a drop of at least 10 positions.

So rather than a further change to how sites are penalised when overusing the same anchor text, it appears that just more and more sites have been discovered by Google overusing the same anchor text and have been penalised accordingly.

In our previous article we looked at bed related anchor texts (which could be either the keyword itself or other bed related anchor texts), and last time we didn't find any evidence that there had been a change in the way backlinks with bed related anchor texts affected rankings. Having updated the positions, we have now found some weak evidence of a change. It appears that using bed related anchor texts has become less effective.

Figure 1 gives the average position versus the number of backlinks with anchor texts containing either bed or beds. (Black - 19th April, Green - 30th April, Red - 15th June).

On all dates we can see that the position improves as more backlinks with bed(s) anchor text increases up to around 200-250. Using any more than that has no further positive effect on the position. The difference between mid April and mid June is that the improvement in the position is not as big any more. Using around 200 anchor texts with bed related keywords had a positive effect in April, with on average positions 10 higher compared to sites using very few bed related anchor texts. In June however, the improvement in position is only around 5 (from 30 to 25). Moreover, we can see a bigger drop in the position when a large number of bed related anchor texts is used. While in mid April sites with 1,000 bed related anchor texts were still better off than those sites that use few bed related anchor texts, in mid June you could end up worse using many bed related keywords rather than using none at all.

Average Position vs. No. of Backlinks


Figure 1: Average Position versus number of backlinks with bed related anchor texts

We also looked at the effect of anchor texts equal to stemmed versions of the keywords. We found that backlinks with these type of anchor texts had become less effective. Looking at June data we found this again, and the statistical evidence is even stronger. The sites that were effected in April haven't seen any further drop in traffic, but more sites have been affected in June. In April 22% of sites were affected, which increased to 43% in June. So rather than any further changes to google's algorithm, it appears that more sites have been affected by the change, and the Panda updates do seem to be just more of the same.

Many sites have been affected by the panda updates seeing a drop in their position, either soon after the first “update” or at a later date. Among the pages we looked at, 40% have seen a drop in the position of 10 places or more when comparing 15th June with 19th April.

So if there are so many losers, there must be winners as well. Looking at about 100 bed related keywords, we have seen an improvement in position for many prominent sites like Ikea, E-bay and Asda, with Ikea having seen the biggest improvement in position (see Figure 2).

Some sites (Next, Oakfurnitureland, TJHughes and Littlewoods) were not in the top 100 for the keywords in April, but have moved into the top 100 in June. So while sites heavily relying on backlinks to get a good ranking have suffered, sites that have a strong brand name and probably have built up backlinks naturally, have benefited, having moved up due to other sites dropping.

Figure 2: Sites with Substantially Improved Positions

Average Positions of Major Brands

Filed under: SEO Research No Comments