Cases of Product Feed Optimization

Cases of Product Feed Optimization

Thomas Byskov Madsen & Lasse Rasmussen - January 8, 2018

Case 1: Adding a Generic Synonym

By using the comparative analysis method described above on a client’s AdWords account, QuantAds discovered that one of two large search term synonyms from the standard Keyword campaigns appeared significantly more than the other among the Shopping search terms. After reviewing the product titles of the relevant products, it was found that the missing synonym (hereafter referred to as ‘Synonym 1’) was not used at all in the product titles. The titles of these relevant products were thus changed so that they included both variations, as follows: “Synonym 1/Synonym 2”. Changes like these are carried out via our feed management tool, as Google’s own Merchant Center Rules do not currently support changes of this type.

Due to the product type, the search volume of the two particular synonyms was highly influenced by seasonality. Therefore, the effect of adding Synonym 1 to the titles was identified by comparing the amount of impressions both synonyms received YoY as an index over 34 weeks.


Impressions Index


As the account was generally performing significantly better than the previous year, both synonyms had very high impression indexes throughout most of the period. But while Synonym 2 peaked at an index of 910, Synonym 1 had an index of +2,000 consistently during the last 9 weeks, peaking at index 6,045. The dramatic increase in Synonym 1’s impression index shows the significant effect of optimizing the product’s titles. Furthermore, the rather stable impression index of Synonym 2 shows that optimizing against a second synonym does not (negatively) impact the original synonym used in the product title.

As will become evident from case 2, adding synonyms to the product’s title (as opposed to the product’s description or category) is essential if we are to reap the positive effects of Shopping ad impressions on a broader selection of relevant search terms. The product title field remains the single most important place in the feed to focus optimization efforts.

Case 2: Adding a Sub-Brand

In a second analysis, the data-driven approach revealed that a specific sub-brand (sold by the retailer), which was one of the largest impression drivers in the search term report of the Search campaigns, was almost non-existent in the Shopping campaigns. Again, it was a matter of the brand not being present in the product titles. In this case, however, the brand was in fact placed in the feed attribute ‘Manufacturer’. This indicates that this attribute field is not weighted heavily by Google when matching search terms with products. By adding the specific sub-brand to the product titles, impressions on search terms, including this one, increased 582% in the subsequent period as opposed to the previous one.

Case 3: Adding a Retailer Brand

In a third analysis, we identified that the parent brand of the retail client was almost non-existent in the Shopping campaign search terms. Search terms that included the brand represented just 0.97% on average of the Shopping campaign’s total impressions. By adding the retailer’s brand to the end of all product titles, impressions on search terms, including the brand, increased to a share of 2.51%. Although this remains a small share of the total impressions, it represents an increase of 158%.


Impressions index-retailer brand


Some will argue that this kind of optimization just simply results in a ‘cannabilization’ of the brand’s text ad results, thereby adding no real growth. You have to remember, however, that the brand of this retailer is also searched for in conjunction with generic terms, and even though your own brand might be in a much stronger competitive position than the other advertisers, brand-generic searches do also trigger competitors’ ads. Shopping results are extremely price sensitive, so a potential customer could end up seeing a competitor’s ad with the same product at a lower price and purchase it from them instead. This represents a strong argument for the implementation of any strategy that will reduce this risk.

From a rather simple feed optimization, the retailer from this case is now much more present on brand-generic searches as a result of having both text ad results and multiple Shopping ad results.