APRIL 28, 2016 STUDY ON THE PERFORMANCE IMPACT ON PAID SEARCH RESULTS DUE TO THE ELIMINATION OF ADS ON THE RIGHT HAND SIDE OF THE GOOGLE SERP A NETELIXIR UNIVERSITY WHITEPAPER
Table of Contents 1. Executive Summary 2. Introduction 3. Methodology 4. Results / Insights for 4.1. Hypotheses on CPC inflation 4.2. Hypotheses on auction participation 4.3. Hypothesis on fourth position ad 4.4. Hypotheses on share of traffic 1
Executive Summary Background The recent change in Google desktop SERP (Search engine results page) related to the elimination of right-hand-side text ads and showing four top ads has spurred lots of speculations in the PPC industry. By removing the right hand side ad units, Google basically reduced the real estate area for search engine advertisers in order to provide a unified SERP experience across devices. However, one point to note is that this change has affected only the queries with a commercial intent. The above change will have serious implications on the retail PPC advertising. Though the NetElixir retail insights team welcomed the design change with enthusiasm, they have anticipated certain performance signals. These include: CPC (cost-per-click) inflation for keywords, overall reduced auction participation and others. To validate our assumptions, we carried out a study evaluating the performance impact of the change in layout. We started with a typical inferential statistical approach of formulating hypotheses and validating them with data. Data was collected on only PC users searching via Google from Jan 1 st till April 15 th. The data, comprised of 12 million impressions, belongs to four different verticals in the retail industry. Summary of findings Affected Metric Results CPC (Cost per Click) CPC has not inflated for text-ads while it has happened for PLAs. CTR (Click through rate) The CTR for the fourth position text ad has increased. Auction Participation The number of auction participants has decreased. Percentage Share of traffic from The share of traffic from PPC has not decreased, rather PPC marginally increased. Share of traffic from The share of traffic from organic listings has decreased. Organic Listings We have shared our conclusions and points of view around the findings in the next page. 2
Conclusions & Points of View 1. The cost-per-click for keywords used to purchase text ads has not increased, in contrary to what we hypothesized. However there is a marginal inflation in the CPC for Product Listing ads. This proves that search advertisers will have to marginally raise their budgets to adapt to the new layout. (It is pure economics. Supply comes down; Demand remains the same. A price increase is bound to happen.) 2. The number of advertisers participating in an auction has decreased. This indicates that many advertisers could not enter the auction since there are less number of ad slots placed for bidding. Needless to say, the above information should not be translated into an idea of decreased competition as the level of bidding has not really changed. However it signals the withdrawal of the advertisers at the lowest end of the spectrum who would be mostly SMEs. 3. Since the fourth position ad is introduced in the top of the SERP, it has attracted better visibility, hence higher click-through-rate. This ad slot also allows extensions which is why it is imperative to include them for all top-performing keywords. 4. The share of traffic for PPC has not been effected. We find this to be quite logical as the additional top ad and better visibility for PLAs has offset the lost impressions and clicks from the ads from the eliminated slots. In fact, we observe that the share of traffic from PPC in an increased state. 5. The share of traffic from organic listings has marginally decreased. We are not concluding on this as organic is a channel difficult to hypothesize. However, we believe, a decline in the organic traffic is a plausible happening as currently there are four text ads in the top of the SERP and organic listings have been pushed down below the fold. 3
Introduction In GOD we trust; all others must bring data W. Edwards Deming Around the middle of February 2016, Google made a major change in the layout of the desktop SERP (search engine results page) by removing the PPC (pay-per-click) text ads appearing on the right hand side of the SERP. This layout change is mainly for the search queries with commercial intent. This was not just a cosmetic change but a strategic one where Google wanted a unified design across devices. This update encapsulated the following changes: Desktop and Tablet can show up to a max of four top ads Right-hand side (RHS) text ads removed (except for Product Listing Ads - PLAs) on desktop Google Search; Bottom ads remain Reintroduction of two-line site-links and deprecation of non-brand 2-expanded site-links on desktop The changes in the layout is not an entirely new concept. The experiments around it have been an ongoing process and the SERP has been evolving to create the optimal user experience. (Check this Search Engine Land blog in 2010 on how Google has come up with four top ads.) Being search purists and believers in the power of data, the NetElixir Retail Insights Lab initiated a data-study to understand the impact on performance of the above changes. We did collective brainstorming to arrive upon certain hypotheses around the performance. Later, just as in a typical inferential statistical study, we validated our hypotheses. There are instances where our hypotheses went wrong, standing against the logical presumptions. We have accepted such cases since those were what the data suggested. Note: In the next page, we have showcased one screenshot which explains the new SERP layout. 4
Old Layout: New Layout: NO TEXT ADS IN RHS 5
Methodology The project consisted of two phases: Framing of hypotheses Data collection & validation Hypotheses We formulated six different hypotheses which could be the possible implications of the change in the layout of the Google SERP. These are scenarios that may have happened after the middle of February around when Google rolled out the change in the layout. There are two hypotheses on CPC inflation, one on auction participation, another one on the CTR of fourth position ad and two on the share of traffic from PPC and Organic listings. The hypotheses are: H1: The CPC (Cost-per-Click) has inflated for the text ads. H2: The CPC has inflated for the Product Listing ads. H3: The auction participation has decreased. H4: The CTR for the fourth position ad has increased. H5: The share of traffic from PPC has been affected. H6: The share of traffic from Organic listings has marginally decreased. Data Collection & Validation Data collection had the following characteristics. Data was collected only for PC users searching via Google from Jan 1 st till April 15 th. The data comprises of 12 million impressions. The data belongs to advertisers of four different verticals. Data validation was done in a typical inferential statistical method where we analyzed the data and checked whether the hypothesized trend is visible or not. Mostly it is a comparison between the trend before the layout change and after that. Please note that we have either accepted or rejected the above hypotheses while we do not claim any other outside trend to be true or false. 6
Results & Insights Hypotheses on CPC inflation H1: The CPC (Cost-per-Click) has inflated for text ads. H2: The CPC has inflated for Product Listing ads. Hypothesis Rationale: As the real estate area (supply) for the PPC ads decreases while the advertising incentives (demand) remain the same, the cost per click is bound to increase. Data Graph: Result: The graph above suggests that the CPC has marginally declined or in some cases remained stagnant for keywords used to purchase text ads. For the product listing ads, the CPC has gone up marginally. So, based on the above observations, H1 is rejected while H2 is accepted. Insights: The CPC incline, as expected, for the keywords used to purchase text ads has not really happened. However CPC has marginally risen for Product Listing ads. From a very high level, this data tells us that there needs to be a marginal increase in the advertising spend to adapt to the new layout. 7
Hypothesis on auction participation H3: The auction participation has decreased. Hypothesis Rationale: The number of ad-slots (supply) up for an auction decreases. This leads to less number of advertisers qualifying for an auction. Naturally the overall level of auction participation will decrease. Data Graph: Note: The number of auction participants in the first period is taken as a baseline. Results: As the above graphs suggest, the number of auction participants decreased, starting from late Feb. However B2B clients may see an exception as the number of competitors are relatively less. Anyhow, based on the above observations, H3 is accepted. Insights: The reduced level of auction participation does not really indicate a lesser competition. It is just that those advertisers who were bidding less in the earlier auctions are not even allowed to enter the auction. 8
1/Jan 8/Jan 15/Jan 22/Jan 29/Jan 5/Feb 12/Feb 19/Feb 26/Feb 4/Mar 11/Mar 18/Mar 25/Mar 1/Apr 8/Apr 15/Apr NetElixir Retail Insights Whitepaper Hypothesis on the CTR of fourth position text ad H4: The CTR for the fourth position ad has increased. Hypothesis Rationale: The text ad in the fourth position was previously appearing on the right-hand-side. Now it is on the top position, sometimes with the site-links as well. This should enhance the CTR Data Graph: 2.50% 4th Pos CTR 2.00% 1.50% 1.00% 0.50% 0.00% Results: As the graph suggests, the CTR for the fourth position has improved. The hypothesis, H4 is accepted. Insights: Now there are four top ads which can be leveraged in terms of a higher CTR. The top slot ads have been the best performing by driving almost 89% of the PPC traffic. With one more in that section, PPC traffic is benefitting. It also becomes imperative to include all possible extensions for the best performing keywords as the chances to be in the top is higher now. 9
Hypotheses on share of traffic H5: The share of traffic from PPC has not been effected. H6: The share of traffic from Organic listings has marginally decreased. Hypothesis Rationale: Historically, the top ads have been driving almost close to 89% of the traffic. With an additional top ad in place as well as PLAs either in the top or right, we estimated that the PPC traffic is not going to be affected. Also, now the extensions can show up in four ads versus three in the earlier case. The PPC traffic is going to remain steady, if not increase. With four text ads in the top, the organic listings get pushed down. This may lead to a marginal reduction in traffic from organic listings. Data Graph: Results: As the graph suggests, the traffic from PLAs have increased and that from organic listings has decreased. H5 and H6 are accepted. Insights: This finding will have serious implications on budgeting. PLAs have always remained the dark horse of PPC and the above results underscores that statement. Increased focus in terms of budgeting may be required in PLAs as it would to lead to more clicks and conversions. This is more evident in one of our retail accounts where the share of e-commerce revenue from PLAs is about 35% after the change contrary to 28% before the change. The traffic from organic listings has marginally come down for these selected advertisers. We cannot conclude on this as it is a small decline and it may require repeated tests on multiple data-sets to validate. 10
About NetElixir Global search marketing solutions firm headquartered in Princeton, NJ. Our purpose is to help businesses profitably acquire customers online through a combination of technology and service solutions. www.netelixir.com 11