Last October, Google announced that it would make secure search (SSL) the default setting for all users logged in to their Google accounts. With secure search, a person’s search terms cannot be tracked by the third-party technologies that marketers have come to rely on to determine how natural search terms drive actions within websites.
This shift to an encrypted search model drastically impacts marketers’ ability to see details about the origin of natural traffic at the keyword level. As a consequence, tools such as Google Analytics, Coremetrics and Omniture are losing the ability to track the performance of some keywords, thus losing a share of data previously used to inform SEO strategy.
At the time of the announcement, Google estimated that encrypted searches would comprise less than 10 percent of all organic search queries. However, a 360i analysis found that this figure is larger than the engine anticipated. Our study of searches across a range of industries and browsers demonstrates that nearly 20 percent of Google searches are becoming untraceable by marketers. By 2013, this number could rise to 30 percent as Mozilla will soon make secure search its default setting for Google users.
This report assesses the impact of Google’s encrypted searches and provides recommendations for how marketers can navigate the change. The findings are informed by 360i’s analysis of natural traffic for several retail and service companies monitored through IgnitionOne, a leading digital marketing suite.
High Level Findings
360i analyzed the effect of Google’s announcement across a sample of marketers. We found that by making secure search the default for all logged in users, Google has created a data loss of nearly 20 percent. This means that marketers were unable to track a substantial portion of natural search traffic driving actions within their sites.
Our analysis showed that the impact varied by industry and browser:
- Industry trends: Overall, websites for service-oriented businesses (i.e. tax preparation services and insurance agencies) were impacted more than retailer websites.
- Browser trends: The change impacted Google searches occurring within Chrome the most. Before October, Chrome encrypted just 0.01 percent of searches on average; after, more than 30 percent of the keyword information was hidden to tracking tools. For searches conducted within Safari and Internet Explorer, about 10 percent of the data was no longer traceable following the update.
Several months after the announcement, we witnessed a plateau in the overall share of Google encrypted searches across retail and service industries – which have since stabilized. Today, close to 20 percent of keyword information remains hidden, with service domains continuing to collect the most encrypted queries.
We predict that close to 30 percent of all Google natural search traffic will be hidden across retail and service domains after the default search encryption is released to Mozilla Firefox users. This is expected to happen by the end of 2012.
Our analysis contains three phases spanning 270 days. The "Before" phase spans 90 days prior to Google's Oct. 18 announcement; the "During" phase covers 90 days immediately following Oct.18 and the "Post" phase looks at 90 days after the "During" phase.
To inform this report, we researched Google natural traffic and encrypted referrals by industry (retail and service). In addition, we studied Google encrypted search queries across top browsers such as Chrome, Firefox, Internet Explorer and Safari.
Before Google defaulted to secure searches, just 0.1 percent of natural search traffic across service-oriented and retailer websites was secure, or not trackable by marketers. That figure soared to 13 percent across the Google announcement period in October –3 percent higher than Google’s 10 percent estimate.
Overall, Safari and Internet Explorer were the least affected browsers. About 30 percent of total Google natural traffic analyzed during the 270 days came from Safari, which remains the least impacted browser to date.
The impact across service and retail industries varied during the announcement phase. Secure searches for service sites soared to 18 percent on average during this window. Before the announcement, secure searches were 0.05 percent on average for service websites.
Prior to the announcement, secure searches for retail domains were 0.07 percent on average. During the 90 days following the announcement, this figure jumped to nearly 12 percent – less than the average for service domains but still 2 percent greater than the Google estimate.
Approximately 15 percent of Google Revenue and Orders for one of the analyzed Retail domains were encrypted during the announcement period. After the announcement period, Revenue and Orders attributed to encrypted keywords increased to 20 percent. For retailers that rely on this data to inform their SEO strategy, the inability to track a fifth of natural keyword referrals creates a very significant challenge.
For example, in examining one retail client, we noted that its top 100 non-brand, non-seasonal keywords saw a 5.5 percent drop in Google natural traffic following the change. Although the shift in search patterns between encrypted and unencrypted searchers is still manageable, additional monitoring is required to establish the threshold where unencrypted Google keyword information will become insignificant.
Impact by Browser
Below is a snapshot of the percent share of total encrypted searches by browser during the 270 day research period.
- Chrome - 34.8%
- Firefox - 21.7%
- Internet Explorer - 7.9%
- Safari - 1.7%
As expected, major browsers saw an increase in encrypted searches during the announcement window, with Chrome surpassing the rest with nearly 32 percent of its searches hidden on average across retail and service domains. Chrome had the lowest number before the change (less than 0.001 percent). This stark increase can be attributed to the fact that the browser is a Google product.
Post-announcement share of encrypted searches increased across Chrome, Firefox and Internet Explorer while Safari encrypted searches dropped by 4 percent. Of the four browsers, Safari’s share remained below 1 percent on average before and after the announcement. During the announcement, the share of Safari encrypted searches rose to 4.25 percent on average.
Given the prevalent use of Firefox and its intent to test default Google search encryption, we predict that the share of Google Secure Searches on Firefox will rise considerably over the next few months – with most of the impact occurring across retail domains.
Next Steps for Marketers
Since keyword research is essential to any SEO program, monitoring which keywords are generating visits to a site remains a critical exercise. In light of this trend in secure search, it becomes increasingly important for marketers to accurately report on top traffic-driving keywords, particularly Google non-brand terms and top converters. Below are some strategies for marketers to consider when compensating for hidden keywords caused by secure search:
Run paid and natural campaigns simultaneously. Consider paid performance and conduct long-tail keyword ad tests to gain insights that will power your natural optimization efforts. In analyzing natural visits, terms that historically maintained a consistent flow of traffic to a retail site had a significant increase in encrypted information during and after the announcement as traffic from keywords labeled “Search Phrase Not Provided” spiked. For example, non-seasonal curtain variants such as “curtain rods”, “door panel curtains,” “sheer curtain panels” and “beaded curtains” reportedly drove less site traffic following the announcement period, as illustrated in the table below. This was due to a widespread data loss across all keywords driving to that domain from within Google Search.
In this example, a deep dive into the domain’s paid search performance gave us insight into additional keyword variations used to gain access to the site through paid search during the same period. Popular exact match variants for the term “sheer curtain panels” in paid search include a trimmed down variation: “curtains sheers”; and detailed variants such as “semi sheer curtains” and “yellow sheer curtains,” as illustrated below.
Marketers should therefore run paid search (AdWords) and SEO campaigns simultaneously, tracking keyword attribution to discover additional keyword variations that will not only deliver traffic to the site, but convert as well.
Use Webmaster Tools to access reports about your pages’ visibility within Google. Monitor the top 1,000 search queries that drive traffic (Clicks or Impressions) from Google Webmaster Tools and use them for lead generation. Grouping terms into Broad (one keyword phrase), Torso (two to three keyword phrases) or Long-tail (four or more keyword phrases) buckets will help identify the most popular variants, especially in non-brand, long-tail queries. Use the most keyword-dense variants within the top queries to expand on keyword variations.
- Supplement third-party analytics tools with additional data. Use Hitwise, comScore and similar tools to monitor competitor or industry performance. Use additional resources to explore both organic and paid search trends for your site to continuously identify the most opportune words or keyword phrases that you might be tracking in your campaign. Some of these tools are listed below.
- Competitive analysis tools, such as Google Insights for Search and AdGooroo, can be used to compare search patterns across different regions, categories, time frames, properties and competitor keywords.
- Quantitative market research tools, such as Google Consumer Surveys, can be used to glean additional market insights.
- Search demographics can uncover usability patterns across audiences and identify high value keywords that will attract qualified traffic to your site.
Consider search beyond the PC. Mobile searchers have different behaviors and motivations than PC users since they are normally on the go. By next December, smart mobile devices will account for 25 percent of all paid search ad clicks within Google (source: Marin). The ever-growing adoption of smart phones, tablets and other connected devices means that non-PC traffic now comprises significant share of overall searches.
Mobile long-tail searches can provide valuable intelligence into what searchers are looking for, such as intent, desires and needs beyond simply information pertaining to location. Moreover, mobile users are more prone to clicking paid ads, as those appear closer to the top due to the small screen size of mobile devices. Marketers can leverage these insights to power the keyword discovery process, as well.