Based on underlying comScore data, the report analyzes visits to the top five websites in several verticals: Media, Retail, Travel, Auto, Finance and Health. It doesn’t discuss the impact or relative position of Facebook at all, possibly because Facebook is not yet a public company.

Mahaney points out that Google’s retention of the position of top traffic referrer is an accomplishment in a marketplace that is so “fluid” and intensely competitive.

The report has a number of key findings, which I’ve distilled here:

Google ranked as the #1 source of traffic for 23 of the top 30 Websites. “By contrast, the #2 (Yahoo!) and #3 (Microsoft) influencers accounted for 11 percent and 6 percent of traffic to the top 30 Websites.”

Google’s “influence is [very slowly] slipping”: Mahaney said that “Google has gone from delivering 17 percent of all the traffic to the top 30 Websites in 2010 and 2011 to 16 percent in 2012.” The one vertical studied in which Google’s influence has waned considerably is Heath (see chart above).

Yahoo and Microsoft’s traffic is flat: the report said that “Yahoo!’s and Microsoft’s share of traffic “has remained essentially flat from 2010 to 2012.” This is regarded as a surprise on both counts.

Mahaney said that flat traffic is positive for Yahoo given the market’s perception of the company/site as a “deteriorating asset.” By contrast this is “arguably something of a negative surprise [for Microsoft] . . . given signs of Bing Search market share gains.”

For the iPhone-app study, just published yesterday, Mediative monitored eye- and click-tracking with 12 people in Canada. They ranged in age from 21 to 45. The participants were asked to find places to get a tattoo in each of several Canadian cities. Here’s the task as described by Mediative:

Every participant in the study was given the same scenario as in the previous Google Places study – an imaginary road trip with stops in Hamilton, London, Winnipeg and Edmonton, with the task of choosing a place for a friend to get a tattoo in each of those cities based on the Google Places search results. From the participants in the eye tracking part of our study, we recorded each individual session, and then compiled the sessions to create heat maps that represent aggregate gaze data and supplement the qualitative findings.

Screen size was a critical difference between the PC and iPhone studies. Beyond this Mediative found that images and reviews were very important in capturing users’ eye movements and clicks.

The image on the left shows yellow bubbles of various sizes. The larger the size of the bubble, the more time the eye spent in that area of the screen. The numbers represent the progression of movement of the eye across the screen during the first several seconds. In other words the eye spent the most time on an image on the right, which was nearly the last thing considered in this sequence.

On the right screen above, red indicates more time spent and green less time spent. The second image in this case got the most attention among the three images. Mediative explains: “It’s the only image on this screen of a tattoo; both the other images are of storefronts.” Thus the image was directly relevant to the searcher’s objective: find a tattoo vendor.

As a general matter, people on the iPhone app scanned left to right and then down the page, not unlike eye-tracking patterns on the desktop. According to Mediative, reviews were an especially significant factor in eye-tracking patterns in the iPhone test:

People typically start looking in the upper left part of the screen, scan from left to right, then move down to the next result, and scan from left to right again. However, given the small space of the iPhone screen, some people will have their attention pulled to the right to look at an image, and may continue a scan down to the next image, before resuming a left-to-right scan pattern. If they start scrolling down, then their gaze will stay on the left side until they hit a listing of interest. In the examples we used in this study, in almost every case it meant that they would scroll down until they hit a listing with a better than 3-star review.

What Mediative observed is that where there were fewer “social signals” (e.g., reviews) among the top results users went further down the page to find places that had not only reviews but at least three stars. According to the report, “This presents an opportunity for businesses whose websites are not listed at the top; they can be more competitive by adding positive reviews.”

Mediative offered the following conclusions and recommendations:

• Reviews and other social signals grab attention: “If a business website is listed in any position other than the top three, and the listing does not include any social signals, it will be relatively ignored, especially if there are other listings that do have social signals.”
• Reviews drive clicks: “Positive reviews on the Google Places iPhone app were the biggest single factor we observed that determine which listings got clicks and which did not.”
• Volume of reviews matter:  Mediative found “that 29 of the 47 clicks went to listings that had at least four reviews.”
• Images matter too: “People . . . look at the images to see if the business looks trustworthy.” The type of image may matter as well: “We recommend giving the image that accompanies a listing some special consideration, and remember that an image that might work on a desktop might not work so well when reduced to a phone.”

You can download the entire report here (registration required).

Related Entries

• Eye-Tracking In Google Maps: Study Shows Value Of No. 1 Ranking & Social Content
• Eyetracking & SEO: Fad, Fact, Or Fiction?
• Eye-Tracking Study: Everybody Looks At Organic Listings, But Most Ignore Paid Ads On Right
• Survey: 6-10 Local Business Reviews Required For Trust
• Survey: Local Reviews Gaining In Importance
• Eye Tracking Study Shows Importance Of Search Snippets

Earlier this month, we released the RKG Digital Marketing Report with this in mind. Our goal is that it serves as a credible benchmark for advertisers who don’t enjoy the luxury of viewing results across multiple sites with multiple channels each. In our report, we offer our insights and data for paid search, SEO, and Facebook, as well as comparison shopping engines and multi-channel attribution.

We’ve covered the big trends, but we also tried to uncover some subtler ones, a few of which we think you’re unlikely to see anywhere else.

Generally, RKG’s Q4 results are directionally in line with the emerging consensus out there on the higher level metrics. We saw paid search spend growth accelerate in Q4 to a 31% year over year rate, up from 21% in Q3.

A significant increase in click-through rates was the primary driver, while cost-per-click actually fell 1.4% Y/Y. With revenue per click increasing modestly, our advertisers enjoyed an 8% increase in return on ad spend.

Paid search growth on Google was even more impressive, with ad spend up 39% Y/Y on a 46% increase in clicks. Click-through rates on Google were 26% higher than in the fourth quarter of 2010, speaking to Google’s efforts and ability to drive more and more users to the paid listings through appealing new ad formats and tweaks to existing ones. CPC fell 5% on Google though, and that may give some pause.

Increased mobile traffic, which has lower click costs, is certainly a contributor here, but we also have to consider what ultimately enables higher CPCs: higher revenues per click. We only see RPC increasing 2.6% Y/Y on Google in Q4 and that serves as a significant constraint on CPC.

Google’s Product Listing Ads format, which migrated to AdWords and shifted to a CPC model in the Fall of 2010, generated 8% of Google spend for the quarter and, for some advertisers, represented a larger traffic segment than Bing and Yahoo combined.

Even though we see Bing and Yahoo making some strides, our figures for them seem to be a little more pessimistic than those we’ve seen elsewhere. In Q4, our combined spend for the two engines was down 6.1% Y/Y, an improvement from an 8.7% Y/Y decline in Q3.

Our revenue per click was roughly 23% higher though, allowing CPCs to rise 12% while our advertisers took home a 10% higher ROAS. Click-through rates did improve on Bing and Yahoo, but not to the same extent as we saw on Google.

At least part of the problem for Bing and Yahoo is a continuing decline in traffic from its search partners. Last fourth quarter, Bing and Yahoo generated over 28% of their traffic from partners, but that has fallen steadily to 21% in Q4 2011. This helps advertiser revenue per click, but if adCenter is unable to deliver much additional ad inventory for subsequently raised bids, we see our ROAS increase, rather than volume.

It’s important to point out that we are now comparing to post-Search Alliance numbers and, given that the Alliance appeared to hurt Yahoo traffic at the time, we would expect a growth rate improvement for that reason alone.

Also, there may be some upside to these numbers as our analysis factored out brand keyword performance to limit the impact of extrinsic variables like offline media buys. Unfortunately for advertisers, we saw brand CPCs soar on Bing and Yahoo.

In the end, we see Google increasing its dominance it paid search, taking nearly 87% of clicks. On the organic side, we see Google with an 84% share, giving us all a couple more reasons to scratch our heads at comScore’s figures. Google is losing some of its advantage in CPCs commanded, but again, mobile is factor here and Bing and Yahoo have almost no mobile share to speak of.

Speaking of mobile, by the end of the quarter we saw traffic from smartphones and tablets generating over 14% of paid clicks, but for the entire quarter, mobile share was just under 10%. This was right in line with our organic search figures which also pegged mobile at 10%.

Tablets are the major driver here — even though the “mobile” moniker may be inappropriate for them — and we saw tablets overtake smartphones in December. The iPad remained dominant with a Google-esque 88% share of PPC tablet traffic, but the Kindle Fire was able to carve out a solid niche for itself with nearly 4% of tablet traffic after Christmas.

While Facebook garnered increased attention and budget dollars from marketers, it largely remained a growth opportunity given the sheer scale of its traffic and its grasp on its users’ time online.

For our SEO clients, Facebook generated a little under 4% of referral traffic in Q4. Those running Facebook ads found that they were able to generate an average of 90% of their unique impressions on Facebook from paid activity.

As we move deeper into 2012, the importance of social media, particularly to SEO, looks like it will only grow by leaps and bounds. Google’s Search Plus update may have been as controversial as it was significant, but don’t expect Google to back down on this one, and if they can finally play nice with Facebook and Twitter, things will really get interesting.

On the paid search side, we are seeing the major Q4 trends continuing into the new year with strong spend growth fueled by CTRs giving advertisers a solid return on their investment.

The ad network analyzed search referrals on “hundreds of millions” of impressions across sites in its network between January 9th and 12th. And the longest search referrals — at an average of 4.81 words — came from Ask.com. AOL users are at the other end of the spectrum; their user queries average barely above four words, by far the shortest of the five sites that Chitika studied.

It makes sense that this would be the case, since Ask.com has been refocusing on questions and answers in recent years — and asking questions tends to involve more words than other types of queries.

It would be great to get real search query length data directly from the search engines, but they’ve never made a habit of sharing that kind of information. In May 2010, Google did release data showing that 54.5 percent of queries are more than three words — which falls in line with the Google (and Bing) data that Chitika shows above. But that’s the last time I’m aware of that either Google or Bing shared such information.

IgnitionOne said that among its retail clients, “Mobile search [  ] accounted for 24% of retailers’ total paid search budgets on Black Friday, compared to 14.2% of the total during all of Q4. This represents enormous growth compared to Q4 2010, when mobile search spend accounted for only 5.2% of total retail paid search spend.”

The following chart from IgnitionOne shows the growth of mobile impressions, clicks and search ad spend vs. 2010.

The Marin Software report says the company is seeing increased adoption of mobile and tablet advertising among its clients. More significantly, however, its clients’ campaigns on mobile are “showing significantly better performance than similar campaigns on desktop computers.”

Marin published comparisons for its aggregated search campaigns across PC and mobile. As the chart below indicates, paid search ads on smartphones and tablets outperformed those on the PC in terms of CTRs.

The results of this research present a good news/bad news scenario for search marketers. The good news: If you have a credible, trusted domain name, you’ve got an advantage, as searchers really do pay attention to the URL in search results before deciding to click. And this is true regardless of the position of the URL on a search result page.

The bad news, of course, is that it’s more difficult these days to acquire “credible” domains now that most single or even double word domains are in use or reserved. Add confounding factors such as personalization, Google changing its core algorithm more than 500 times a year, and the fact that most searchers don’t move beyond the first or second page of results and you’ve got a major headache for most SEOs.

Nonetheless, the study is worth a close read for anyone wanting to understand more about how to capture the attention and clicks of searchers, thanks to its wealth of data generated by observing real people and their search behavior. Probably the most significant conclusion from the study:

Surprisingly, we find that despite changes in the overall distribution of surfaced domains, there has not been a comparable shift in the distribution of clicked domains. Users seem to have learned the landscape of the internet and their click behavior has thus become more predictable over time.

In other words, even if search result rankings change due to factors like personalization or algorithmic tweaks, searchers don’t seem to care. They’re demonstrating a clear preference now for credibility and trustworthiness in a domain name now over simple ranking on a search result page. This is the strongest evidence yet that I’ve seen that an obsession with ranking is not only futile, it completely ignores the reality of how your site attracts users.

Key takeaway for bosses/clients: rank really doesn’t matter, if you’ve got a quality (trustworthy) domain name.

The study also has merit for anyone doing paid search, and considering what display URL is most appropriate for an ad. While advertisers are always limited to a display URL that corresponds with a top-level domain, the additional keywords shown in the display URL may be crucial in getting searchers to click. Also, even if searchers don’t have favorable “domain bias” for your main site, it may be possible to secure another more favorably-perceived domain for your paid search campaigns that serves as a microsite that ultimately funnels searchers into your main domain.

The report is thick with math and numerous citations to related work, but it well worth the effort for anyone involved in competitive search marketing.

Domain Bias in Web Search (pdf), Samuel Ieong, Nina Mishra, Eldar Sadikov & Li Zhang, Microsoft Research (Sadikov is now with the Department of Computer Science, Stanford University, but the study was done while the author was at Microsoft Research.

Bing gained 3% between September and October, going from 12.8% market share to 13.23%, while Yahoo also rose slightly from 15.27% to 15.39%. Combined, that boosted Bing-powered share among major search engines to 28.62% … still a far cry from Google’s estimated 65.38% of the US search market.

Meanwhile, more interesting to me is Hitwise’s take on search query lengths. In the same news release today, Hitwise estimates that 27.23% of searches were one word long in October. That number is up from 26.45% in September. Hitwise tells me this is desktop searches only — mobile search is not included. (Mobile is where you’d expect shorter queries to be more prevalent.)

Not much of a month-to-month gain, but a little over a year ago, I wrote about query length on my own blog and shared a Hitwise chart that covered searches in September 2010. Notice that, at the time, Hitwise estimated one-word searches to be only 22.72% of all queries.

So, over 13 months, Hitwise is saying that one-word search queries have risen from 22.72% to 27.23%. There are a lot of implications in that for business owners and marketers — implications that are best left for a separate article, or for the comments below if you prefer.

The company says it serves 10 billion monthly ad impressions. The data in the report were collected between July and September. Here are a few key points:

• 37 percent of general mobile consumers use their devices for local search, while 75 percent of smartphone owners conduct local searches (Google has reported even larger numbers)
• Local search growth on xAd’s network has seen triple digit increases this year (with faster growth concentrated in apps vs. the mobile browser)
• xAd sees 90 million local-search requests per month and over 250 million local-search requests per quarter
• 62 percent of users conduct at least two local searches per month, while nearly a third (32 percent) of users do at least five local lookups per month
• Tuesday and Wednesday are the most active days for local-mobile search, which peaks generally during the middle of the day

The top local search categories or lookups seen on xAd’s network in Q3 were the following:

1. Restaurants
2. Gas stations
3. Shopping
4. Auto repair/dealers/rentals
5. Fast food
6. Cafes/coffee shops
7. Travel & lodging
8. Health & medical
9. Bars & clubs
10. Finance & legal

The network also reports on engagement and ad performance data, both in terms of apps vs. browser and iPhone vs. Android. The report indicates that CTRs on ads in apps are 8 percent vs. 5 percent for browser.

Call-through rates, however, are higher on browser-based ads. This difference is attributed to feature phone users calling businesses rather than seeking more information on their handsets, which offer a more constrained user experience than smartphones.

The company reported that CTR rates on the iPhone and Android are roughly comparable, but that “secondary actions” are greater on iOS. These secondary actions can include calls, map/directions access, review lookups and so on. Calls are the most popular “post-search” secondary action (62 percent), followed by maps and directions lookups (35 percent).

The xAd report also discusses demographic data. As one might expect, the heaviest users of local-mobile search tend to mirror smartphone demographics generally: a slightly more male audience (55 percent), concentrated in the 25 to 44 age range, which is also more affluent ($75K+) and educated than the general mobile population.

Early last month, Mediative did eye tracking on 12 participants and online click tracking on another 90 people as they interacted with Google’s local search results as shown on Google Maps. (The number of test participants on the eye-tracking side seems low to me, but let’s soldier on.) Mediative had the entire group act as if they were on a four-stop trip across Canada, looking for a place to get tattoos in each city.

If I had to sum up the findings from this study, it’s that the famous “golden triangle” does exist when searchers scan Google’s local results, but it doesn’t necessarily apply when lower business listings have more content than top-ranked listings. (If you’re not familiar with the “golden triangle” concept, see the image in this article by Mediative’s Gord Hotchkiss.)

Golden Triangle In Google Maps Results

The first stage of the test involved looking for a place to get a tattoo in Hamilton, Ontario. The top result had one review, but no image. The second listing had an image, and the fourth result had a review with text snippets. Searchers looked at this search result in the classic F-shaped, “golden triangle” scheme.

Mediative’s study of online click-tracking for the same query produced somewhat similar results: Most clicks went to the first listing, with the second and third businesses getting progressively fewer clicks. The fourth listing, the one with a review and text snippets, appears to have had more click activity than the third result.

When Social Content/Signals Change Behavior

Things got more interesting in the next stage of the test. The participants did the same search, but for London, Ontario this time. In these search results, the first two listings were basic with nothing more than contact info and a website URL. The third listing had all that, plus a ratings indicator showing three red stars, an indicator that Google has eight reviews for this business, and a text snippet from one of those reviews. And in Mediative’s eye-tracking study, those social signals and content helped the third listing attract more attention.

And the online click-tracking study for this query also showed the third listing getting a substantial amount of click actively — certainly more than the second business and a comparable amount of clicks as the top-ranked business.

As the study participants reproduced these queries using two other Canadian cities, the results were similar: The top-ranked business always garnered attention and clicks, but listings further down the page did well when they had additional social content like star ratings, reviews and text snippets.

As I said above, it may be hard to draw any definite conclusions from a study with 12 eye-tracking participants. But Mediative makes two observations that seem relevant despite the small size:

1. When the top results have fewer social signals such as reviews (e.g. London), lower results get more visual attention.

2. If your website is listed in any position other than the top, and your listing does not include any social signals, it will be relatively ignored, especially if there are other listings that do have social signals.

The study also seems to reinforce past reports about the growing importance of reviews for local searchers, and perhaps also a different eye-tracking study that showed searchers do pay attention to snippets (although those were traditional organic snippets, not local search snippets).

Mediative’s whitepaper is available for download. You’ll need to provide contact information to get it.

EMarketer has forecast that total US search ad revenues will reach just over $17 billion in 2012. If we do the simple math, and assume that mobile spending doesn’t add anything more to overall revenue growth, here’s what we get under EFF’s two mobile scenarios:

• 22 percent: $3.7 billion
• 16 percent: $2.7 billion

It’s worth noting that tablets already account for a significant portion of mobile search ad spending, as well as mobile clicks:

Efficient Frontier is seeing a sharp ramp in its search spend on Tablet devices. Over the past three months (Aug 2011 – Oct 2011), Tablets accounted for 43% of Efficient Frontier’s mobile search spend, and 50% of click share.

Google stands to be the beneficiary of mobile search revenue growth, at a disproportionate level vs rivals. Among EFF’s client base more than 95 percent of mobile search ad spending goes to Google.

This is commensurate with and a direct result of Google dominating mobile search. According to StatCounter Google controls about 95 percent of browser-based mobile search in North America.

Summarizing EFF’s lengthy discussion of mobile paid-search performance, mobile ads have higher CPCs and better CTRs but mobile ROI is less because of lower conversion levels:

• Across its advertiser clients, Efficient Frontier found that the average CPC on mobile phone search campaigns was actually slightly above (8% higher) than CPCs for desktop search campaigns.
• Mobile phone and tablet click-through rates for Efficient Frontier’s search advertisers are higher than for its desktop search campaigns, at 166% and 137%, respectively, of CTRs in desktop search.
• Lower conversions for mobile search campaigns result in ROI that is only 50% of the ROI for desktop campaigns according to Efficient Frontier’s data. Notably, 50% represents a significant improvement for this metric compared with mid-2010, when mobile ROI was only ~10% of desktop levels.

There’s a major caveat here however. ROI on mobile campaigns is being measured by traditional “conversion” metrics used on PC campaigns. EFF points out that a broader and more mobile-specific set of conversion metrics needs to be developed to evaluate the true efficacy of mobile search ads:

Until advertisers can more accurately attribute a wider range of “success” events such as offline store visits, offline and (delayed) online (desktop) sales, and phone calls to a mobile ad campaign, marketers using a traditional ROI-based approach to make campaign budget decisions will be less willing to increase mobile ad budgets. In the meantime, we expect “early adopter” mobile marketers will continue to test mobile as both a performance-based and branding ad medium.

EFF also compared query length on the PC, tablets and mobile devices. As was somewhat predictable, EFF found that PC queries are the longest, followed by tablets and mobile handsets in that order. It’s possible, however, that voice search may create longer and more specific mobile query strings in the future.