(1) More control to Google: Google now controls the popularity of words more than ever before. With auto complete surfers were predisposed to complete queries on Google recommended terms. Now, with Google returning search results real time, one is even more predisposed to click on both organic and paid search ads the Google recommends. True, the suggestions are supposed to be based on what surfers are predisposed to type probabilistically, but Google’s suggestions and results only increase the bias !
(2) More impressions reported: Google reports that it will count any result displayed for 3 seconds as an impression. Surely, slow typists will record multiple impressions. This will increase reported impressions and hence lower Click Through Rates (CTR) from Google.
(3) Traffic Shifts: There could be big traffic shifts among paid ads. Consider this example:
When I type “hotels” in Google currently, I see the following results
When I type the same with Instant Search turned on, I get:
Google is showing me ads for hotels near Disneyland. This will increase the impression volume of “hotels near Disneyland” keyword and reduce that of “hotels”.
(4) Quality Score implications: I have noticed in my tests that many times, Google prefers showing brand over generic keywords. When I typed tr, Google showed ads for “Travelocity”. Given that travel is far more popular than Travelocity (based off Google’s own Trends data) we should expect the impression volume for Travelocity to shoot up. This would drop the Click Through Rates of “Travelocity” and could impact its quality score adversely. Our research has shown that CTR explains about 90% of Google’s QS formula. Hence, to prevent advertisers like Travelocity from suffering with higher CPCs, Google would have to adjust its quality score formula to account for the effects of the instantaneous mapping.(5) The Brand Bias: Here is what I got when I typed “a”
The first four results are all brands. Only the fifth result is “a”. Based on Google is saying “a” should be far less popular than Amazon. Lets have a look at Google Trends results:
Amazon is blue and a is red. Amazon is only 8% of “a”’s traffic !
Clearly there is some ordering bias in instant search here. My empirical tests with other letters (try x and z for instance) returned results with a similar brand bias.
(6) Your Brand Name Matters: Let us take my Travelocity example. If I were another competing OTA like Priceline, I would expect significant traffic from the “travel” keyword. However, now with Google’s Instant suggestions potential Priceline traffic could be diverted to Travelocity. Your brand name matters ! The open question is which brands would come on top ? Brands with unique names like Cisco would still get targeted traffic as you need to type “Cis” to get Cisco specific ads while brands like “travelocity” will get a lot of untargeted but high volume traffic from words like travel. Will all this additional traffic convert ? Unlikely. So, is all this additional traffic welcome ? It remains to be seen. On the flip side, I have to believe that Priceline will be unhappy about them not being able to compete on a level playing field for the "travel" keyword.
(7) For SEO, here is what Jason Cooper our Director of SEO has to say:" In some ways it reinforces the emphasis on head terms, the holy grail of keywords that companies spend years and many thousands of dollars going after because the fewer characters that you enter in the search box, the more likely you are to see search results for a head term. It also makes being in the top 5 more critical than ever since those are the positions most likely to be above the fold in the search suggestions. In terms of actionable strategy, SEOs need to look at the keywords that drive the most natural traffic to their site, and know what suggestions are made at every keystroke leading up to that keyword to see if there is an opportunity earlier on the process to get ranked. "
In the last two years, Google has been “optimizing” their query inventory by displaying ads when they didn’t show ads before. One could view this as a “spatial” optimization of ad inventory i.e. showing ads where there were white spaces before. The latest innovation is a “temporal” optimization i.e. showing ads during the time it takes to type your query. And as Marissa Mayer, Google's VP of Search Product and User Experience said, this could be a game changer.
My experiments with Instant Search indicate that the suggestions have a definite commercial bias; branded words are preferred over generic terms despite Google’s own Trends data showing much lower traffic on the branded words. Further, Instant Search could affect impression volumes as well as CTRs on many terms.
In the coming weeks, I expect to see higher impression volumes on branded terms and traffic pattern shifts among unbranded ones. Google will also have to adjust its Quality Score formula to account for the impression shifts among keywords, specially branded ones to enable advertisers maintain their ROI. As with any significant change, there will be winners and losers. However, who wins and loses remains to be seen.Dr. Siddharth Shah
Director, Business Analytics
Will Google Instant Search increase their yield per click?
Posted by: Tim Cohn | September 09, 2010 at 07:48 AM
Sid: My comments are included
Hello Siddharth,
Thank you for sharing your view on this. In the first example ('wa'), I am not sure that anything really changes. User intent is the powerful thing with Search – if the user is going for the word ‘washington’ then the user will continue to type until they see what they are looking for – if WalMart comes up before the user finishes typing, I don’t think that will have any effect on anything. My guess is that if you spent a lot of time, you could come up with examples that are more interesting than this one - but if anything, for me, this example suggests that not much really changes.
I think your point about Google controlling the popularity of keywords even more is directionally accurate, but I also think it will have an incremental effect, not a night-and-day effect, partly because they've been suggesting searches with auto-complete for a long time now.
Sid: Agreed. But I think that they will try to profit from spillover. Lets say I am looking for a cheap vacation. Google controls the ordering of ads for cheap vacations to Hawaii/Alaska/Disneyland. They can, perhaps to a small degree change the popularity of these locations. Perhaps the fluctuations will be small, but at this point Google does care about 1-2% more revenue !
In terms of more impressions being reported, I think this will be marginal – 3 seconds is a long time, and Google may change this criterion depending on the data. For an impression not to be registered, all a user has to do is type one more letter within 3 seconds. At the end of the day, I'm not sure that the number of impressions will matter much – a rising tide lifts all boats, and vice versa. Quality scores are relative, and are based on relative CTRs - accounting for ad position - for a search.
Sid: I am not sure if everyone specially the older demographic can type in three seconds. Perhaps this is a small demographic of users . Also, they are now counting spelling corrections as impressions among other actions. Not sure if they did that before. Don’t you think this will have some incremental effect ?
For the Disneyland example, you omit auto-complete. Most users have that turned on – auto-complete would also show Disneyland as a suggestion, making the difference versus instant search more incremental.
I also disagree that the 'hotels near Disneyland' impressions would increase and 'hotels' impressions would decrease. If the user’s not interested in Disneyland, they’ll type another few letters into the search query box, and it won’t take them more than 1-2 seconds to do so – I think folks go searching because they are searching for something – they go browsing when they are checking things out – if they’re open to many So Cal vacation ideas, I think they are more apt to browse around a travel website, etc.
Sid: Agreed. The change should at best be incremental.
Section (4) seems very interesting, but I’d want to check the data and the results over time – we know Google shows personalized results, and we also know that Google tweaks it’s algos – and my guess is that at the end of the day, Google will stay true to their ultimate value prop, and give users the most relevant results. But obviously, going with Travelocity will certainly produce a more commerce-driven result for the user!
I don't think QS will be impacted much. Google has the 3-second rule for this reason. If the user doesn’t like the results they are seeing, they’ll type another letter or 2 or 3, and QS is relative. So, the other advertisers that show up on a ‘travelocity’ results page would also see their CTR decrease (if the user isn’t really going after this result), and Travelocity’s QS would stay the same, relative the advertisers it's competing with in the auctions. You do infer that Google would likely adjust its QS formula if advertisers are adversely affected (or, Google could adjust the criteria for an impression to be registered) - agreed!
Sid: Yes I think Google will account for the change in impressions due to the changes so as to not affect QS. Otherwise, a lot of advertisers would get pissed. However, I do think Travelocity will benefit from incremental leakage traffic.
In your section (5), I wonder if this is a good example of normal user search behavior? Who types in ‘a’ as a search request? Folks who hit the return key too quickly. ‘A’ isn’t generally a ‘real’ search request, and I think it may be easy to type random letters into the query field and see all sorts of weird and anomalous things, but I'm not sure we should extrapolate from these scenarios.
But it *is* interesting about moving all the brands to the top! Agreed! If they continue to push the envelope in this way, they risk serving up less relevant results, and losing users in the long term.
Sid: I have a better example. If I have bought from abebooks before and I don’t particularly care which brand I buy used books from, isn’t there a chance I would click amazon when I typed a ? Also, am I to believe Travelocity is more popular than travel ? I think Google has figured out the ordering so as to maximize their paid clicks.
Best,
Terry Whalen
CPC Search
Sid: As I said before, ultimately, the data will speak for itself ! Thanks a lot for your feedback.
Posted by: Terry Whalen | September 09, 2010 at 05:14 PM