Impact of Google Instant

So I’m jumping on the impact of Google Instant bandwagon…

I hope to add more data to the conversation and bring something new to the analysis of Google Instant. It’s important to note that it hasn’t been long since the change, so we’re not working with a lot of data yet. However, I think we’re seeing some interesting insights so far.

I’m building on top of conversations started by SEER, Conductor, and SEOmoz; and used the spreadsheet posted by Brett over at SEER.

11 Sites, 75K+ keywords, and Half a Million Visits

Here are a few quick stats about the data set I’m using. I haven’t looked at the data for our biggest sites yet, but Ben is working on pulling down all that information.

Data Set

  • 11 Sites.
  • Various sizes: 3,500 weekly visits to 75,000 weekly visits.
  • Various niches: Ecommerce, Software, Gambling, Travel, and more.
  • Over 75,000 keywords.

I ran analysis on each of the 11 sites.

It’s a bit different than Conductor’s method. I’m looking at the number of keywords that sent traffic, instead of the amount of traffic sent from each keyword group.

Example Analysis

Below is the analysis of Distilled.co.uk.

Percent of Total Keywords

Google Instant Percent of Total Keywords

Let’s graph it out.

Google Instant Distilled Graph

Now, this is just one site. What gets really exciting is when we repeat this analysis on several sites.

Change in Percentages for All Sites

I calculated the change for each keyword length for each of the 11 sites and summarized them in this nifty little heat map.  The values are colored to demonstrate the magnitude and direction of the change.

Google Instant Delta Heat Map

What the colors mean:

  • Red: change < -0.5
  • Orange: change < -0.05
  • Yellow: change between -0.05 & 0.05
  • Light Green: change > 0.05
  • Dark Green: change > 0.5

Now this shows some interesting insights that sums and averages hide. You see a lot more red to the left and more green to the right. Nine out of the eleven sites saw increases in 7+ word keywords. Eight out of eleven sites saw decreases in 2 and 3 word keywords.

Overall Percent Changes

Finally, I aggregated all the data and performed an overall analysis of all the keywords. So here are the overall results from all of the data.

Overall Google Instant Keyword Graph

To clarify what’s happening here, here are the changes graphed out.

Impact of Google Instant

I think this graph combined with the heat map shows some interesting results. The data, so far, suggests that Google Instant might be pulling people away from 2 and 3 word keywords and towards 6 and 7+ word keywords. This shift doesn’t appear dramatic, but there appears to be a slight shift toward long tail keywords.

This is still early data and doesn’t include some of our largest sites, so I’ll continue to take a look as time goes on to see the impact Google Instant continues to have.

Get blog posts via email

17 Comments

  1. Hi Justin,

    Great read! It's interesting that you're seeing long tail traffic improve (we're seeing similar results with some of our bigger clients as well) after many people assumed exactly the opposite!

    Glad the spreadsheet was helpful to you :-)

    reply >
  2. That is interesting... The gut reaction from a lot of SEO's was that it will kill the longtail - but seems to be going the other way.

    Time will really tell though.

    Guess it makes sense - when average user instantly sees that each word they add makes their search more targeted... they add more words rather than backspacing or trying something else.

    reply >
  3. Daniel

    Nice work! I'm going to run some similar studies on sites that our company manages to see if I see similar results.

    reply >
  4. Justin

    @Brett

    Thanks! We're going to keep an eye on it, but its interesting data at least. We're looking at some of our larger clients as well, but I'm letting Ben automate some of that.

    And thanks for the spreadsheet. Hope the data helps add something to the conversation.

    @Wingnut

    I've speculated to myself on reasons why the data looks the way it does, but they're purely my guesses.

    There is also a good issue a coworker of mine has mentioned, but he's working on a post for that.

    @Daniel

    Thanks! The SEER post shows a little hack to get 5,000 results, but you can get up to 20,000.

    reply >
  5. Have also seen an increase in the long-tail on all of the network of websites that I own. Also an interesting thing was that people came through different keywords to many of the same pages as well which could probably be attributed to new recommended search queries through Google Instant. I think instant is actually a positive for most SEO especially those who rely heavily on the longtail such as Demand Media, Mahalo, etc.

    reply >
  6. The appearance of a shift toward 6 and 7+ (not exactly the same as long-tail) seems almost counter-intuitive. I was among those that expected the opposite to be true: that people would start typing with the intention of putting in a longer query and Instant would auto-complete and search a popular 2-3 word query, the user would see a relevant result and click through.

    But I guess it makes sense that because of the predictive auto-complete, Google may be showing longer queries with more KWs than the user originally intended to type. If the suggestion makes sense (more specific to what they are searching for, etc.) it would make sense that the user would go with the longer query.

    reply >
  7. It totally makes sense. I've found myself comparing the instant results to what Google suggest says, and usually choosing one of the suggestions (which are traditionally longer-tail keywords). I don't like using Google instant because it slows down my experience, but that kind of feedback is pretty awesome.

    reply >
  8. Don't think many people would have expected what your research is suggesting. But I still do think that these are early days for Google instant.

    reply >
  9. Im really glad we are seeing these type of results, I didn't think it made sense to predict the end of long-tial, The SEOMoz seminar in London last year quoted that 80% of search is long-tail.

    The whole fact people search long tail should have suggested that their isn't a two word solution to get to your result. Google can't reduce the length of a search term and become more accurate.

    reply >
  10. VERY INTERESTING Justin! This must have been very time consuming.

    Thanks for sharing this it is very valuable data. What's left is to compare the results with other countries and on a larger scale to see if it corellates.

    Are you going to post an update later this month?

    reply >
  11. Very interesting analysis! Google Instant is somehow changing users behavior when searching by Google, so it's worthy to analyse the way Google Instant affects it. Thank you for sharing!

    reply >
  12. I like how you utilized the heat chart to display the changes for each website. Although I only work on one (for the time being), I hope my post will be just as valuable as yours. Thanks again!

    http://seochristopheryee.wordpress.com/2010/09/29/aftermath-of-google-instant-on-search-behavior/

    reply >
  13. Justin

    @Paul

    Yeah, Rob and I chatted a bit about the reason we were seeing what we saw, but it was just speculative. I didn't have any data behind why it was happening. I think too many blog posts have come out around Google Instant based on opinion and a few examples. I have guesses.

    @Nikhil

    I'm looking forward to taking another look in a few weeks to see if we're seeing anything different.

    @Index Web Marketing

    It wasn't too bad. I couldn't manually pull all the information for our large sites, so I asked Ben to do that via API. The spreadsheet from SEER was a huge time saver.

    I'll check in on the data in few weeks and see if I have anything else to really add. If we see something worth posting, I'll put together another post.

    @Christopher

    Thanks!

    Always great to see people sharing data.

    reply >
  14. Some excellent insights. We've a lot of customers selling niche products that live and die by long tail search. Your results are comparable with our the sales volumes those sites have since instant came along.

    Thanks again

    reply >
  15. But how does Google Instant affect the local-search space? Is it "instant LOCAL search"? Thought you all might benefit from the following paper: "Is It Instant Local Search?" It's available for free at http://tmpdm.com/ourwork/whitepapers.asp.

    reply >
  16. Great post Justin. I like how you broke it down by the heatmap - very interesting!

    Here's some more data from a similar perspective, and also segmenting it into non-branded keywords:

    http://www.lunametrics.com/blog/2010/10/05/google-instant-effect-on-long-tail/

    (24 sites, 3,182,487 visits, 1,197,374 keywords)

    reply >
  17. I had a feeling that google instant would increase the amount of long tail searches. The only issue I potentially see arising is the same 'set' of longtail keywords being forced into the instant results.

    Many companies have seen this happen with the addition of "scam" after their brand name, and if that appears then its more than likely users will click it, creating a vicious circle.

    reply >

Leave a Reply

Your email address will not be published. Required fields are marked *

*
*

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>