It’s been just over a month since Google Instant rolled out and we published a first look a week in at its impact on searcher behavior. A comparison of 880,000 visits from ten high traffic websites by search term length the week prior to Instant against the week after showed virtually no change in search traffic distribution:
We promised an update after a month, so here goes.
We looked at search traffic for the same ten high traffic websites, expanding the view from one week pre and post Instant to two weeks pre and four weeks post. Keyword distribution data was analyzed as a percentage of the total negating the difference in periods analyzed.
We analyzed 2.7 million visits over the 6 week period. Aside from seriously testing the row limits in Excel and ensuring the Analyst who worked with me on this will stop taking my calls, the data showed things have remained steady:
The data is telling in what it doesn’t show: if Instant was having a significant impact on searcher behavior we would expect to see things like a percentage of longer tail visits (4+ words) leaking back into the head (<=3 word terms ) as searchers click on Instant results before they get to the end of their intended query.
If Instant was having a significant impact on searcher behavior we would expect to see long tail visits migrate up the head as searchers click on Instant results before they get to the end of their intended query.
Time on site metrics show little change other than one word visitors spending slightly less time on site which may or may not be the result of Instant’s influence.
Moving down the head to longer queries, there are no changes in time on site or page views post Instant, so those who suggested Instant would increase relevancy in the long tail thereby decreasing time spent on site do not seem correct.
Looking at conversions, other than around a 3% shift from 2 word phrases to three, little has changed. With no corresponding distribution shift in visitor traffic we would be hard pressed to attribute this movement to anything other than a normal flux in conversion distribution.
By now, after looking at chart after chart of visitor metrics that appear nearly identical from one period to the next you are probably getting the sense that little has changed after Google rolled out Instant with much fanfare a little more than a month ago. The traffic data suggests searchers are searching the same way they always have and, with a month of getting used to Instant already behind them theres nothing that indicates that will change any time soon.
Are you seeing anything different?
Senior Research Analyst