The basics of Google Insights For Search
Google provide a very nifty tool called Insights For Search, that allows you to compare relative search volumes over time and by territory. Here we look at the basics.
- A bit of background
- A worked example
- Watch out for misspellings
- Think about time of year
- And also search location
A bit of background
Google, as we are all aware, is the dominant search engine across much of the world and has been for ages. They receive a huge amount of traffic, and record a massive number of search queries. As such, their data on search volumes over the past few years can be taken as statistically valid so long as the search term being assessed is of sufficient volume.
Insights For Search provides access to this data, allowing you to query the relative volume of searches for a given set of phrases over time. Relative is the key element here; volumes shown are shown in proportion to both their own historical values and those of the other terms queried.
A worked example
For example, if you compare the terms “hotmail” and “yahoo”, the results give you a graph of the relative strengths of each phrase. The right hand axis is scaled so that the largest result of either data series is 100, and all other data points are relative to that. This is shown above, or you can visit the live data yourself.
Notice that the system automatically default to the longest time scale, which is five years. I've actually removed the forecast and the new headlines references for simplicity - more of these later.
You can also see the average volumes over the entire period for each term just above the graph on the right. So you can see for this query, yahoo is beating hotmail 76 to 52. You'll only see this numbers if you are logged into a Google account, without them you'll just get the graphs.
Watch out for misspellings
Now query misspellings of these terms, for example “hotmal” and “yaho” - you now get another set of results. Again the graph shows how these terms compare, but don’t make the mistake of comparisons between graphs.
Instinctively, the correct spellings obviously get more searches although as each graph is drawn to a different scale this isn’t initially apparent. If we redraw the graph with all four terms included, you can immediately see the huge difference in volume - the misspellings appear as flat lines stuck on the bottom axis.
Think about time of year
OK, now let's think about time. Some search terms, such as hotmail and yahoo above are faily independent of the time of year. Whereas others can be very time critical.
So let's take one very big event from the middle of the year, the Wimbledon tennis championships, and one from the end of the year, New Year. If you look at these on the graph above, you'll see how the search volume on Google mirrors the relative importance of those events across the year. You'll notice that I've changed the time scale to show just the last 12 months.
You can even see how the Chinese New Year on January 26th creates a smaller spike.
And also search location
You'll see top right that you can filter by country, and in some cases region also. To demonstrate the differences you'll get, I've shown two graphs above. They both for the last 12 months, and both are for the search term "summer".
The top graph is for searches from the UK, the second is for the same searches from Australian users. This is a vivid example of how different parts of the world, in this case hemispheres, can show very different search trends.
I tend not to pay too much attention to any geographic region lower than country level, but I base this purely on my suspicions of accuracy and have no data to prove this either way.