Analysing search terms in Google analytics

Google analytics provides a useful method for analysing the effectiveness of your website’s search through its capture of the search terms used and their accompanying data.

Approx 15% of visits to our intranet involve search, with 10,000+ searches run on average a month. It is therefore important that the search results are working for users.

I carry out a monthly exercise to identify searches which aren’t performing as well as expected. By improving the keywords, structure, titling and language of any relevant content I try to improve the search results. A month later I come back to see if the changes have had the desired impact and to repeat the exercise.

Using the search terms tab within the site search section of the content report within Google analytics I identify the search terms with a high results pageviews per search (see the circled column within the image below). My thinking is that it is bad for a user to have to click to more than one results page for a search (I have always been taught that the majority of users won’t even look beyond the first results page).

Google analytics screenshot of the top 10 search terms used

Using the results pageviews/search column (circled) can assist in identifying terms which users are having trouble searching for

I tend to look for search terms with a value above 1.55 (the site average), giving greatest attention to the terms with the highest values (I have seen terms with values as high as 4.00). Alarms bells should ring if your site average is around the 2.00 mark. Terms with high values in comparison to those search terms with a similar frequency catch my attention more than infrequently used terms with a high value.

The impact of any changes take time to have a visible impact and our intranet searches seem fairly fluid, 50-60% of the top 10 terms used change monthly, meaning this is a valuable exercise to revisit monthly.

I also restrict the number of modifications to 10-15 terms each month, we have 6,000+ unique terms used each mothh, if I can get the most common used terms performing well the less common (but similar) terms should hopefully improve similarly.

There is a caveat to this work, you are working with one main assumption: that you know what content users are looking for when they search using a particular term. Revisiting the changes you have made after a month is important to ensure that the changes you have made have had the desired impact, suggesting your assumptions about the content being looked for were correct.