How to sift "site search" data from Google Analytics for trends

[rtkl]

I apologize in advance if this has been asked a million times but I'm just not able to find anything on it for some reason. Probably the words "site" and "search" come up a lot in this area... Anyhow, my question:

How do I find trends in "site search" data from Google Analytics?

I set up "site search" a long time ago. I have thousands and thousands of searches people have made on my site logged and squirreled away. The plan was to review them on a weekly basis, find the trends and start writing content to address interests people seem to be having but not finding on our site. Sounded great at the time.

The problem I have, of course, is that among my 10,000 searches (many shown in Google Analytics as "no-results:cats and dogs", etc), there are slight differences that make it difficult to total up search trends. Let's say the list is like this:

Term | Search Count

Cats | 500
Dogs | 500
Cat | 250
Dog | 250
Cat food | 5
Dog food | 5
Birds | 1
Bird | 1
Cats are great | 1
Cats are really great | 1
Dogs are great | 1
I like birds | 1
Seriously, I like Cats | 1
Turtles | 1

... 10,000 more entries, every single one only 1 search per term.

OK, so it looks like people like Cats and Dogs a lot, but also Birds and Turtles. But maybe there are snake searches. Maybe there are "cat pajamas" searches and variations on all of the above. Who knows what else is really trending in there??? The review of this data is MIND-NUMBING. Especially when you get into plurality and misspellings, this rabbit hole has no bottom.

Is there a tool people in the SEO jam use to take a big ole CSV dump and have it magically sorted by at least potential trends?

I mean, there's gotta be, right? And I'm silly for not already knowing what it is.

[BlueCorona]

Hi Thom,

Like Mike mentioned, I think that culling through this information yourself with Excel will probably be your best bet for analyzing the data you mentioned.

I'll give you a few ideas on how I would start if I was you:

You mentioned that many of the queries were only searched once. I would start by organizing your information into two columns: one with the query, and one with the search volume. Then, I would filter those down to only queries that have been searched more than [a certain threshold]. I think that while queries that have only been searched once can give you valuable information, especially compared to others and analyzed in aggregate, you will be less overwhelmed if you start with the most important (and a smaller set of) queries first.

After you've done that, I would start to filter your data by query. You can sort by queries that contain, begin with, end with, or do not contain a term, or any sort of custom filter. Given your example, I would probably start with filtering by queries that contain the word "cat" or "dog" or "turtle," and then try to find trends within that smaller set of data. They will be much easier and less overwhelming to find when you are not dealing with so much data.

This next step is a little more tedious, but I would also recommend adding another column that you manually populate. You mentioned that there are misspellings; you could account for those by creating another column in which you populate the word "cat" for all queries that contain "cat" and also all queries that contain "cta" or any other misspelling. That way, later, you can easily sort by keywords that were about cats, but didn't necessarily contain the exact word "cat." Or, you could populate the column by grouping all queries that were about one animal, two animals, three animals, etc.

Once you have a ton of information, I would recommend creating pivot tables and charts that help you compare the data.

Ultimately, what I'm recommending is probably very time intensive and tedious, but I really think it will be rewarding once it's completed, because you will very thoroughly understand the trends going on in the data. One of the benefits of doing it yourself, rather than using a tool, is that you will understand the context around these words; you will be able to better interpret trends or make connections than an automated tool would be.

Here is a great article that walks you step by step through some of the steps I mentioned, and it also goes into depth about other ways you can use Excel to analyze this kind of data: http://www.sitevisibility.co.uk/blog/2015/09/11/4-excel-tips-applied-to-keyword-research/

Hope this helped!

[MikeRoberts]

Honestly I do it all by hand in excel using conditional formatting to highlight  various core terms and change text color/bold/italicize for certain common combinations that accompany core terms. Then I sort by those variations, copy/paste them into new tabs, and break them down further as needed from there.