Let's talk about Google Trends. A lot of different industries/communities use this tool, but let's discuss it in the context of researching the news and social trends.

First, let's get this out of the way. Google Trends is a tool for measuring the relative search volume for keywords or topics over time. You can adjust the lookback window, geographic segment, and sometimes choose between the "search term" or "topic" versions of a keyword.

With that said, Google Trends does not measure Twitter/social media trends, posts, articles, elections, engagement, or ranking. This might seem confusing at first because, after all, it's called Google *Trends* right?

Well, yes and no. It's called Google Trends, yes, but the term "trend" meant something different in 2006 when both Google Trends and Twitter were launched. Back then, SEO was the cat's pajamas so the search marketing industry pretty much controlled how terminology was used.

That meant the term "trends" applied specifically to organic search trends. Since Google was already the elephant in the room this technically meant "organic search trends on Google." It was shortened to just "trends," which is where we run into our first problem.

The Name. It seemed like a great idea at the time, but the context and usage for technical terms change. Thus, you may find yourself with a confusing brand name. In this case, people assume it's social trends because organic search isn't the main way to find information, anymore.

Some of you may not even remember the organic search wars. You probably don't shudder reading this list: Florida, Austin, Panda, Penguin, Venice. Back then, we all thought these things would matter forever, but times change.

So, the brand could stand for a refresh, but it's so well established it's not really a viable option. These are the kinds of choices you often have to live with as a brand, so choose your name wisely. Now that that confusion is out of the way, let's just cover what it does.

A comparison with no baseline is meaningless, so I recommend picking some "control keywords." This tool will ONLY give you RELATIVE search volume. It doesn't tell you the sample nor sample size (more on that later). My control kws are "tacos" and "cats."

Why? They're popular and "taco cat" is a palindrome. The key is to just be consistent. You're already pretty much comparing apples to grapefruit, so be consistent with your control. Good control kws have relatively flat growth, and keep pace with the expansion of the userbase.

If you want to get into the nitty gritty of the taco cat lines, I wrote an article on this years, ago. I need to update it, but the information is still relevant. https://www.chow-bryant.com/analytics-tracking/cat-line-brand-reach-google-trends/

The reason the control kws are important is it helps us put a trend into perspective. Notice how "owl facts" seemed like a thing until we added "tacos" and "cats." No matter how many facts I write, it's probably just never going to be that popular. It's something to do for fun.

While it's all well and good to have a tool for measuring relative search volume, it's also VERY relative. What do these numbers mean?! What's the scale for the bar graph?! This should all be clear. So let's check the help tip.

The average user is going to understand these values as percentages. 100 means that's 100% of the total search volume. This is clear as mud, and it already assumes its audience is familiar with the terminology and its usage. This is pretty bad.

"Search interest" here means "search volume." The aggregate total of unique searches on a keyword on a timeline. Search volume *does* correlate to "interest" but not always. For instance, a botnet searching a keyword is not "interest" but it sure is volume. Also, no FAQ?!

I hope you can already see how misunderstood and misused this tool is. It's very limited, and more useful for organic search marketing than it is for online research. But that doesn't mean it's worthless for this purpose.

If you use this tool to track trends on a daily basis, you are probably reading tea leaves. Tools like this are best used in larger intervals over a long period of time. Think monthly across a year. It's really sketchy for trying to research emerging trends.

The kw you're researching also needs enough volume behind it to even show up. There are plenty of "no data" kws out there. So already these are going to be some major limitations for researchers.

But this is a great place to discuss the "search term" and "topic" options. If the kw has a lot of volume and different contexts, you're going to see this dropdown. Note: when comparing kws always make sure they're all the same type. Try not to mix search terms and topics.

The topic options are useful because those include search volume for the selected kw AND closely related search terms. Other topical options like the proper noun version help us filter out searches for cat pics and only see interest in 'Cats' (2019). See, no one was interested.

With all that said, Google is clearly assuming users already know terms like keyword, search term, topic, volume, search interest, search trend, semantic search as well as their usage. That's unrealistic, and SEO is arguably too outdated to command that kind of position anymore.

Ways to improve this include:
Adding an FAQ
Clearly labeling charts and graphs
Including information on sample size
Any documentation at all.

While the layout and UI/UX are clean, it's not at all clear what this tool is for unless you're thinking in 2006 terms.

Before I close this out, we have to talk about what search interest is and what it can be used for. It's not going to be a reliable way to track the spread of an article or obscure term. It's also not good for predicting elections. It's just not that simple.

But this is where I need to give Google credit. They do provide some clickable examples in a prominent position. These are all great, but they're probably not clear to everyone upfront what they're teaching.

From left-to-right:
We can gauge search volume in a "Diva Fight,"
Analyze interest in a worldwide event segmented by region
Look at regional uses of keyword variants like "Football" and "American Football."

So yeah, these are great examples, and they can teach new users how to use the tools. But that kind of requires you to already know the things these examples are testing in the first place. But key takeaway is they all have testable hypotheses.

How would this apply to research? One example is looking for a correlation between a breaking news topic and search on a keyword. This gives an INDICATION of how popular the story is, and how quickly or slowly user interest in Google organic search is growing.

If you're lucky, the story you're tracking will have been published by multiple sources with sustained reporting over a few weeks or months (rare). If so, then you can use the dateline to look for a LOOSE correlation with a specific pub date to see who had the greatest impact.

But most of the time you're not going to be so lucky, so this tool won't help you there. That said, you can still see if interest on a topic is growing or waning in organic search, and which regions produce the most searches.

Something that's important to keep in mind is that search volume = / = approval. Outrage is one of the best vehicles for traffic, so don't read too much into interest. Think of it as neutral, or a combination of all possible sentiments. This is why it's a bad tool for elections.

That means you cannot use Google Trends to gauge or examine sentiment very easily (or much at all). You can sometimes accomplish this via longtail keywords that express some kind of sentiment, but it's rare for people to tell a search bar their feelings like it's a chat client.

Don't read too much into the above graph. While it suggests that far more people are searching for how to get the vaccine than to avoid it, there are better ways to gauge this. We don't know all of the terms people are searching on this topic, so who knows if these even correlate

That graph can be interesting, but better methods for answering these kinds of questions are surveys/polling, tracking vaccination rates, and analyzing sales trends on things like disposable masks (but if you factor in use case, even that's murky).

The biggest takeaway there is to always reproduce your test multiple times, compare your findings with others' findings, use consistent methodology, and try to corroborate a hypothesis with multiple data sets.

I hope this helps you gain a better understanding of how to use Google Trends, what it can and can't do, and whether or not it's actually useful for your research. It can be a fun and powerful tool, but it can also lead to fortunate-telling. Now in the spirit of '06--LASER CAT!