3 Ways to Extract Meaningful Audience Insights from Twitter
I'm a big fan of Twitter - it's the social network I use the most and I get a huge amount of value from it. So it confuses me when I see complaints about the platform being too confusing or too difficult, rendering is largely useless in the eyes of many. And while not every brand will get value from tweets, there are a heap of ways to utilize Twitter data that are often overlooked.
Of course, Twitter itself doesn't do a great job at promoting such use cases - Twitter does have a data blog, but most of the research they share is fairly generic.
Now that summer is officially over, here's a look at this year's summer music festivals according to Tweets: pic.twitter.com/EBmXVIZz6m- Twitter Music (@TwitterMusic) September 28, 2016
Those are interesting, of course, but they're not highly applicable. Most users are looking for ways to make sense of the data as it relates to them specifically, for a way into the tweet stream to find more relevance and meaning within the masses of tweet info.
And there is a lot of insight value there - for example:
- The US geological survey uses tweet data to track earthquakes and their potential impacts
- Academic researchers have used Twitter data to predict flu outbreaks and better prepare resources
- Tweet insights have been used to predict civil unrest and crime in various regions
While Twitter may not be as popular as other social platforms, there are still more than 500 million tweets being sent every day, a real-time stream of discussion and information that includes a heap of data signals, and potential value for brands. The challenge, then, lies how to unearth the insights relevant to your business.
Here are some tips on how to use Twitter data for such purpose.
1. Simple Complexity
When looking through that list of examples, it's easy to dismiss them as being highly complex, insights that could only be gleaned by teams of researchers combing through masses of Twitter data. But actually, almost all of them come down to very simple fundamentals. Finding the key points which unlock the relevant data takes work, granted, but once you have found the relevant triggers, they actually come down to fairly basic variables.
For example, the US Geological Survey, when using Twitter data to track earthquakes, utilizes only four variables:
- They look for mentions of 'earthquake'
- Tweets of seven words or longer are excluded (people in an earthquake zone tend to keep their messages short)
- Tweets sharing links are excluded
- Tweets which include a number (like 'magnitude 8') are excluded
That's it - those are the basics for their earthquake tracking system. As you can imagine, it took them some time and research to narrow down those elements, but once they did, they were able to set up a search stream based on those measures and track relevant activity moving forward.
You can take the same approach to your Twitter search efforts - while it may take some time, initially, and some creative thinking around how people are discussing your brand or products, you can feasibly create an indicative search stream that can highlight the most relevant tweets, as they happen, which you can track in real-time.
To get started, you first need to establish how prospective customers are discussing your products, what words and phrasing they're using which then leads to them making a purchase.
Here's a basic example - let's say you've just opened a new pizza shop in Tampa, Florida and you want to get an idea of people who might be interested in your business.
To start, you could set up a basic search string, which you can track in TweetDeck.
In TweetDeck, click on the '+' symbol at the left of screen to add a new column, which will bring up this option window.
Select 'Search' (as highlighted) then type in 'pizza' as your search term.
You can then create a column which will start ticking over in real-time as tweets come through. But this is, of course, far too broad, you need to narrow down your search.
Click on the option to edit column at the top right of your search stream and you'll get a drop-down menu with more options, including location filters.
Enter the location filter, then click on 'Clear' to start your stream fresh with the relevant filters applied.
Now you have a listing of when anyone mentions 'pizza' on Twitter in your region. Simple.
But you probably need even more than that - what about people who are considering what to have for dinner, people who are tweeting about specific pizza toppings, slang terms, even emoji. You can search for all of these too - you can set up as many search columns as you like in TweetDeck, giving you a clear overview of all the related terms and queries in your region, or anywhere.
For this, it's worth taking note of all the various ways people search for your products. You can look at the tweets and replies on your competitors' profiles to see how people are getting in touch with them, or even use an SEO tool like keywordtool.io to give you an idea of related searches (I'm using the free version here).
And yes, you can search by emoji - for this, an easy way is to find a tweet with the relevant emoji in it, then cut that emoji character and paste it into your TweetDeck search stream.
Just as the researchers have done with earthquake activity, the key from all this initial search activity is to narrow down your streams to the queries that are highly relevant to your interests - what are the commonalities in the tweets from people who are actually looking to buy pizza? How are people using emoji in a way that indicates actual purchase intent, as opposed to jokes or other use?
This takes some study, but as noted, you can eventually narrow down your results to specific info of relevance, then build more complex tweet search strings based on that.
For instance, a tweet like this in your region could be a great prompt for you to reach out and offer your services, maybe even a discount coupon.
I wonder what me and Meg are gonna get for dinner tonight. Hoping for pizza or chick fil a- Gordon McDonough (@GordonMcDonough) January 6, 2017
You have to be careful with this, you don't want to come across as intrusive or creepy, but acting on such queries in real-time, and providing an offer based on those tweet mentions, can be a great way to generate awareness.
If you wanted to narrow down tweets like this, in addition to your location filters, you could use qualifiers like:
This removes any tweets with links from your search, refining it further than he basic mentions.
Over time, with various search streams in place, you'll be able to work out which searches are more relevant than others, enabling you to hone your alerts further using additional qualifiers so you're only being alerted to highly relevant tweets.
2. Relevant discussion
Another key way to use Twitter is as a means of locating industry relevant discussion - which, again, is not something Twitter has done well at promoting as a function. They've been working to improve this, uncovering more relevant content through Moments and refining their algorithms to highlight relevant content, but there are existing ways to find the most relevant info for your business.
One of the easiest ways is to use Nuzzel - Nuzzel is a Twitter aggregation tool which highlights the most popular topics of discussion among each users' followers.
As you can see in this example, when I log onto Nuzzel it shows me the most shared posts from the 24 hours among my connections on Twitter, which, if you've refined your Twitter list to highly relevant users and audiences, can be very helpful at tracking key trends and discussions.
But Nuzzel also enables you to see what's popular among other users' feeds. In the above image, see in the bottom right where it says 'Recently viewed feeds'? I know that someone like Jay Baer for example is followed by many industry influencers, so the topics of discussion amongst his audience will likely highlight key industry news and discussion.
By monitoring what's trending amongst these influencers, it can help you keep abreast of the most discussed news and issues - and the same principle applies for your business.
You know who the influential brands and voices are in your niche - using Nuzzel can be an efficient way to track the key topics of discussion amongst these communities, without having to track Twitter 24/7 on your own.
That said, if you do want to track trending news amongst your own communities, or via specific keywords, you can also set up a TweetDeck stream for this purpose.
Using the built-in engagement filters - minimum retweets, likes and replies - you can highlight key mentions and topics generating the most interest and discussion.
This can help you locate important, trending issues in your niche.
3. Topics of Interest
This is still one of my favorite Twitter research notes - a simple way to learn more about key topics of interest amongst the people you want to reach.
Each Twitter user has their bio where they can note down the details of what they're interested in, which is great for finding relevant users to connect with and follow, etc.
But what people say they're interested in is not always the same as what they share on the platform, which is likely a much more indicative measure of their interests.
So how can you find out what people are tweeting about and their key topics of focus?
Using a tool like Allmytweets, you can enter in any Twitter handle and it'll give you a listing of up to 3,200 of their latest tweets.
Using this, you can then cut all those tweets and paste them into Wordpad or similar.
You'll need to clean up the data by removing things like the date info - each tweet also has the posting time listing attached, but those likely won't be relevant for keyword search purposes (be careful only to remove specific date mentions, not all mentions of those letter combinations, like 'may', for example).
You can then take your cleaned tweet list and analyze the most mentioned terms via your app of choice - you can even make a word cloud out of the info to see which terms this user most commonly mentions.
Now, obviously I haven't cleaned my data enough, I can go back and remove come of the more common terms to narrow it down and generate real insight, but it's a quick and easy way to get a better understanding of the topics of interest being tweeted by the people you want to reach.
As you can see, there are a heap of ways to use Twitter data for insight, and most of them are not overly complex or expensive, they just take some time to establish. But once you've done that initial research, you'll have a much better idea of the potential value of the platform for your business, enabling you to set up ongoing alerts to ensure you're always aware of the latest mentions and tweets of importance.
Follow Andrew Hutchinson on Twitter