In this series I will be exploring the ways you can measure your digital science communications, starting in this article with Twitter. I'll show you how to get started with data capture, measurement, and analysis as well as taking insights from your data.
Chris J Bennett
February 25, 2017
In this series I will be exploring the ways you can measure your digital science communications, starting in this article with Twitter.
I'll show you how to get started with data capture, measurement, and analysis as well as taking insights from your data.
Some of these methods were used to help us win an Excellence with Impact Commendation Award from the Biotechnology and Biological Sciences Research Council (BBSRC) for our work on communicating research on the wheat genome.
Without being able to measure our work, we wouldn't have been able to demonstrate it's impact.
Let's get started.
Twitter is one of the best platforms for engaging people with science because it is fast, open, and easily measurable.
For those new to science communications, it is a great place to start. There are also plenty of more powerful features and strategies for advanced users.
A tweet is the unit of communication on Twitter. That's pretty obvious. People send tweets out and other people read them and there's your communication.
But there's a lot more to this than first appears, so let's look at this in more detail.
You can see lots of things going on here. Most of them are self-explanatory but, just in case, here's the rundown:
Account Handle - a Twitter account identifier. Separate from the name of the account. If you want to send a tweet to an account, you need to use this, and not the account name, which isn't unique. These are. They always start with an @ symbol.
Account Name - the name of the account. Different from the account handle and are often not unique. You can have lots of accounts called John Smith, but only one @johnsmith.
Tweet Copy - this is just the text of the tweet. In the communications profession, text is often referred to as 'copy'. This is the body of the tweet, and can include links, hashtags, and @ mentions.
Hashtag - Can't do better than the dictionary on this one. A word or phrase preceded by a hash sign (#), used on social media websites and applications, especially Twitter, to identify messages on a specific topic. Use these for visibility. If you are tweeting about space, you can use #space in your message, for example. Your message is then grouped with everything else that contains this tag.
...you need to have clear objectives for your program.
Link - a URL, which can point to anything that a URL can point to, such as a website, video, PDF, image etc. Note that Twitter will automatically shorten your URLs to make them easier to read and to keep lengthy URLs from cluttering the message. These shortened links are called 'shortlinks'. Genius. There are other shortlink options available, such as the popular Bitly platform.
Image/video - you can add an image or a video (even a GIF) to a tweet.
Twitter Card - a rich snippet of content presented in a predefined way that adds a more professional feel to the tweet by making it into a "card" of information, specially designed for sharing and better information presentation. Tweets only appear as cards if the publisher of the content being linked to has configured this in their infrastructure. I'll go over these in a later post, but for now Buffer have twitter cards covered nicely on their blog.
Reply - how people reply and comment on others' tweets.
Retweet - how you can share other tweets on your own account.
Buffer Button - optional, appears when you have a Buffer account linked to you Twitter account for content sharing. See my post on content curation for more on this.
Pocket Button -optional, appears when you have a Pocket account linked to your Twitter account for content sharing. See my post on content curation for more on this.
Like - how users indicate they 'like' a tweet. Likes are stored on your account so others can see what you have liked. Think of it as a form of approval crossed with bookmarking.
Tweet activity - this button is only available if you have enabled Twitter Analytics for you account, which I will cover in the next section. It's essentially a popup with brief data on your tweet and allows you to Quick Promote your tweet without logging into Twitter Ads (more advanced features I will cover in another post).
A tweet also has a lot of technical stuff going in (read: code) that lets it function on the internet in the first place. A great example of just how much information is carried in a single tweet can be seen in the following diagram:
The good news about all this is that all these things we have looked at so far are measurable as a way to assess your science communications on Twitter.
So let's take a look at how to get started doing this.
Before you measure, analyse, or interpret anything you need to understand why you are doing your science communications in the first place.
"Because you have to" isn't a good enough answer.
"Because science is interestingly, amazingly, wonderously, consciousness-raisingly, intelligence increasingly, intellectually liberatingly, civilisation-buildingly awesome" will do.
I won't go on and on here about why you should be doing it at it's core, but you need to have clear objectives for your program.
Some example of the ones we focus on at Earlham Institute include:
• Grow an audience of people from the general public who are interested in the work we are carrying out
• Put our science out into the public domain aimed at the general public
• Aim our science story content (articles, videos etc) at a science-interested general audience
• Encourage these audiences to engage with our science
• Get these people to visit our website and find out more about us
• Have a dialogue with these people (two-way, of course)
There are many more, but these are just a few that we can directly measure in the digital space, and certainly through our work on Twitter.
If you haven't seen our Twitter account yet, go now and look. I'll be using it as an example for the rest of the article, so get familiar with it. Brownie points if you follow us.
Ironically, I can measure if you do. I'll show you how later.
So what are your goals? What do you want to achieve by being on Twitter, tweeting about science?
Individual researchers might have different goals to organisations, so whatever works for you. I'll be writing about goal setting and other strategic approaches later, so check back for those if you need help here.
If you are really stuck and can't wait, use the good ole' SMART method: Specific, Measurable, Achievable, Relevant, and Time-bound.
For this article, I will use as examples some of the goals we focus on at Earlham Institute as listed above.
Let's assume you have some goals in mind and have put out some tweets.
But how do you make sure all the data we talked about is captured?
The first thing you want to do is to enable Twitter Analytics. Simply visit the link, and sign in to enable analytics on your account.
Twitter Analytics is an entirely separate platform for data analysis that lives in the Twitter product infrastructure. Once enabled, you can either bookmark your analytics page, or visit it easily from your own account by clicking your profile image and looking for the 'Analytics' option.
From here you can track your activity on Twitter, from monitoring follower growth, to seeing which tweets performed the best.
Most of the basic activities users make on Twitter are automatically tracked and available here.
You can see that on my own Twitter account analytics dashboard that my top tweet this month generated 3,661 impressions. (I've shown my own because some of the data associated with the Earlham Institute account is confidential).
An impression is when the tweet appeared in someones timeline. This is a good indicator of the reach your tweet had. It doesn't mean that every impression is the same thing as a view.
Like with news coverage, think of this as circulation - but you don't necessarily know exactly how many people actually saw it. The only way to know this is to stand behind all these users while they use Twitter and make detailed notes on what they are paying specific attention to out of the no doubt hundreds of tweets appearing in their timelines daily. This is clearly impossible, so we take impressions with a pinch of salt.
I want to stop for a moment and talk about dark social. It sounds scary, and it is - for data analysts.
Dark social is what is not measurable, which is important to know. Have you ever shared a great article or video with a friend via SnapChat, text message, email, or just told them to check it out over coffee?
This is dark social. It's essentially untrackable and unmeasurable (at least directly) and presents a major issue for communicators that is still yet unsolved. It's data that cannot be captured, leaving a large gap in your analytics. Around 75% of all content shared online is shared in these dark social channels.
All is not lost, though. There are still plenty of data to look at and direct engagements with content are the easiest ones to measure, a small saving grace.
Back to Twitter Analytics.
Dark social aside, we can see both a panel of engagement metrics to the right, as well as information on the individual tweets themselves. I've included a screenshot below of both the individual tweets dashboard as well as a closeup of the engagement panel.
Pay close attention to engagements and engagement rate, as these are very useful for measuring science communications performance.
An engagement is any time a user interacts with the tweet and can include the following:
• Detail expand
• Link click
These are all engaging actions that users need to make a choice to perform, indicating a higher level of involvement over someone who sees the tweet, perhaps reads it, but moves on without interacting.
The engagement rate is simply the number of engagements divided by the number of impressions, expressed as a percentage. This is the percentage of impressions which led to an engagement. It's a good way of tracking how engaging a tweet or campaign was.
In the communications profession, this engagement is part of a typical conversion funnel.
This is important to know, because we want to measure how our audience interacts with our science communications activity.
Without going into too much detail at this stage, a conversion funnel starts off with an audience who are not aware of a product. In this case, the 'product' is simply our science communications 'thing', like some research that was carried out and explained in a web article. The article is the 'product'.
The audience then move through stages down the funnel, becoming increasingly engaged until an action is carried out.
The action is the thing we want to achieve and is tied to the goals you will have thought about earlier. Here's a rough outline:
Let's say we are researching rocket systems to get to Mars, and want to communicate this. Our funnel would be:
AWARENESS - Exposing people to the research on rocket systems
INTEREST - Getting the audience interested in this topic ongoing
DESIRE - Making the audience want to come back and keep engaging, to learn more
ACTION - Users doing the things you set as goals, could be signing up to your newsletter, visiting your website
You will now see that measuring how your audience engages with your Twitter activity is a good indicator of how far down the funnel they are, and thus how your objectives are being met.
We now want to look at the data and see what is going on.
Let's work through a full example, from a piece of content at Earlham Institute.
I'll show you lots of different ways the data from Twitter can be collected and used.
The content in question is an article called "10 Things you Need To Know Before Starting A PhD Degree".
• Put in front of potential PhD students, so undergraduates who have science degrees
• Get them to engage with the post
• Get interested students to get in touch and apply for PhD studentships with us
So we know the audience, we have the content (thanks to our super writer Pete Bickerton), and we know what our objectives are. We put some tweets together and sent those out with a link to this article:
Some of the things I am immediately looking for on Twitter are:
• Number of posts
• Number of engagements overall
• The breakdown of what these engagements were
• The engagement rate (%)
• The number who visited the full article on our website from the Twitter campaign
• The number of those above visitors who got in touch about a studentship
Let's jump in.
I can find out my reach by looking in Twitter Analytics for the tweets that are to do with this content, then adding up the impressions.
Without a proper social media management platform that can segment tweets by campaign (we use Oktopost), this can be time consuming, but it's worthwhile if you need to know at this level of detail.
I recommend creating a Google Spreadsheet, with a tab per month, and adding in your numbers. Then use a formula to add them up.
You can even include multiple campaigns per month or data from other digital sources. Keeping it all in one place really helps for reporting, and is a real help if you don't have an enterprise solution to hand (which is recommended if you want to do this work seriously, of course!).
Ok, so looking at our tweet data for the PhD article I see we generated 116,000 impressions, from 33 posts overall.
There were 284 engagements overall, breaking down into 31 likes, 25 shares and 2 comments.
This gives me an engagement rate of 0.24%. We average around a 0.1% engagement rate, so this is almost twice what we usually get, a great improvement and an indicator that this campaign was well received.
I can also see that it generated 226 clicks back to our website.
Tracking whether these web visits ended up in a studentship enquiry is an advanced form of attribution tracking, and I'll deal with this in a separate post.
For now though, it looks like the campaign did well.
But what constitutes good?
When you are analysing data, you need to know where you stand in a wider context. How do I know that 284 overall engagements is good?
Well, because we usually get a lot less than that. I run the numbers on everything we do, so I can tell by looking at the results of multiple campaigns what we can, on average, expect to get for each one.
We essentially benchmark against our past activity. It is difficult and often unrealistic to compare with others, who may have significantly different budgets, staff, expertise, time etc. Too many variables which skews the comparison.
If you know a partner organisation who is similar in size and scope, particularly in terms of communications activities, by all means get in touch and share your data for comparison.
We can't consider Twitter in isolation.
We are doing so right now, for the purposes of this article, but you will want to know how well your Twitter activity is doing compared with your whole social activity. If it's not comparing favourably, you might need to rethink how you invest resources in the platform, for example.
You can see that for our PhD article campaign, we dominated with Twitter, so we know to keep pushing these sorts of things on the platform.
One trap you need to make sure you steer clear of is looking at underperforming channels and assuming you need to abandon them, or reduce investment in these.
They could be underperforming precisely because you are not either investing enough time, or carrying out the appropriate strategy for that platform.
For this you need to take a holistic view of you activity over all social channels, reflect on your own strategy and take insights from this.
In the case of Earlham Institute, Twitter is our largest and most effective channel, known because of the large amount of data we collect and the work we do on each one. So to see LinkedIn and Facebook lagging behind Twitter here isn't really a surprise for us.
It's important to be able to understand your audience.
In Twitter Analytics you can see your follower growth as well as demographics and interest breakdowns of your audience.
Follower growth will allow you to keep an eye on how your activity is affecting growth.
You are looking for spikes around the same time as key campaign activity.
When looking at pure audience growth, you need to calculate what your base growth is first. This is the rate at which your account grows if you don't do anything, and it's easiest to work out by looking at your inactive or stable periods of activity and seeing how many followers you gain per month.
You will gain more followers per month gradually as your follower growth increases anyway, but it's important to be aware of your base rate of growth and how this changes so that you can effectively measure any increases which are due to specific campaign activity.
In the image below you can see the different demographic information which is available about your audience.
You can see that my audience is interested in TECHNOLOGY and TECH NEWS, which is what I want to be targeting, so I know that I am attracting the right kind of audience.
I can also see SCIENCE NEWS on there, which is good because I want to communicate about design and digital (tech) with a focus on the science sector.
You can find information on the gender and age ranges of your audience as well as their geographic location. This can be very important if you need to communicate about research that is relevant to a particular place in the world.
You can also compare this with different audiences, across all sorts of interests. In the below image you can see that you can filter audience interests even down to specific areas of science.
This data, while only taken from US users, is still useful and can be broadly applied to audiences in the Western world in general. We see that the demographics around users interested in BIOLOGY also includes the following information:
However, this does not account for users who are not interested in science - how do you measure whether you have actually reached a general audience without a pre-existing interest in your science area?
A good way is through the analysis of engagements, which I'll get into shortly.
I'll finish this section with a tool that let's you analyse your followers for free, called Followerwonk. There is a paid version, which I recommend, but the free follower analysis tool is a good place to start.
It will analyse your followers and give you a detailed breakdown of where they are from, their activity with you, their interests and much more.
Another nifty tool for seeing more information about those who engage with you is the Klear Chrome Extension, which shows you detailed information on the users directly from Twitter. In the below image you can see it adds in info about the account being in the top 2% of social users, as well as being active in the 'science' field.
You can also look at your network of engagers, through a relatively unknown, but really amazing tool called Netlytic. This tool let's you map your social connections and see patterns and engagement.
Here's a link to Earlham Institutes network for you to explore as an example.
You can analyse you entire account profile using a free reporting tool from SimplyMeasured. It also contains a lot of information about your followers and audience.
Looking a the wider picture is important in science communications. We want to see how views and opinions on science, science-related policy and other cultural factors change over time.
A good place to start is by looking at audience sentiment - what are they feeling about things?
Sentiment analysis essentially uses lists of trigger keywords classified by emotion and scans social activity (mostly text-based messages) and cross-references these with the keywords.
An example would be a message containing the word "hate" - it's likely this would be given a negative sentiment.
Let's explore the issue of GM Crops.
If we are running a communications campaign to raise awareness about the real science issues behind GM Crops then we want to look at the sentiment around this topic before, during and after the campaign to see if there has been any effect, such as a change of views (or average sentiment scores to represent this) along with engagements to support this change.
To start with we can use a few tools to do this:
1) Mentionmapp [retweet and mention network]
You will need to create an account and sign-in with Twitter. Once you have done this it will create a map of all the times you have been mentioned (using your @ handle) as well as a map of all the retweets you have had, showing the network and connections of users responsible.
You can investigate users in this way too, so you can look out to see who the people are who are responding to your science communications.
2)Sentiment140 [sentiment analysis]
This free tool lets you search for anything on Twitter and it will display the messages and corresponding positive or negative sentiments.
This data is useful not just for measuring, but for campaign planning. Unsure what message or users to target? This will let you know what they are thinking and who is thinking it.
3)SocialMention [sentiment analysis, message mining]
This is an amazing tool. Simple, but powerful. Simply enter your search query and it will pull in messages, users, keywords, hashtags and more.
Bonus: you can download this data as a CSV file and work with the data in Excel or Google Sheets. try building an 'insight database' with it for future campaign planning.
Another to measure is the actual discussion and comments that people make on Twitter, which shows a really good level of engagement and demonstratea a two-way dialogue with an audience.
At Earlham Institute we use Mention, and below you can see the main dashboard with all the mentions of us and our work appearing.
These sort of tools are usually similar, so try a few and go with what you prefer. They also track any other coverage online, such as your content, press releases, videos (etc) related to keywords.
If we are monitoring discussions and comments around a certain topic, we will then tag those messages with the topic, so we can run a report on the dialogue at the end of the campaign.
Best of all, because we can see people talking about us and our campaign in real-time, we can join in the discussion ourselves, or encourage our researchers to do so.
Bonus: connect a monitoring tool to a Slack channel for real-time updates you can discuss and share with your team.
By now you should have a really good idea of how to access data from Twitter.
Now you need to put that into action.
You should focus on taking away informative and actionable insights.
Just looking at some data from the SocialMention tool shows us that if we want to communicate about GM Crops on social media that it would be a good idea to include the keyword "GMO" to increase exposure. We can also talk about "tolerance", "reevaluation" and the "benefits" of GM Crops - as these are all words being used in current discussions.
Once we know more about our audience in general, those who engaged with our campaign, and what they are talking about, we can then develop our overall strategy for how we can communicate better.
When you put all this together you should be able to notice an increase in any of the metrics we have discussed - the point is to do the analysis work and see what has happened. This is the hard bit and requires patience, accuracy, and an unbiased viewpoint.
Let the data tell the story, don't impose your own.
Much of this article has focused on communicating to a certain audience, and that is your current, earned audience. Though these methods work for anything you do on Twitter for any audience, what do you do if the audience you want to reach is not a part of your 'owned' audience?
There are two things which most help here, but because each of them is an entire separate field of communications in their own right, its beyond the scope of this article to cover it.
I'll give you a hint.
One is paid advertising (yes, on Twitter).
The other is influencer marketing.
Both are full-time, well salaried jobs, so it's not easy to cover them here, so stay tuned for future posts which cover these topics in depth.
Eventually you will be able to connect my posts together for the information you need to deliver world-class scientific communications.