Social Media Analytics FAQ
Tackling the FAQs from the last Analytics Meetup, from ROI to sentiment analysis, from campaigns to "Content is King".
After a long hiatus from meetups, conferences and “digital events”, I finally returned to the London Web Analytics, a great London-based meetup for people interested in all things analytics.
The subject for this month’s meetup was social media analytics, pretty topical considering that I work in social media analytics day in and day out.
I met quite a few great people there, including the lovely Samantha Sile…
I had lots of interesting conversations, and I also had my first-ever analytics selfie…
Needless to say, it was an amazing night.
However, there were more questions than answers during the meetup. Granted, I should’ve raised my hand and intervened a bit more often than I did, but I took notes of the “hot topics” and unanswered questions that were raised that night. While I did my best to answer those who raised their questions after the meetup, I thought - what better way to tackle this than with a blog post? So, here it is.
Twitter Analytics vs Facebook Analytics: Reach and Impressions
The first subject that was raised was on the new Twitter analytics. With all the new metrics that Twitter introduced, two stand out: reach and impressions. When asked about what they really mean, the answer was:
“Reach tells you how many people have viewed your tweets, and that is calculated as the sum of all your followers, the people who have interacted with your Tweet and their followers; impressions tells you how many times your Tweets have been viewed by your followers’ connections.”
Wrong. Bearing in mind that the majority of people in attendance were more familair with web analytics, here’s an analogy that would’ve worked:
“Reach and impressions are to social analytics as visitors and pageviews are to web analytics.”
Short and sweet. Later, a debate ensued on why you can’t directly compare Facebook analytics with Twitter analytics, because of the Facebook algorithm and the inherent features in each platform. So, for instance, while Twitter shows you tweets in real-time, Facebook filters what content you see in your News Feed.
I understand where some of this confusion comes from. Facebook analytics give you the exact number of people reached (reach) and the number of times your page and its content have been viewed (page impressions and post impressions). However, until July 2014, Twitter didn’t really help much with these metrics - in fact, they didn’t show those metrics at all (with the exception of Ads Analytics). Due to this, a lot of marketers and analysts resorted to 3rd party tools to calculate an approximate reach for them.
Unfortunately, not all of these tools calculated reach in the same way. There were mainly three formulas used: under-calculation, approximate calculation, and the most common one - over-calculation. Over-calculation required the sum of all your followers, the people who have interacted with your Tweet and their followers. This is statistically incorrect, as it’s based on the wrong and overly-optimistic assumption that all of your followers and the followers of people who have interacted with your Tweet are online at the same time, looking at your content. (I’m not going to get too much into detail, but you can read all about it in this blog post.)
Due to this, it was almost impossible to compare an exact Facebook reach with just an approximate (and exaggerated) Twitter reach. However, that changed in July, when Twitter launched the second iteration of their Twitter Analytics dashboard on analytics.twitter.com. (You can read all about it here.)
With the introduction of Tweet impressions, you can now compare Twitter analytics to Facebook analytics. Yes, Facebook has an algorithm that filters what content you see in your News Feed, while Twitter doesn’t (well, at least not yet). However, a diligent analyst knows how to safely compare two similar metrics from two different platforms. Comparing how your impressions are increasing on Twitter while they’re dwindling on Facebook is a great start for a case to ask for a bigger “social marketing budget”; comparing how your impressions differ between Twitter and Facebook can help you craft two separate but equally effective content strategies for the two platforms (because what works on Facebook won’t automatically work on Twitter, and vice versa).
If you’re using a 3rd party tool for your analytics, don’t worry - they’re gradually adapting to this change, adding the new Twitter metrics to their platform. The first analytics provider to have done so right away was Simply Measured (you can read more about their clever workaround here).
Campaign Tracking on Social Media
While not everyone was familiar with social media analytics, there’s one thing we all agreed on - most campaign tracking on social media sucks. Heavily!
Whether you’re a Google Analytics power-user, or perhaps you only use it occasionally, you know that campaign tracking can be a lifesaver to manage your owned traffic. You wouldn’t send a monthly newsletter without properly tagging your links, right? You wouldn’t advertise your site on a magazine or a TV ad without campaign tracking, right?
Campaign tracking helps you assign traffic to your specific campaigns. Failure to do so means that not only will you not be able to properly segment your owned traffic, but you’ll also make it difficult to get actionable insight from your campaigns. (And you’ll also make a social analyst very unhappy.)
At the moment, Google Analytics lets you apply up to 5 campaign tracking parameters. For instance, if you’ve ever clicked on a link that was shared via Buffer, you’re probably familiar with the following pattern:
If you’re wondering what all that means, here’s a quick breakdown:
- utm_campaign: campaign tracking: this lets you group all the links from one campaign in your analytics tool. Imagine all your campaign links in one big folder - the campaign tracking name is the name of that folder;
- utm_source: source tracking: this helps you track which website is sending you traffic. So, going back to my Buffer example, I can see that this link comes from Twitter (utm_source=twitter.com);
- utm_medium: medium tracking: this lets you track the marketing medium the links is featured in. So, utm_medium=social tells me that this link was shared on social media, while utm_medium=sms tells me that this link was used in an SMS marketing campaign;
- utm_content: content tracking: if you have more than one link pointing to the same destination, this tag lets you differentiate the links or entry points bringing traffic to that destination. So, if your homepage has two links to your contact page, yet one is in the sidebar, one is in the header, and one is in the footer, you may want to use utm_content=sidebar, utm_content=header, utm_content=footer respectively. This is commonly used on pay-per-click (PPC) ads and A/B testing;
- utm_term: term tracking: while you might not see this often, this parameter is used to track what keywords you’ve paid for in a pay-per-click (PPC) ad.
As for the ampersand (&), that’s only used to join these parameters together, hence why you end up with long strings of parameters in a URL.
So, what’s the problem? It’s twofold:
- A lot of social campaigns are often carried out without thinking of these parameters. Sure, you have Google Analytics on your site, and sure you can have a referrals report through Google Analytics, to see which sites are referring traffic to your own site. However, this isn’t always effective - what if someone clicks on your link from a Twitter client (e.g. Tweetdeck)? What if someone cicks on your link from an email client, or a Facebook client etc.? They’re not clicking from a website, so they’ll normally show up under “direct traffic”. That aside, campaign tracking makes it infinitely easy for you to see the impact of your campaign on your website;
- A lof of social publishing tools aren’t conducive to writing parameters properly. So, for instance, you’ll have tools that promise to shorten and schedule links for you, when instead their default parameters aren’t helpful when it’s time to report on traffic. This will then leave you with unhelpful and ambiguous parameters.
If you need help in creating custom campaign parameters for your links, Google Analytics have a handy tool that can help you with that - click here to access it.
“How do you calculate social ROI?”
This is the jackpot question. This is still a grey area for many: while a brand’s ROI is easily measurable, social media’s ROI seems to raise more questions than answers. A lot of people know how it works theoretically, yet fail when it comes to putting it in practice.
ROI stands for Return On Investment - what do you get back from everything you’ve put in? This could be for a campaign, an ad, or your overall social strategy. So, if you pay £500 for a promoted post on Facebook, how much do you get out of it at the end of the promotion?
The main problem arises when you try and fit in more factors, factors that are peculiar to social media: what about influence? What about advocacy? What about engagement? What about new fans and followers? What about employee advocacy? How do you factor all of these metrics into ROI? What’s their monetary value? This is where a lot of marketers get stuck.
Some have come up with their in-house formulas to resolve this, attributing a monetary value to each of those elements. Some tools can do this for you, notably Adobe Marketing Cloud and Salesforce1 ExactTarget Marketing Cloud. These tools can show you how much money your followers’ traffic is worth. You can then calculate the average value of an influencer, which may differ from platform to platform.
Often the need to attribute a monetary value to all these social features and attributes stems from the need to speak in the same language as managers and directors: money, money, money. Sure, your campaign had a great clickthrough rate, the engagement was incredible, the positive sentiment went through the roof, and the fanbase growth was exceptional - but how does that translate into money? All these social attributes mean a lot to those who work in social media, but often to justify investing in social media, you’ll need to translate your efforts into “money talk”. “Sure, social media is great, but will it help me make more money or will it only cost me money? Either way, how much money are we talking about?”.
Targeting on Social Media
“I heard a rumour from a friend of mine that you can target social posts and even ads by email - he didn’t go too much into detail, but I’m definitely interested - can brands and marketers do that?”
The room was quiet, and that’s when I knew I had to put down my bottle of cider and raise my hand up.
I confirmed - yes, you can target by email, but not how you think. I quickly touched on the subject email targeting and lookalike audiences on social media, using Facebook and Twitter as examples. At the end of it, it seemed that I had raised more questions - “so you mean people can target me if they have my email address? Doesn’t that go against privacy and ethics?”
Let me clarify with the following example.
Say you have a newsletter, and with it you have a CRM with all the email addresses of people who have explicitely chosen to to receive emails from you. While you cannot share email addresses for privacy reasons, your CRM will safely encrypt each email address into a 32-character hexadecimal string. This string is called an email hash. Twitter will then match your email hashes against their list of Twitter accounts. You can then target these matched accounts for Promoted Tweets. These people won’t see more Twitter ads - if anything they’ll see better targeted ones. These people can always opt-out of seeing these ads by unchecking the “promoted content” field in their accounts.
You can do the same with Facebook too, although the options will differ from the ones you’ll find on Twitter Advertising.
Email hashing isn’t a new practice; in fact, hashing dates back to 1992 - just google MD5 hash.
Right after that, someone said,
“Wouldn’t it be cool if you could pick a person and tell their personality based on what they post and their online behaviour? Is that even possible?”
The answer is yes! There are tools out there that can do just that; besides, you also have other tools that have great psychographics integration. My personal favourite is Brandwatch, a powerful social listening tool that integrates social analytics and psychographics. With it I can tell what someone is into, based on who they follow, what they talk about, and their general online (and social) behaviour. Additionally, as a marketer, I can also tell what the followers of my brand are interested in - ideal if I want to tailor my content to what will resonate better with my audience.
That aside, if you’re looking for a specialised tool that lets you target and market based on psychographics, have a look at Certona and Monetate.
“Social media sounds so interesting, but with all these social networks it can be quite confusing and quite difficult to keep up with, especially when it comes to analytics - is there a tool that can join all these analytics together?”.
Unfortunately the gentleman’s question was quickly shot down by a resounding “no, there isn’t, it’s impossible, and even if it were possible it would be futile”.
At the end of the meetup I quickly approached the guy and introduced him to two of my personal favourites in this area - Simply Measured and SumAll. I told him that YES you can have a tool that aggregates all these data points for you, but NO you can’t expect these tools to do everything for you. What I mean is that, whether you’re an analyst or not, it is your due diligence to still look at the data - aggregated or not - to make sure you’re doing a good job with your social marketing.
When’s It Best Time To Post?
Mid-cider I heard the following:
“Post outside of working hours and during lunchtime, for people who want something to read during their lunch break…”
I personally have a problem with generic recommendations on best times and days to post, as well as “scientific guides on best and worst times to post on social media”. Here’s why: just imagine if we were all to post at the same, on Facebook, altogether. That “best time” would soon become the worst time to post - ever. Facebook would then filter out all the content that it deems “uninteresting” for users, and your post reach will quickly, but surely, go down.
There are best times to post for a brand, and that time differs depending on industry vertical and brand. Ideal times differ from platform to platform too - perhaps your Facebook fans are more engaged in the evening, while your Twitter followers are more engaged during the day. So many factors can come into this, but the main one is your audience - where is your audience? Perhaps you’re based in the UK, yet you have a big following from Australia; or perhaps you’re based in New York, but your Facebook is getting a lot of attention from a lot of Canadians.
Luckily most platforms, including Facebook, Twitter, Google+ and YouTube, offer insights into demographics, so instead of making educated guesses, you can make a data-driven decision as to when is the best time for you to post on what social network.
Someone mentioned sentiment, which prompted me to put my bottle of cider down and examine the opinions of the people in the room around this topic. The general consensus is that social sentiment analysis is interesting yet impossible to measure.
While I could go on and on about this topic (no honestly, I could), I’m just going to leave you with this post I wrote not so long ago on the subject of sentiment analysis in social media: here.
“Content is King”
Shortly before ending the meetup, I heard those 3 words together.
“Content is king”.
This expression was great the first time I heard it. Maybe the second time too. It stopped amazing me the third time onwards.
This usually comes with the variation of:
- Content is king, and context is queen
- Content is fuel
- Content is fire
(For more associations, just search for “content is” on Google Images.)
Personally, I don’t believe content is king. Sure, I understand the perspective of those who use this adage and actually understand it - “content is king” justifies what we do as marketers, whether we’re social media managers, analysts, advertisers or PR professionals. However, content isn’t everything - content is what happens in conversations.
Author and activitst Cory Doctorow puts it nicely,
“Content isn’t king. If I sent you to a desert island and gave you the choice of taking your friends or your movies, you would choose your friends…. If you chose your movies, we would call you a sociopath. Conversation is king. Content is just something to talk about.”
…and that couldn’t be further from the truth. Content helps you build your portfolio, but conversation makes your marketing go round. Because if you have some great content to share, yet no one shares it, what happens to your great content?
Conversations are what make social media what it is today; the constant change in how we carry conversations is the major reason why social media continually evolves. So, if you have great content and you also have conversations around it, your content may lead to relationships, which may then lead to acquisitions (which will make everyone in your company happy). If you have great content and there’s no conversation happening, you’re just broadcasting and/or advertising - nothing more, nothing less. Sure, you’ll have an audience, they’ll read your content and perhaps they’ll remember it someday down the line… but if no one shares it, if no one talks about it, that’s where your content stops - in your audience’s memory. Throw conversations in the mix, and you’ll see your content getting a new boost, something that only conversations can give it - your content hasn’t changed, it’s still the same, it’s still great, but it now has a life of its own.
So please, next time you find someone saying “content is king” as a buzz-phrase, stop in your tracks and think about it - is it? Is it really?