Four-dimensional social media analysis (no, really)


Quadrants. Marketers love ’em. Actually, I like them too. I like the way you can draw two axes and plot things on them, and get an instant idea of often quite complex issues.

I’ve been using this approach quite a lot recently to plot social media. There are many, many things you can measure once you start grabbing data. For example the Facebook insights dashboard is very rich, plus you can download the data and do your own analyses. The Twitter analytics are good too. However, if you really want to know what’s working you need to measure across channels, and across competitors. Measuring across channels means that you need to use metrics that work across all of them, to compare like for like. And measuring across competitors means you need metrics that are publicly, consistently, readily available.

So what to measure then?

Well, as I’ve already said, there’s plenty you can measure but that doesn’t fall into these categories. The most important measurements are obviously what you have decided your business needs to look at, which might be extremely specific such as reduced time to market, improved support outcomes and so on. But a good start is audience size and engagement.

Do this: identify a handful of competitors, and in a spreadsheet note down how many Facebook likes they have, and what their ‘Talking about this’ total is. Then go into Excel and plot them in a scatter chart. You can do this either way, with likes going across and ‘talking about this’ going up, or vice versa. What’s important is that you now see where you lie in relation to the competitors, just for Facebook. Your objective is to move across and up. In three months do the same exercise and you’ll see whether you’ve succeeded.

This is very basic, and I can just hear some of you out there wincing at the idea of reducing social media down to this. But sometimes you do need to distill to key metrics, not least for internal reports. Time-pressured CEOs might not want breakdowns of every possible metric. If they can just say a chart that shows you’re moving across and up, that might be enough.

Across channels, across competitors

So that’s just Facebook. Now think about plotting the other owned channels, and how audience size and engagement might be measured. For example, Twitter audience size is followers, and engagement is retweets or replies (hint: use Topsy to count these). YouTube audience size is subscribers, while engagement is channel comments. And so on. Note again that these all must be what works across all channels, and is readily, consistently, quickly available. I agree that view count might be attractive on YouTube, or loop count is impressive on Vine, but there’s no equivalent of these on, say, Facebook. You could go through individual comments for each video on YouTube, but that would take ages. And you could look at the number of views your blog gets, but you can’t for your competitors.

Three-dimensional analysis

Now you’re looking across your owned channels, and comparing them to competitors, and that’s a good start. But if you’re getting into pulling data via APIs and suchlike, you can also draw more insight and add more dimensions. For example, if you’re pulling in user data, you can identify the number of unique commentators. Change your scatter chart to a bubble chart, and now your audience size can be across, your engagement can be up, and the size of the bubble can be the number of unique commentators.

Or, if your data includes sentiment analysis, you can use that in some way. A nice way to show this could be to have engagement going across, sentiment going up, and the bubble size representing audience size. But be careful: automated sentiment can go wrong. That’s why I tend to ignore it, and just deal with the other three axes.

Can you beyond three dimensions?

Can we have four dimensions? Audience size, engagement, unique commentators and sentiment? Unfortunately not it would seem. It would be great to have a sliding scale of colour intensity for the bubbles but I don’t think Excel does this. If it does, please let me know! Also, it could just be a bit too complicated.

What about time? That’s another dimension, right? This can get quite interesting when you plot over time. You can do this in Excel using macros to go through the data but it can get very complicated and slow, plus your data has to be in exactly the right format for the macro to work. So I’ve been using a Windows macro recorder such as JitBit to update the date in a spreadsheet, grab the resulting chart, paste it into Photoshop in a new layer, and build it up that way. Then export as an animated GIF and you can start seeing the ebb and flow of how your owned channels are behaving. It’s a bit like watching one of those cool time-lapse videos of clouds scudding across the sky or flowers growing, blossoming, and dying within seconds.

This is what you can see at the top of this post. It’s from work I did quite a while ago and I think it’s old enough to share publicly now. You can see how the bubbles move around and I can tell you now that they do correspond to marketing activity. This actually goes beyond just charting using owned channels and in fact takes all mentions across all channels, so giving us an idea of where we lie in the marketplace of conversation.

In this instance I was able to show that the work I did had an impact, at least within just the social media-sphere. I’ve since used similar methods to prove similar effectiveness and actually secured more funding for social media initiatives. If nothing else, this shows that data analysis can lead to ROI. Now, let’s see if we can plot that…

#NewsSonnet 2: KKK, Rochester, NoTW, Brains

News sonnets: current affairs in three  iambic pentameter quatrains of alternating rhyme and a couplet.

The KKK has been Anonymized
Their concealment, by the masked, unconcealed.
So will disguise be, by the disguised
Unmasked? The pointed, burning truth revealed?
Andy Coulson served all five months, is freed.
Hang on. Eighteen into five doesn’t go.
Rebekaaaaaaaaah Brooks meanwhile is to succeed
Robert Thomson as News Corp’s CEO.
Niggle Farridge has a brand new MP.
Flashman vows he will regain the seat.
Milibean sacks turbulent Thornberry
For flags, a white van, and a reckless tweet.
Memory lapses, fainting fits, some pain.
It’s nice and tasty living in a brain.

#NewsSonnet 1: Sainsburys, Miliband, Putin, Federer

News sonnets: current affairs in three  iambic pentameter quatrains of alternating rhyme and a couplet.

So Sainsbury’s is selling fruit and veg
By exploiting World War One. Still, it pays.
What next? It’s just the thin end of the wedge.
The Holocaust to sell us holidays?
Talking of wars it’s getting very cold
As Pooty-poot expands his manly chest.
This posturing is already quite old,
Quick, fetch the ageing premier a vest.
Ed Miliband is hanging out to dry
On a lightweight night-time ITV show.
The mansion tax you see is far too high.
It’s Pure and Simple Ed, you’ve got to go.
Federer could brave such an attack
But can’t because he’s got a dodgy back.

Data, you need

This is a cross-post from Ranieri Communications…

Actual output from one of my dashboards

Have you seen Particle Fever yet? If not, you should. There’s a seminal moment when, on achieving collision, a Cern star states triumphantly: “We have data.” It’s the point at which the theorists craned their necks eager to see what the experimentalists could actually prove. Suddenly, this wasn’t theory any more.

If you’re in any way serious about your social media, you need to make sure you have data. Without data you don’t know what the current situation is, so you can’t measure where you’re heading, so you don’t know whether or not you’ve been successful. You need data to know whether your strategy is working.

What data exactly? Well, that depends on what you want to achieve. Say you want to use social media to improve your SEO. What makes you think you have a problem with SEO in the first place? What needs fixing? Better find out first, because that’s how you’re going to measure success. Or perhaps you want something more qualitative around reputation management. How are you going to quantify this? Where are you going to get the data from?

There are three approaches to getting data depending on how much time, expertise or cash you’ve got: manual, semi-automated, and fully automated. Here’s a quick rundown of each.

Manual: get typing

Everyone loves a spreadsheet. They’re amazing things and you can go a very long way by manually entering data that is publicly available and then drawing insights from it. The key here is to use data that you can compare like-for-like across social media channels to get an idea of how they’re doing. So, while Facebook’s dashboard for example is rich in data, and you should certainly be using it to improve your performance, a lot of the analysis isn’t available for other channels such as Twitter, or Instagram, or your blog.

At the very basic level, you can look at two essential metrics that work across all of social media: audience size and engagement. The audience size is the total potential audience you could reach with your message, so that’s fans of your Facebook page, followers of your Twitter feed and so on. Engagement is when people actually do something in response to reading about you, so they retweet you or they comment on your Facebook page.

Do this for your competitors too, build this up over time and you can start seeing patterns in the data. You’ll see spikes that correspond to activity, and how to develop more advanced metrics off the back of these. How about dividing engagement by reach to get insight into how engaged your audience really is? How about adding frequency so you can start forming an idea of tweet quality? How about requesting access to the client’s Google Analytics and looking at how social media referrals to the website are behaving? Develop your own charts, stamp them with your logo, and you’ve got a bespoke measurement system. Port this to an online resource such as Google Docs, and you’ve got an online dashboard. Nice.

Semi-automated: learn APIs

If you’ve got an in-house geek (the one you keep in the cage in the corner and occasionally feed with Haribo) then they might like this: you can start getting involved with Application Programming Interfaces (APIs).

An API grabs the data directly rather than going through the manual procedures. So, by using the Twitter API you could directly interrogate the Twitter database and get follower figures, retweets, times of tweets and so on delivered direct to your machine rather than having to input it manually. You can also use the APIs of other social search engines such as SocialMention and Social Searcher that do a lot of the grunt work for you, by searching across multiple social media sources and aggregating them.

So, by downloading the results of API calls, you build up a store of data that you can then aggregate and analyse, again in Excel. With a canny combination of download managers, batch files and macros, you can do this all with just a couple of keystrokes.

The difference here is in quantity and types of data and therefore insight: you can accrue literally thousands of data points detailing who said what, and when, and you can start understanding who your influencers are, and what your issues might be – plus those of the competition and therefore the industry at large. At this point you really do start understanding the landscape.

If you have a smattering of statistical knowledge you can also start charting the ebb and flow of debate. Moving averages show the underlying trends. Crossovers of moving averages are highly significant. And so on.

Fully automated: bring in the Big Guns

If fully manual requires investment in time and semi automated needs investment in expertise, then fully automated is the money play. Here, we’re talking systems such as BrandWatch with millions of sites categorised, crunching huge amounts of data using dedicated server farms. It’s the rocket science approach and while this is mostly the domain of large companies that provide consumer services such as telecoms companies, there’s also a strong argument to be made that smaller agencies can use them profitably by sharing the cost across several accounts.

Hands, APIs, BFGs: Which one’s right for you?

If you’re not storing and analysing any data currently, then you need to start, right now.

At the very least start recording reach and engagement, ideally alongside competitors. It’s a useful exercise as of itself because you really start to understand cause and effect, and get to grips with the concepts.

When you get the hang of that, and you’d like to dive deeper, see if you have a geek in your organisation, or a latent geek, or know someone who keeps one. They might be able to ramp you up to the semi-automated solution and then you become something of a social media data guru.

And when you’re finally seeing the shiny green numbers coursing through the very fabric of the Matrix itself, and you’ve landed that major social media account – or you’re a postdoc working at Cern – it’s time to hoover up as much data as you can possibly get your hands on. Even if you don’t uncover the secrets of life, the universe and everything, you’ll know what drives conversation, and that’s a decent second.

Time to saddle up again: Brandinnnnnng

Right. So, the past few years has seen a lot of water under a lot of bridges. I’ve been a social media consultant. I’ve been a copywriter. I’ve been a marketing manager, PR manager, social media manager and content creator all at the same time. I’ve watched c-beams glitter in the dark near the Tannhäuser Gate. I’ve been undressed by kings and I’ve seen some things that a woman just ain’t s’posed to see. I’ve quoted far too much from films and songs.

And throughout all this, I’ve completely and utterly lost the blog habit. I couldn’t really see the point because I’ve been actually doing all this, so why bother write about it? I’ve decided to take it up again because:

  • It keeps me off the streets. Like Ishmael, it prevents me from deliberately stepping into the street, and methodically knocking people’s hats off.
  • While things got boring a couple of years back, they’re a bit more interesting and write-worthy now. Mobile has changed a lot of this game, and new platforms such as PInterest, Vine and Instagram are making a real impact.
  • Whereas I thought blogging had died, people like Shel Holtz think I’m wrong. I still think blogging is really only useful when you get comments back, rather than just a load of shares or likes, but I’m prepared to try this again and see if I genuinely do get what I would call ‘engagement’.
  • I probably should do this because I’m back in the self-employed game – or, rather, I contract now, under the guise of SuperCooper Comms. Catchy isn’t it? I thought long and hard before that one, threw away company names that sounded like cures for constipation (actually, nearly all of them were like that strangely enough), and really should get the domain pointing to this blog but I can’t be arsed.

I know, I’ve already fallen at the first blogging hurdle: I’m talking about myself rather than things you want to hear about. That’s because I’m quite rusty.

So, from now on, this blog’s going to be mostly about comms – earned, owned, paid, mobile, desktop, advertising, PR, social, digital – but from time to time I’ll put what I like on it. For example, isn’t the Philae lander an amazing thing? And aren’t otters cute?

Branding: The truth

Anyway, back to comms. Branding. That’s not comms so you could say that having blundered through the first hurdle I’m now staggering through the second. But wait. Branding is the root of comms. It informs what you’re going to talk about, and how you’re going to say it.

I’ve somehow become a branding person too along the way. What happens is, you go further and further upstream, beyond copywriting, up through social media, then messaging, then marketing, then you realised there’s more, so you start looking at branding. It’s a bit like Captain Willard travelling up the Nung River.

The key to branding is figuring out a way to make yourself different from your competitors, and suited to your audiences. It seems deceptively simple, but it’s not easy. Just Google ‘brand positioning’ and you’ll see that everyone has their own take on it, from the big stuff like what a brand is, to the nuts and bolts such as brand values, vision statements, mission statements and the like.

The important thing is that it’s the truth, which can sometimes be harsh. And as we all know, many of the truths we cling to depend greatly on our own point of view. So you need to get a lot of points of view.

The Outside-in view

  1. Choose a handful of key clients, partners and friendly peers.
  2. Ideally call them, not ideally mail them to say something like “We’re doing some thinking about our brand and would really value your input”, arrange a call for, say, 30 minutes or so.
  3. In the call, ask them:
    • What do you think we do well?
    • What do you think we could do better?
    • What three words describe how we work?
    • What do you think you’ll want from us in five years’ time?
  4. Create a quick slidedeck breaking this out into the core (ie what do all of them say), the common (what do most of them say) and the unusual (are there any that stand out for saying something the others don’t).

The Inside-out view

Do this internally too. Make people feel this is something they contributed to so that it becomes ‘real’ and ‘owned’. Also by comparing your responses to the clients, you start to identify gaps.

There is one additional question here too, for the MD or CEO, and maybe your senior leadership team, which is: where do you want to be in five years’ time?

The Industry view

  1. Spend an hour looking for articles that talk about what the future holds for your industry or sector
  2. Pull out the most significant summary points into a slidedeck
  3. Go through them and identify the top three

The Competitor view

  1. Spend an hour going through the websites for your competitiors and figure out:
    • What is their SEO description (if any)? Right-click on their home page where there is white space, and then click ‘View page source’ and search for the text <meta-description>
    • What is their strapline (if any)? If not under their logo, then sometimes they’ll have a sort of strap line on the top of their ‘about’ page.
    • In their ‘About’ page, what are the top three things they talk about, ideally relating to how they work rather than what they do?​
  2. Create a quick slidedeck breaking this out into the core (ie what do all of them say), the common (what do most of them say) and the unusual (are there any that stand out for saying something the others don’t).
  3. Email your team and ask them to give them a score of 1 to 10 where 1 is least and 10 is most, for the following:
    • Impact
    • Sophistication
    • Contemporariness – awful word I know, but what I mean is, how up-to-date do they look?

Web view

Create a ‘taxonomy’, by going through everything you can think of – websites, social media, wikipedia etc – and create a cloud of the most commonly used words in your industry.

Bringing it together

  • The outside-in view is the most important, by far. It’s your opportunity for you to find out, genuinely, what your strengths and weaknesses are.
  • The inside-out view is quite important, so long as you trust your staff not to be too blinkered in their outlook. By comparing this with the outside-in view you get an idea of whether or not you’re deluded.
  • The industry view shows you what the main challenges and opportunities are. Now that you know your strengths and weaknesses you can start developing a position that plays to your strengths, while addressing any potentially disabling weaknesses.
  • The competitor view gives you insight into how to differentiate. Do any of your competitors mention the industry challenges or opportunities? If not, they’re not branded very well, and you don’t need to worry about them too much. If some of them really do have a good position then you need to work out how not to overlap with them.

Now, you have three bubbles: your culture (the three words that describe how you work); the industry issues and opportunities they present; and your strengths. You need to find a common position that works at the intersection of those three bubbles, and doesn’t impinge too much on a competitor.

See? It’s all about bubbles. And it isn’t easy. That moment of insight can take quite some time. As is often the case, the process involves leg-work, but the insight means being in the right place at the right time, usually with a G&T on the go.

Finally, the web view gives you an idea of the kind of words you should use when you come to express all of this. You can be a maverick, and that’s fine, but you also need to make sure the words you use, are the words people are looking for.

So it’s about words too.

Good luck, and hopefully it won’t be three years before I post again…