How to measure social media returns

“Not everything that can be counted is worth counting. And not everything that is worth counting can be counted.” Albert Einstein

Social media is a seductive tool for business. The thought of “engaging” closely with customers and prospective customers warms the most unromantic business heart. And that is why investment in social media keeps on rising.

However, social media seems to be different from other digital channels. Digital is completely measurable isn’t it? Well no; but social media is particularly hard to measure.

Once upon a time, social media evangelists used to say: “Measurement is simple. Just look at the number of Facebook Likes you have. See how many of these people convert into sales. Divide the value of those sales by the number of Likes and that’s your ROI.”


That assumes that the act of Liking results in the sale. And that is pretty questionable. (After all perhaps they Liked you because they were intending to purchase anyway.)

So if that isn’t a valid metric, how can you measure social media?

I think the answer is to say that there are a number of different types of metric, and some of these can be have monetary value attached, while others can have non monetary value attached.

(I know most digital commentators hate non-monetary metrics but the reality is that things like brand preference and share of voice are useful as they provide indications of success and failure which can be acted on.)

I am going to propose six different types of social media metrics:

  • Vanity KPIs
  • Indicative KPIs
  • Brand KPIs
  • Strategic KPIs
  • Marketing KPIs
  • Sales KPIs

Let’s take a look at each of these.

Vanity KPIs

Vanity KPIs are of little, if any, value. They include:

  • Followers and Fans. Many of these people will be irrelevant to you; perhaps they interacted once (typically this is how often they do interact) and then forget all about you. (If you think your Facebook Fans are constantly interacting with your brand then why is it that engagement rates almost always drop as Fan numbers grow?)
  • This metric represents an “opportunity to see” (OTS) a post; but only in an unsatisfactory way as the OTS represents accounts who have access to the post not accounts who have any real chance of seeing it. (Take Twitter. On average each Twitter account follows 100 users, tweets 5 times a day and is on Twitter 5 minutes a day. That means that in those 5 minutes, the average user will have had access to 500 tweets posted that day – they would have to read two each second to read them all!
  • Trivial engagements – Likes, Favourites. These are easy actions that mean very little to the user and so have little value to brands

Indicative KPIs

Indicative KPIs are a bit more useful. They tell you that things are moving in the right (or wrong) direction but not much else. They include:

  • This is rarely accurate but a “blip” in sentiment may indicate something of importance; and of course it can also be useful to compare your sentiment with that of your competitors as this may tell you whether you need to take extra efforts to out compete them
  • The total number of weak engagements – Shares, Retweets. There is some small benefit in cementing a relationship and driving visibility through these engagements, although it hard to put any real value on them
  • The total number of “strong” engagements such as comment and posts mentioning the brand: these are likely to be of greater value than weaker engagements but it is difficult to put a credible value on them
  • Website traffic from social media. Likely to give you a warm feeling if rising, but of little value in itself

Brand KPIs

These are KPIs that indicate brand support; unfortunately some of these have to be measured using primary research (e.g. surveys) so they may be hidden unless budget is available. Examples are:

  • Positive brand perceptions e.g. thought leadership (“sentiment” won’t be a useful proxy for this)
  • Claimed purchase propensity
  • Number of Follower/Fan “influencers” (i.e. people with a certain number of followers) including the number of influential prospects/customers and the number of media influencers (journalists, bloggers)

Strategic KPIs

Strategic KPIs are particularly interesting. These KPIs can be extremely important but as they don’t relate directly to marketing investment or sales success they are very difficult to put a value on. That doesn’t mean they should be ignored. They include:

  • Competitor benchmarking parameters such as Share of Voice and comparative sentiment
  • Keywords around satisfaction as these can indicate areas for new product development as well as helping the development of appropriate advertising messages and search terms; particularly interesting are:
    • Keywords related to dissatisfaction with competitors
    • Keywords/topics that generate engagement
    • Keywords in any comments

Keywords themselves are not strictly “measures” but the depth of insight they provide can be measured through volume and number of categories in any one period

Marketing KPIs

Marketing KPIs are related to marketing success (or failure) and can often have a monetary value attached. Some of these parameters will be related to the website traffic generated from social media including:

  • The cost of generating the equivalent amount of traffic volume via paid search
  • The value of non-sales conversions (e.g. product sheet downloads) and leads (e.g. leads from email submission); to create a value these will need a further calculation based on the average number of sales that are made following a non-sales conversion or lead
  • The value of back-links achieved from social media; again there will be a need for a further calculation based on the way that the value of SEO activity is measured

In addition there are some “pure” social media parameters that can be related to marketing:

  • The number of “qualified” Followers/Fans (i.e. people identified as prospects) who can be given a value based on average conversion rates
  • The number of people identified as being dissatisfied with competitors; these can be treated as leads and over time a value based on average conversions can be established
  • The number of contacts with customers that have arisen as a direct result of social media activity; these can potentially be valued using the average cost per customer contact for channels outside social media

Sales KPIs

And finally there is a KPI that can be directly linked to sales: the value of sales that have been converted directly from SM leads. This is of course where the eyes of many businesses will focus. And it is undeniably important.

A caveat though. Social media is just one channel among many. All marketing touch-points in a customer’s journey to a purchase (search, advertising, email, social) will contribute to making that sale, not just the last one.

Just as it is wrong to ignore the contribution of social media when the final touch-point is an email or a search result, so it is wrong to ignore the contribution of email and search when the final touch point-is social media.

Any sales KPI therefore should accept that and follow the attribution method you use for assigning value to other channels.

It’s not easy. But then, who said it would be! This method of tracking social media returns won’t be 100% accurate. But it will at least be more accurate than a method that says that the only value is in a sale made as a direct result of a social media contact.


Measuring engagement

Many years ago, the on-line marketing industry shot itself in the foot by proclaiming that the web was the only truly measurable medium. It demonstrated this by showing how it was possible to track the viewing of an on-line advert through to a sale.

Great. But marketing has always been a bit more complex than that.

For a start not all media activity needs to end up with a sale in the same medium. And indeed, not all marketing activity needs to end up with an immediate sale (although mostly it will want to deliver a sale at some point).

But the claim had been made. And so a lot of people still expect digital campaigns to lead directly to a measurable increase in on-line sales.

In fact we need a more sophisticated way of measuring the effectiveness of on-line campaigns.

Even the “simple” business of sales attribution isn’t so simple. Many companies still count the last click as being responsible for a sale. And yet the sales journey might have started offline with a TV ad, moved on-line via some video ads, progressed with a search of an advertising strap-line that led to a website, and then ended with a search for a brand that led to the same website and an on-line sale.

Companies like MC&C are experts in attributing sales value across different media. But there is an opportunity to go further. We can also look at measures of effectiveness that are “softer” than sales.

I am talking about brand “engagement”. Measuring this is important for many marketing activities.

For a direct response campaign it is useful to measure engagement as a proxy for value created elsewhere – off line sales or sales at some point in the future.

For TV, press and on-line video ads, effectiveness can show the value of campaigns that aren’t designed to lead to immediate sales. And the same is true of PR and event marketing.

So how can we measure engagement? Well it depends on what we want to measure: brand awareness, brand perceptions, intentions, behaviour? And it depends on the medium.

For offline media it can be difficult to go beyond established methods of pre-and post surveys and surveys looking at brand recall and favourability. There are potentially ways of going beyond this though:

  • social media analysis can give clues about audience reactions, although this can be time-consuming to do properly
  • website analytics may disclose search terms that seem to be related to an offline campaign
  • eye-tracking could be used to identify the degree to which press ads or brand names and logos in TV ads hold the viewer’s attention.

It’s easier with on-line media. Where web display ads, or other marketing activity, generate a click but not a sale there are opportunities to examine a combination of page depth and/or dwell time with return rates: this is because people who stay a long time and return several times are likely to be more engaged with a brand.

More complex measures involving the measurement of accumulated indications of engagement are also possible. These indicators might include:

  • Use of search box for product searches
  • Visits to product pages
  • Downloads of product PDFs or vouchers
  • Abandoned baskets
  • Use of store locator tools

Image a visitor who clicks on two product pages, downloads a product PDFs, and then uses a store locator. They could be assigned a value of 6, derived as 2 (a score of 1 for each product pages) plus 2 (for a pdf download), plus 2 (for using the store locator). In contrast a “busier” visitor who visited five product pages but didn’t download anything and didn’t use the store locator might only be awarded a value of 5.

On-line video provides yet more opportunities. Certain measures are obvious: plays, completed plays, average length of plays, click-through rate. But it is often possible to go further. If a YouTube channel has been set up then you can look at many more indicators such as:

  • Likes
  • Shares
  • Channel subscribers and channel views
  • Feedback and comments
  • Video hotspots

In addition you can gain useful insight from identifying how the video was discovered – embedded in your website, via Google search, via YouTube search etc. And where the source was a visit to your website via Google, tracking the visit back to the search term used may also be valuable.

What would these insights tell you? On their own, perhaps nothing. What does it matter that the average visitor has an engagement score of 7? What does it matter if a video has 9900 “likes” or (beyond a trivial media value) has been shared 136 times?

But if these scores are used as benchmarks then they can indeed be valuable: by comparing data over time it is possible to track back to campaigns and make reasonable assumptions about how particular campaigns have increased engagement; there may also be potential opportunities to relate levels of engagement with sales.

Engagement is a hard thing to measure. But it does have some value. And any attempt to measure this is better that simply saying “I can’t measure it so it must have no value”. To rely solely on media that deliver hard sales data would be to miss out on many valuable opportunities.

Valuing brands through the customer experience

I was at the CASS business school Global Brand Forum earlier this week and heard a fascinating talk from David Haigh of Brand Finance on brand valuation and the new ISO standard in this area (ISO 10668).

Brand valuation needs financial and legal input of course. And it also needs a whole lot of market research such as market trends and consumer trends.

But, to get it right, it also needs a robust, quantified understanding of how consumers experience a brand.

Understanding the value of the consumer experience will involve measuring a number of “rational” and “emotional” parameters and these could include:

  • Benchmarking the utility of products
  • Benchmarking the usability of products
  • Evaluating emotional responses to brand imagery and language found in advertising, in retailers and on the products (and indeed in other touchpoints such as call centres and bills)
  • Measuring how forgiving people are when things go wrong
  • Identifying the extent to which consumers are loyal and further, are advocates for the brand
  • Identifying the extent to which the brand is enhanced by mimicking the functionality used by other leading brands (for instance does the brand imitate Amazon’s one-click purchasing or Google’s approach to search)
  • Identifying how engaged consumers are with the brand and any brand communications (including the website)
  • Quantifying the value of social media buzz (for instance are people kinder about your brand than they are about other brands?)

Not all of these are particularly easy to measure in absolute terms (although some are) but all should at least be relatively simple to measure  comparatively against competitor brands.

Companies like Amberlight ( who evaluate customer experience are not generally asked to link this to brand valuation. But without this type of expertise the values attributed to brands will surely be flawed.

Lies, damned lies and internet statistics

Sitting in the Intellect Consumer Electronics conference I heard one speaker claim that most media usage was (still) TV and then another speaker claim that people used the internet more (I think he said that people at home used the internet for 3.4 hours a day and watched TV for 2.4 hours).

I have long been interested in some of the data claimed for internet usage: it has often seemed surprising how much people use (or claimed to use) the internet – especially in the days before the big social media sites.

I am not sure where the particular statistics I saw came from but there are a number of questions that need to be asked when considering such data. These  include:

  • Do the data compare internet users with the whole population or are they based on internet users only? This is important as around a third of the UK population don’t go online.
  • How is the data gathered – is it all self reported, all of it from the same panel, or is some of it self reported?
  • Does internet use actually mean PC use?
  • How much IPTV watching is included in the web usage figures? This time is really analogous to TV viewing so maybe should be taken from internet usage and added to TV usage.
  • Does internet use reflect the time that the internet is “on” – i.e. a PC is connected to the web, even if it is not being used?
  • Is it reasonable to include activities like email, chat and messaging (akin to letter writing or telephoning) when comparing internet use with TV use for media comparison purposes?
  • Do people overclaim internet usage in order to feel better about themselves?

I am pretty sure that internet use is still growing in popularity (!) but I am not sure anyone is helped by unrealiable statistics that paint a false picture. After all TV is a very important medium commercially and planning it  closely with the Web will generally have a synergistic effect.

Apple pi?

It’s not often that Apple is out of the news. But recently a couple of quite negative stories about their mobile strategies have been circulating.

First publishing, and Apple are facing accusations of censorship. Apple, in pious pursuit of a moral web, have been upsetting some major and perfectly respectable German publishers. The story goes that not only have Apple refused to allow certain mobile apps they disapprove of to be sold via their apps store, they have also demanded a zero nipple count on content accessible via iPhones.

This, some say, leaves them open to the accusation of trying to censor printed material!

Well maybe. But remember, Apple is an independent corporation and is, surely, quite entitled to sell whatever it likes in its own shops. And if publishers don’t like that, then they need to get their publications distributed elsewhere.

The second story is about advertising. According to Addictive’s excellent weekly Mobile Fix email, Apple have effectively banned Google and Microsoft from serving ads with any iPhone apps. This means that advertisers will need to serve ads via Apple’s own iAds system. 

That is likely to be a pain for advertisers. A new system to design ad formats for; new media and ad serving processes to understand; and perhaps worst of all the danger of yet another stream of data which doesn’t use quite the same assumptions and measurement methods as other data streams.

Not ideal!

And yet does it all really matter? To many Apple is synonymous with mobile apps. But the truth is subtly different. There are other mobile platforms out there. And they are already delivering substantial numbers of apps to mobile users.

After all, while i-phones may have 25% of the Smartphone market, they don’t have 75% of it!

So if other platforms are more flexible than Apple then they have an opportunity to outflank them and attract publishers and advertisers frustrated by the restrictions imposed by Apple. And that could well lead to a mobile content market that is larger and more competitive than it would be if a single retailer has a quasi monopoly. And that would be to everyone’s benefits. Even Apple’s.

De-fuzzing social media measurement

In the bath thinking, as one does, of social media it occurred to me that measuring the effectiveness of social media is potentially a very fuzzy activity.

Putting a little structure around it might therefore be of some value.

So I thought – well, there are three main types of measurement you can consider. There’s absolute data – data that measures size.  There is trend data – data that measures change. And there is comparative data – which you could use for benchmarking.

Absolute data is factual – I have 5000 Twitter followers (I wish), or 37 people have commented on this blog post (ditto). Or 6000 people viewed my video on YouTube. Or my brand was mentioned 5000 times in forums. Etc etc. Some of this can perhaps be assigned a value in media terms. And while it isn’t always easy to assign an appropriate value (what is a Facebook fan worth?) at least you can make a start.

Trend data is also factual – but as you are comparing two sets of data, as you can with tools like Alterian’s SM2, the important fact is the change in the data. Thus if there are 10,000 positive mentions of my brand through social media in August that may or may not be good: it’s hard to know. But if there were 9000 in July and 12,000 in September I can be pretty confident that I am going in the right direction and that the value of those mentions in September is 125% of the value in July. (There are some big assumptions being made here, but the principle is I think valid.)

And there is comparative data. I might have 12,000 positive brand mentions in September. But if my competitor got 50,000 that month I’m not looking so bright!

Then there are the results of the social media conversations: these may be conversations, actions or effects. I’ll explain.

Conversations are simply mentions of your brand or your competitors  in various places – forums, blogs, file shares etc. They can be good, bad or indifferent (and you should be measuring that). At its simplest it equates to your PR agency counting the press clips.

Next there are views. Sometimes it’s possible to measure the number of people who have seen some of your social media conversations – for instance by measuring visitor numbers to your blog. You won’t always be able to measure that – but where you can this may be a helpful metric. 

Often these numbers will be small – perhaps too small to be relevant in media terms. But they might not be. If Sony Bravia gets a couple of million views of its ad on YouTube then that’s worth something. Probably more in fact than 2 million OTSs on TV.

Then there are actions. This is when people have done something – taken an action of some kind, perhaps signing up to follow you on Twitter, or responding to a comment you have made in a forum.

Again often the numbers here will be very small – and the real value may not be in terms of media but in  terms of the opportunity they bring to engage with brand advocates.

And then we have effects. This is when you can see that social media activity has had an effect on something else you are doing. For instance, and at its simplest, you could measure the effect (or at least some of the effect) of social media on web traffic by tracking people from appropriate sources such as social networking sites. Some web analytics tools like Coremetrics are able to automate at least part of this.

Other effects might only be measurable through data analysis, for instance identifying links between social media campaigns and calls to a call centre or online sales. This is the sort of analysis that direct response media agencies do all the time for their clients.

Of course the effect might well be softer than a measurable and identifable action. It may be a shift in purchase propensity or brand favourability on the part of people exposed to your social media. That’s harder to measure although not impossible using standard quantitative research techniques. (The branding effects of online display are often measured this way.)

So there you have it. A little 3 by 4 matrix that should help you put some rigour into the process of evaluating social media. It isn’t perfect by any means. But at least it isn’t fuzzy!

I’m not bitter. I’m on Twitter

Several clients have asked me recently about how they can evaluate their social media activity. I’m tweeting like mad, they say, but I’m not sure it’s doing much good. Well, unless they have thought hard about what they want to achieve on Twitter it probably isn’t doing much good. But that’s another story.

What they want (and need) to understand is how they can measure what they are doing and whether it is having any positive effect.

So how can you measure social media?

Well, the simplest thing is to monitor conversations. Personally, I use Alterian’s excellent SM2 tool (; it has a great Freemium version) but there are plenty of other tools, some free and some that you have to pay for.

You can use a tool like SM2 a little like a search engine (SM2  “listens” to conversations in social media including Twitter, Facebook, LinkedIn, thousands of blogs, Wikis, product review sites etc) .  Type in a keyword (your brand, a competitor, a concept…) and see how much “noise” it is generating, how many people are talking about it.

SM2 will generate a list of conversations about your brand, as well as letting you cut  the data in various ways such as by location, author, type, time, sentiment, publication etc.

Using this sort of tool you can look at changes in the overall “share of voice” that your brand has in the social media space. Have conversations about your brand increased during your latest campaign? Has the launch of a competing brand reduced the interest shown in your brand significantly?

This won’t provide you with an “absolute” value but it will show you whether the value you are generating is rising or falling.

The difficulty in providing a robust “absolute” value is caused by the fact that, in most social media conversations you simply don’t know how many people have seen the mention of your brand; you can only measure those people who have actively interacted with it – and these are likely to be the minority. Perhaps though it is justifiable to make an assumption based on the generalities of social media behaviour.Thus if a hundred people have contributed to conversations about your brand it would probably be reasonable to assume that two thoudand people have seen those conversations.

You can also look at the share of voice your brand has compared with competitors. SM2 is great at this but if you just want to look at blogs then blogpulse ( is a brilliant and simple-to-use tool which provides great graphing.

Again this is a comparative tool that won’t give you an answer in pounds, shillings and pence but it will tell you if you are doing better or worse than your competitors.

You can also start to measure the “sentiment” associated with the conversations. Are people saying nice or nasty things about your brand? Increases in positive sentiment within conversations is obviously a good thing! 

Measuring sentiment  isn’t an exact science (what does “wicked” really mean?) and ideally you will use a tool that allows you to customise the sentiment measurement, rather than one that is fully automated.

Another very useful tool is Search. Use Google Trends to monitor changes in search activity on particular relevant search terms. If new social media activity coincides with an increase in search terms relevant to your brand  then it is possible that the two are related.

You can start to prove the relationship between social media and search by examining your web analytics.  If you are seeing an increase in website visits you need to look for two things.

First, are the search terms being used to find your site relevant to the conversations going on in social media?

And second, how much of the new traffic is being driven from social media content such as blogs and communities? And is this traffic more (or less) engaged and valuable. 

You may find this hard to measure without the right site analytics tools and processes (but if you haven’t got them right you are probably missing a whole bunch of other stuff too!)

Take care though. It may be easy to see the traffic driven directly to your site from social media as you can see that they are arriving from blog sites etc.  But other traffic will be arriving at your site as a result of your social media conversations, but not necessarily directly from those conversations. In other words social media is probably delivering more than you can see.

Another way of valuing social media is to look at “brand participants“, people who are actively engaged with your brand online. For instance how many followers do you have on Twitter; how many fans on Facebook?

And how many people  have shared or forwarded content your content – e.g. retweeting, email this page, forward this email etc.

It may not be possible to put an exact value on these activities but again comparative values are simple to achieve and with some imagination absolute values can be guesstimated.

At the moment though the only robust way of measuring the effect of social media will be through new research where changes to brand perceptions can be linked to social media activity though the use of quantitative research tools. For instance you may be able to prove a change to ad awareness, especially when a social media campaign has been used to enhance an ad campaign.

It’s early days for social media in marketing. Attrributing accurate absolute values is difficulties and requires many assumptions. And you are as likely to undercount as to overcount! Nonetheless, there are plenty of instances when social media activity has coincided with a rise in sales or brand values. Social media isn’t just about disarming brand critics. It’s about building real brand value. Even if we can’t always prove it!