Expert's Corner

This article written in 2010

Social Media Metrics

In the good old days of the early MBAs and Madison Avenue, publicity, PR, branding, and advertising were the ways to get the word out.  And it really was a one-way street:  TV and print media were perfect for propaganda.

"Half the money I spend on advertising is wasted.  
Trouble is, I can't tell which half."
-- John Wannamaker

Everyone knew that these mechanisms couldn't really be measured or connected to sales.  In any case, PR and ads couldn't directly effect the true end goal -- positive word of mouth.  WOM was the holy grail, and it was every bit as elusive as in the Monty Python story.

Along Came the Internet

The importance of the internet is not that it provides measurability to ads.  That's good.  The power of the internet came form the profusion of portals, blogs, and customer-enthusiast sites (totally out of the vendors' control), that made the web a huge and vibrant crucible for two-way (or even multi-path) conversations.  WOM could be focused and seen by all parties.  The vision of The Cluetrain Manifesto has come true, for those smart enough to execute on it.

OK, ok, I can hear you sighing...old news.  So how am I supposed to measure all this stuff so it becomes actionable?  Let me answer your question with a question:  I've been writing this newsletter for 8 years -- why do you think it's taken so long to write on this topic?

The Three R's 

And the answer to that is... this stuff is incredibly unstable.  It would be generous to say that the best practices in this area are "emerging."   So I'll say that.

Better than asking "how do we measure..." is the question "what should we be measuring?"  Because most of the stuff that's easy to measure just isn't that important.  Does it really make a difference how many Twits are following your VP of Marketing?  Maybe.  Probably not.

So the first order of business is to get out of the Social Media weeds and look at the situation top-down.  Let's look at the Three R's of Social Media Metrics: 

  • Relevance - The first part to examine is "how relevant is this audience to you?"  There's no real point in measuring the sentiment of an audience that has grown randomly -- it won't be clear whose opinions are worth paying attention to.*  The metric here is "how closely matched is our audience's goals and interests to our target segments?"  This cannot be answered with just demographics (as you might get from MySpace or Facebook).  You'll need to survey them (explicit surveys or implicit page-path analysis) to understand your audience's goals, interests, knowledge level, and purchase intent.

    But relevance is a two-way street:  you need to measure "how relevant is my message to my audience?"  If the audience wants only to understand the product category or the plusses and minuses of different product types, sending them a bunch of "Buy Now!" messages will at best annoy your audience.  You need to measure how in tune your messages are with the audience's state of mind. Do this through surveys, page-path analysis, AdWords/Google Analytics, and A/B split testing of content.  And of course, the doomsday measure:  unsubscribes.

    Don't forget, vendors are pretty low in the pecking order of credibility.  As discussed before, the only way to climb Regis McKenna's marketing influence pyramid is to become a trusted source of objective information.  This is every bit as true for products as it is for services.

  • Reputation - Your reputation is both "do people even know who you are?" (aka visibility) and "what do they think of you?" (aka position).  The visibility metric is pretty easy to measure:  how often does your company or product name show up in the relevant communities, publications, sites, or web ontologies.

    Your audience's opinion of you is a lot tougher, because there's very little data presented as a clear vote.  Sure, there are thumbs-up/thumbs-down indicators (like Digg, StumbleUpon, or, but without careful examination it's pretty easy to misinterpret them (e.g., the thumbs-up may be about a blog post that criticizes you).  Getting a true measure of your reputation from the web requires (at the very least) some lexical analysis and fancy search algorithms.  There are some startups that have worked on this, but I don't know how successful their engines are (please email me if you have any good experience with these products, I'll add it to next month's newsletter).

    Of course, you can survey your audience for their opinions about you.  But this suffers from some huge statistical flaws:

    • Self-selection:  you are choosing specific people to ask.
    • Self-selection revisited:  they are deciding whether they want to answer you.
    • Again with the self-selection:  you don't have a way of contacting your entire segment, only those audience members foolish enough to have registered as part of your community.
    • Survey interruptus:  the survey is not part of the audience's normal workflow; they are likely to answer only under certain conditions, and their answers may not be "natural."
    • Surveyor bias:  if you ask them questions directly, the fact that your logo is on the survey page will change some of their answers.
    • Question design:  you are likely to word the questions in such a way that the results will be biased. 

    But go ahead, knock yourself out.

  • References - Ultimately, your audience's willingness to reference you as a credible, reliable vendor is the most powerful and relevant metric there is.  As has been endlessly measured by Forrester Research, the Direct Marketing Association, BazaarVoice, and others, prospects trust the recommendations of customers (even anonymous ones) over all other sources on the web or in the media.  More importantly, prospects make purchase decisions based on reviews and references.

    Since referencability is so clear and is such an immediate success factor, the priority is to measure it first.

    There's good news here:  you can also directly affect your referencability by taking care of your customers and explicitly managing their references (both in cyberspace and real life).  If you do a good job of designing and executing your references program (you do have a references program, right?), you can actually use the statistical flaws of surveying in your favor.  You can't be brutish about it, but subtle use of push-survey tactics can work to focus the customer's latent opinions about you.

    References and testimonials should be measured (mostly) by human judgment, and recorded in your CRM system (as I describe in chapter 10 of my book).  Make sure to keep the references carefully monitored so they don't get burned out by overuse.  Unfortunately, references have a staleness date and you have to manage them just as you would any other perishable inventory.


I can hear you cry, "Dave, you still haven't told me how to measure this stuff, specifically."  That's because I've run out of room.  And I'm going to keep you in suspense until next month.

Phone Us +1 650 326 2626