Analytics Is King

There has been a lot of discussion over the last few days about public relations and social media measurement. Actually, there has been a lot of discussion about public relations and social media measurement over the last few YEARS let alone days. Todd Defren got the ball rolling with his post about PR measurement fails, and using Web metrics as a gauge of success in PR. The truth of the matter is that Web metrics could be one potential gauge of success, but talking about it in a vacuum won't work. Then last night's #pr20chat discussed a very similar topic, and it didn't take long for the discussion to go slightly off track. Instead of talking about the process of identifying metrics, we ended up talking a lot about individual metrics. That, my friends, is what we call the measurement rabbit hole. As soon as you start down that path it's very hard to pull yourself out of it. By now we should all have the steps toward successful measurement beaten into our heads, but if not....

Benchmark research
Developing measurable goals and objectives
Creating a strategy and tactics
Implementing said strategy and tactics
Measuring results

There's a lot of explanation under each of those phases, but that's not the subject of this post. I know several people in this industry have tried (including the wonderful Katie Paine) to get people's focus off of one metric and onto the process above. This should be pretty self-explanatory for most of us but one metric is not an appropriate measure of success. Most often when one metric is used to define success it is impressions, and we know how imprecise that is as a metric. Why is there this over reliance on one metric? Is it just because of our hatred of math (which I think is a total cop out by the way)? Are we not spending the time to create a list of metrics using the steps outlined above? Do we not know what metrics we should be using? It is probably a little bit of everything, and that's where this post comes in... One of the things I raised in Todd's post and during last night's chat was using a series of metrics to create an index score. Now, before you start wigging out over the use of the word "index" give me a second to explain what I mean. I think if we put our minds to it and didn't cop out about our hatred of math, we could come up with a list of metrics that not only fit our particular campaign but are widely accepted as metrics in the field. This could be anything from impressions (I hate them, but you should know they are used still by many companies and are accepted), sentiment, mention prominence, spokesperson quoted, mention in a target publication, raw number of overall mentions and the presence of key messages in the articles. Those are mostly traditional PR metrics, but I'm sure you could come up with a similar list for social media. These metrics taken individually don't mean a heck of a lot, but brought together into some sort of score tell a pretty powerful picture. So how do we go about creating this "index?" The process is actually much easier than you might think...

Come up with a list of metrics - Ideally, you'll use the process outlined above to arrive at those metrics and not just cherry pick off of what you've always done or what you know others are doing. Frankly, it doesn't matter if you have 5, 7 or 10 metrics. Just come up with the right ones for your campaign.
Determine which metrics matter most - If you have a list of 10 metrics, you should think about which of those matter most as the ultimate gauge of success. Is it change in sentiment? Is it mention prominence? Whatever it is, you should identify some sort of ranking for those metrics.
Applying a weight - As I said, this isn't scary math. After you've come up with your ranked list of metrics, you should apply a weight to each of them totaling up to 100. So if you have 10 metrics the weights of those 10 metrics should equal 100. No, you can't just give everyone of them a weight of 10. :)
Creating the index - Truthfully, this is hard to replicate in a blog post using Excel but if you're using this post as a template feel free to drop me a line and I can help you setup the spreadsheet. It actually isn't hard, but I'll just confuse people trying to explain it. Suffice it to say that you should have three columns for each metric - the metric itself, the index and weighted index. Anyway, drop me a line if you have specific questions on that.
Coming up with the score - At the end of the day if this is done right every "mention" in your database is going to have a score from 0-100. You should figure out the ranges for what you'd categorize as a home run and what you might categorize as a bad hit.

That wasn't hard, right? You've incorporated a series of metrics into a cohesive PR measurement story. You can easily see now how a similar process can be applied to social media, right? This isn't tough stuff, if you build the process up front. Have you ever built anything like this? If so, come and tell us about it so we can all learn.

Hopefully you'll permit me a quick shotout to a pretty ridiculous movie with that title...You'll understand why in a minute. Defining and measuring influence in social media is definitely the hot topic right now. Heck, Jason Falls' blog covered this topic at length last week with posts about how to improve online influence measures, revealing a new online influence tool (mPact) and then providing some background on the influencer identification tools marketplace. Radian6 also joined the party by releasing an eBook on defining and measuring online influence on Monday (co-authored by yours truly).The opinions about how to tackle this problem are varied. The varying opinions about an area this (relatively speaking) new is only leading to confusion in my estimation. People are confused about how to develop a process of identifying influencers, what tools to use, what's the appropriate amount of rigor you should apply to your analysis, is there a common set of influencers we can identify for all brands (put that one to bed right now - the answer is no). There are even those who openly are wondering if influencer analysis is even necessary. Let me be totally clear -- not only is it necessary, but it is absolutely imperative that we get the analysis (and the framework) right. And by getting it right I mean some combination of quantitative and qualitative. I'm a self-professed data geek. The more data the better (as long as there are actionable insights, of course). We need data to develop strategy and then ultimately tactics that make us look amazing to our bosses. My love affair with data doesn't mean I'm going to throw qualitative analysis out with the bath water. Quite the contrary. Influencer identification needs, and is currently crying out for skilled analysts who understand the brand(s) better than any computer possibly could. What the appropriate balance is will really be dependent on the brand or agency. My preference would be close to 50/50 (I'm not going to play Solomon here), but you have to make that call based on avalaible resources, which includes budget and people obviously. I think we can all agree that this combination of qual and quant is necessary, yes? It's interesting then to follow along with chatter (aside from the posts linked above) about how lists are developed, let alone used. I don't get an opportunity to participate in Twitter chats as much as I'd like, but one of my favorite ones is #pr20chat hosted every Tuesday night by Heather Whaling and Justin Goldsborough. During this week's chat, they asked a question (posed by yours truly) about how the participants come up with their influencer lists. Check out the conversations around the topic here if you'd like. What you'll notice is that of the 10 (I'm making that up for the sake of conversation) people who responded to the question, the majority outlined different approaches. Interesting, huh? I'll grant you that there are likely to be different ways to slay the dragon, but a majority with different methods? The number of varying methods must make a client's (or boss') head spin. I know it does mine and I live this stuff... With that as a backdrop, I thought it might be helpful if I gave you a couple of "standard" things you should use for any influencer analysis you're conducting. Yes, it is my contention that some of these more "structural" things can be implemented across brands. I'm hoping you'll drop in a comment if you disagree.

Be tool agnostic - We all know that there are a lot of tools available in this space. Jason outlined several of them in the post linked to above. Don't be hampered by one tool or algorithm (unless it is one you've developed yourself). If you want to use one of the tools that are available to measure Twitter influence or blog influence that's great. But, don't get into the habit of relying on their metrics as anything more than a gut check.
Develop your list of metrics - I mentioned above that every influencer analysis is a combination of quantitative and qualitative analysis. Before you go digging for people to identify, make sure you spend a little time upfront identifying the types of metrics you want to use to judge people's influence.
Create a method for defining brand relevancy - This is absolutely critical, and it is something we mentioned in the white paper linked to above. Some of the world's largest blogs are great, but chances are that they've never talked about you. That isn't to say that they will never talk about you, but how much time do you want to spend on a site that doesn't seem to be talking about you or the subject matter you are most interested in? Probably not a lot (at least I hope).
Set your influencer timeframe - There are a lot of places where influencer analysis falls down, and the amount of data you gather tends to be the largest offender. Listen, there's a lot of "gut feeling" in any kind of influener analysis but please don't make the cardinal mistake of pulling in too little data. Spend the extra time to get six months worth of information instead of three. Your output will be significantly better.
Understand what your list is going to be used for - This is a pretty common misconception in the digital marketing world - everybody identified as an influencer MUST be an outreach target. *Cue ridiculous game show buzzing sound here.* That is just simply not true. If you have a list of 100 people, you might have 20 who are direct outreach targets, 20 who would only receive a press release, and 60 that are listen only. That might be an extreme example, but I think you get the gist. There are going to be folks who are influential to the online community that you may not be able to reach. That doesn't mean the intelligence they are providing you by talking online isn't valuable. Listen to those people's conversations, and don't worry if you can't engage directly with all of them.

Anyway, those are just some thoughts from me. I tend to think there is more confusion than ever around online influence. Hopefully through more educated dialog we will come to some sort of consensus. However, knowing the online space as I do I wouldn't look for that anytime soon...

Building Your Own PR or Social Media Measurement Index

Dazed and Confused: The Story of Online Influencer Identification

About

Archive

Get Updates

Sites I Like