August 24, 2011

Understanding How My mPACT Measures Influence

By: Patrick Hurley

Understanding How My mPACT Measures Influence

We recently talked here about how to get the best results from a search in My mPACT. As we said then, our system and methodology are quite different than other influence measurement solutions you may have encountered before, so we wanted to provide some tips and tricks for using My mPACT the right way.

In order to provide some more context to that previous discussion, we thought we would talk in more depth about our influencer identification and measuring algorithms. Specifically, we’d like to talk about methodology used by many other influencer ranking solutions – techniques that we purposefully don’t use – and then talk about the way our algorithms do work.

What we don’t do:

1. Run a popularity contest: This is worth repeating – we don’t just measure “raw” influence based upon popularity or number of followers, etc. We do, of course, take those things into account. But we feel you can’t really tell if someone is influential on a particular topic or in a particular market without isolating and examining his or her writing on that topic itself. So while we do create some general influence metrics internally (and use them in our measurements), we only present a custom-derived influence score based on the keywords you choose that match against the keywords used when someone writes in a blog, article or Tweet.

2. Try to define a topic or market for you:  We don’t focus My mPACT on any particular market segment or topical area. And we don’t think we know your topics and markets better than you do – in fact we’re sure that we do not. So rather than trying to arbitrarily group influencers together into buckets of our own making, we let you define the topic/market you’re interested in with your own keywords, and provide you with a robust advanced search mechanism if your topic is too complex or ambiguous to define with a single keyword or phrase.

How we measure:

1.  Topicality is the key: Unlike other solutions, we don’t look at just “raw” influence, but rather topical influence. So you never see a score until you’ve entered some keywords into My mPACT. Taking it further, we don’t just look at broad topics like “sports” or “technology”, but rather measure influence as granularly as you wish to search, like “NHL Goalies” or “Wi-Fi routers”.

2. It’s not just Twitter: We don’t just measure across one or two social networks (i.e., Twitter or Facebook), but rather look for influence wherever it may be exerted online. That means we include articles in traditional media (like newspapers, magazines and trade publications), in the blogosphere and also in social media.

3. What metrics we include: What we measure depends upon exactly which medium we’re looking at (i.e., blogs have different influence metrics than does Twitter, just because of the nature of the media), but generally we look for a few things. First, we measure the raw influence of a voice, based on the criteria we set for that medium (things like # of followers/retweets/mentions, circulation, blog traffic, number of commenters, etc.). Then – and this is very important to remember – we look at the frequency and prevalence of your search terms within the media we examine to determine the relevance of your voice (and other leading voices) to that topic.

4. Time is a factor: Not only do we look at the frequency and prevalence of a keyword in your writing, but also the timeliness of it. The longer it has been since you (or another influencer) has mentioned a keyword, the less weight we give to that instance when calculating influence. We strongly believe that true influencers are those who talk about a subject frequently and often – AND recently. If an influencer hasn’t mentioned a keyword in many months, we believe she or he is less influential right now on that topic. Note: the exact time weighting we use depends upon the medium in question, so we put more weight on older mentions in articles than we do in blogs, and we purposefully don’t count old Twitter mentions if there are no recent ones.

5. Who we rank: for each keyword or keyword set, we rank the top 500 influencers, based upon the criteria listed above. So if you search for yourself for a keyword that you don’t mention often, you may simply not be in the Top 500 for that topic, no matter how many followers or readers you may have. Try using keywords that you mention often and have used recently for the best results. And if you still don’t make the Top 500, notice the top 25 Influencers we provide in your search results – getting involved in conversations with these people can help raise your personal score.

If you’re looking to dig even deeper into discovering the most influential voices, our mPACT Pro product goes even further with additional algorithms including Authority scoring and other metrics designed to ensure you are receiving the best list of voices possible. Additionally, mPACT Pro lets you create multiple levels of keyword-based filters, so you can look at your market or area of interest both generally and at a very granular level. You can try mPACT Pro for a free two week trial here.

We’d love to know what you think and we definitely want to hear if you don’t get the results you expect from My mPACT. We’re always happy to help you refine your search terms and make sure you’re getting the results you’re looking for. You can contact us, including via live chat, here.

Patrick Hurley is mBLAST's Director of Marketing Communications, having previously performed various marketing and sales engineering roles at the company. Prior to mBLAST, Pat was an analyst and the Director of Research at TeleChoice, Inc. a leading telecoms strategic consulting firm and is the author of over a dozen technology related books in the John Wiley and Sons "...For Dummies" series.

More Posts by Patrick Hurley

Filed in Category: mPACT Pro, My mPACT