Assessing Kredibility

Greg Stevens examines whether the Kred online influence scoring system is as credible as it seems.

The entire notion of measuring “online influence” has been an endless source of speculation and hand-wringing in tech magazines and the blogosphere over the past few years. What does it mean? How might it be used? How much should it matter? Should it matter at all? Because Klout arrived early on the scene in social influence scoring, and is currently the leader in that market, most of the criticism has been directed at that company.

Is Klout accurate? Can one’s Klout score be manipulated? What are the appropriate ways that companies might use Klout scores – and what ways are inappropriate? Should it be used to decide who gets special discounts? Should it be used to decide who gets to board aeroplanes earlier? Should it be used to decide which employees to hire?

Reactions have been emotional and extreme. The problems have also been aggravated by Klout’s confusing marketing efforts and poorly-understood business model. Kred, on the other hand, has a methodology and a business model that appears almost immune to these criticisms. To understand why, one has to look at where Kred has come from and how it is different from the other “social influence” scores available.

The Kred scoring system was created by PeopleBrowsr. PeopleBrowsr is an analytics company that began in 2007. From the beginning, their focus was “big data”: they wanted to use advanced data mining techniques on the massive data sets made available through social media platforms. They approached Twitter, asked for, and were granted, access to the “firehose”: a direct and real-time connection to all of the public tweets on Twitter.

As a result, the databases at PeopleBrowsr have been receiving every single public tweet posted to Twitter since 2008. That amounts to more than 100 billion tweets to date, with new data currently streaming into their systems at a rate of about 2 or 3 billion tweets a week. For its first few years, PeopleBrowsr focused their efforts on solving the problems associated with handling that massive amount of data.

With a combination of cutting-edge hardware, massive parallel processing, smart algorithms and hard work, PeopleBrowsr developed a system that was able to process, index, store, and rank tweets in real time as they came into the system.

At the start, PeopleBrowsr was an analytics platform, not an “influence scoring” platform. This focus is what may give them an advantage. These are “big math” people solving “big math” problems. Their clients are brands who are looking to find out useful information about their customers (or potential customers) from the massive sea of social media chatter out on the internet.

By 2010, PeopleBrowsr was able to address real-time changes in sentiment toward brands and people, measuring the “collective consciousness” of the online world. For example, they could look at time-sensitive changes in sentiment toward brands that advertised during the Superbowl, or toward political candidates involved in an election.

For companies that approached them as clients, they could generate customized reports that included information such as audience profiles, topic trends, website activity tracking, and even regional information for conversations about their brand. PeopleBrowsr provided an entire suite of analysis and reporting services, all built on the infrastructure of their “big data” engine.

Then, over the course of 2011, they developed Kred. PeopleBrowsr envisioned Kred as a tool for their clients: the companies and brands that came to them to understand their customers. To be useful as a marketing analytics tool, Kred needed to have certain features. It needed to be transparent, so that there was never any doubt about what was being measured or why. It also needed to be able to separate social influence from social outreach.

Kred defines “influence” as a measure of what other people do because of you: your ability to spread the content that you create. The complement of that is “outreach”, a measure of your generosity: the amount that you share or engage with other people’s content.

Any time someone retweets or replies to you, you get points that contribute to your influence score. Any time you retweet or reply to someone else, you get points that contribute to your outreach score. Some things matter more than others: for example, if you are retweeted by someone with more than 10,000 followers you get more influence points than if you are retweeted by someone with fewer than 10,000 followers. The exact details of which events give you how many points is all spelled out on the “rules” page on their website.

More recently, Kred has also added the ability to include Facebook interactions in the score. The basic concept is the same: certain things add to your influence points (for example, when someone mentions you in a post) and other things add to your outreach (for example, you post on someone else’s wall). Some actions add more points than others, and the details are all spelled out on the “rules” page.

Kred has the ability to extend even further, using the same basic concept of “points.” One feature that is in “beta” on their website is the notion of “offline Kred”: factors that demonstrate influence not in the online world. You can upload certificates, awards, degrees, and other documents that show that you have influence, and the Kred team assigns influence points to that accomplishment.

After all the points are added up, a person’s final Kred influence score is “squashed” using a logarithmic function – which is also described on the “rules” page. By taking the total and running it through that logarithmic  function, Kred is able to produce an influence score that has an upper limit (which happens to be 1000), and that becomes harder to improve as your score climbs.

In January 2012, its public-facing website was finally launched. So you can now go to the Kred website, link your Facebook and Twitter accounts, upload “offline Kred” documents, and participate in communities.

A quick look at the Kred website highlights a few differences between Klout and Kred. Whereas Klout updates your score once per day, Kred is updated in real-time and can even show you a running leaderboard of every social media transaction that has impacted your score. Where Klout doesn’t tell anybody how their influence score is calculated, the Kred score is completely transparent: they even have a “rules” page showing exactly what factors contribute to a person’s score and how much they matter.

Klout provides a single score; Kred provides separate scores for “influence” and “outreach,” and also can calculate your score within separate interest groups or communities. These differences don’t necessarily mean a lot to the average Twitter user.

For example, unless you are unusually compulsive about your narcissism, you probably don’t need to watch your own influence score change in real time. However, real-time updates can be very useful for brands, especially during large public events that might be generating publicity for them. As a demonstration of their real-time capabilities, Kred has successfully tracked the changes in influence during a number of major public events, including the Grammys, the Oscars, and South-by-Southwest.

The distinction between “influence” and “outreach” might not be particularly exciting for the average Twitter user. Apart from a slight nerdy satisfaction you might have by being able to assign yet another number to yourself, knowing your “outreach” score as distinct from your “influence” score doesn’t have much use.

Yet, for brands, the difference is critical. If someone has a high influence score, that means the content that they provide has a large impact on their network; but if their outreach score is low, it means that they are not that likely to respond or share things from other people.

Someone with a high outreach score is probably more likely to respond to a company that reaches out to them and asks them to answer questions about their brand, or to talk about their brand. When looking for possible brand champions, a company might want to filter their options by outreach first, and then examine which customers have the most influence only after that.

Another critical feature of the Kred platform is the fact that they can identify communities and calculate scores for influence and outreach within particular communities. Communities basically work as complex and intelligent filters that only take into account points from tweets and followers that are relevant to a particular topic, brand, or group.

By looking at keywords in profiles and tweets, and examining patterns of followers, Kred is able to use advanced filters to “re-score” your Kred based only on those tweets that are most important within a defined community.

The concept of having “community scores” overcomes one of the biggest intuitive criticisms of Klout and other “online influence” scoring systems: what can be called the “answers versus LOLcats” problem. Anyone who is active in the Twitter world knows that there are at least two types of content that get massively shared and talked about.

On the one hand, there are people who are actually making meaningful contributions, such as linking to a website that answers a question that is interesting to a large community. On the other hand, there are people who share cute and funny pictures of animals. How can Klout, or other scoring systems, tell the difference between a technology guru who provides popular answers to technology questions, and a popular LOLcat distributor?

Kred solves this problem with communities. Both of these users may have a high “global Kred” score, but only one of them would have a high Kred score on the “technology” community. (The other, presumably, may have a high Kred score in the “cat” community.) Similarly, although Justin Bieber may have a very high global Kred score, his Kred score in (for example) the “sports” community may be much lower.

This feature is especially important for brands who are looking for a particular market segment that they want to target with their advertising. For example, suppose a sporting company approached PeopleBrowsr and said: We want to find people in Detroit who are interested in junior college basketball.

PeopleBrowsr would be able to create a custom community for that client, and score every single Twitter user based on their influence on that particular market segment. They would then be able to track influence and outreach in real time for this highly customized “community Kred” measurement.

Kred will be developing its public-facing website over time: for example, it will be adding the ability to link other social networks. It also has other features that you can explore on the site, such as the ability to identify community leaders or gift people Kred points. You can see the obvious differences between Kred and Klout and other social influence measurement platforms by exploring their website, but, as we have seen, the differences in features are not the deepest difference between Kred and its competitors. The deepest difference is that Kred is rooted in big data analytics.

In the online world, there is endless speculation about how brands and companies will eventually decide to make use of Klout. There are a few anecdotal claims about people getting upgrades at hotels based on their Klout scores, and a few brands sending out laughable gift-bag “Perks” to people based on their scores, but for the most part Klout seems to still be trying to figure out how to monetise itself.

One the one hand, Klout has spokespeople telling interviewers: “Klout’s aim is to help people understand and benefit from their influence.” On the other hand, their main targets for funding are the brands and companies who are still trying to figure out how to make the score most useful.

Kred doesn’t have this problem. To date, PeopleBrowsr is fully funded by the founder, and has not had to rely on external financing. Its business model is focused entirely on the analytics services that it provides to its client brands and companies. As a result, PeopleBrowsr doesn’t have to search for an answer to the question: “How can we monetize Kred?” because commercial organizations don’t have to ask the question: “What can we use Kred for?” The Kred score is, and always has been, a tool in a suite of analytics services that PeopleBrowsr provides for its clients.

In a controversial space, Kred is an interesting and seemingly credible, if much lower-profile, player. Much will depend on Klout’s ability to prove its mettle to brands anxious that a Klout score is little more than another social media vanity metric.