Measuring online influence

[This article was originally posted on the IIAR blog a few weeks back]

This second post on online influence looks at how one might measure influence using online metrics. It follows on from last week’s post which posed a lot of questions, but few answers. Fair cop.

But first, I think there are a couple of principles of influence to consider:

1. People buy people. Therefore influence measures need to identify individuals. It’s not sufficient to conclude that Gartner (for example) is influential – duh. Vendors need to know (a) who within Gartner is influential, (b) what’s their influence relative to other analyst influencers, and (c) what’s their influence relative to other non-analyst influencers. Influence isn’t distributed equally, either within organisations or throughout the market.

2. Influence is multi-dimensional. Some influencers are subject gurus, some command statutory authority, some are thought leaders and idea planters, some structure the financial elements of procurement, and so on. It’s important to understand why someone is influential, as much as the fact that they are influential.

So. Let’s look at some of the ways influence claims to be measured online:

- Citations – this measures the number of times a source refers back to an originating source. Google PageRank works this way: it rates pages highly if other people link back to it. It’s also how academic research works: a recent paper will refer to previous papers, and the more references a paper gets the more influential it is considered to be. Its strength is its weakness – it will persist in referring back to previously cited sources, even if they become superceded. It also build in something called the Matthew effect, where longevity is favoured over originality.

- Connections – how many outbound links a source has. LinkedIn, Facebook, MySpace, Twitter (following) and other social networks work this way. Count the connections to determine how well connected the person is. It’s also easy to fake, by link swaps, indiscriminate “friending” and so on.

- Subscriptions and readership – Technorati works this way, measuring the number of readers a blog has, and Twitter also publishes this information as followers.

- Noise – references to subjects and/or individual firms. Radian 6, Techrigy, and a bunch of other providers do this, measuring the number of times your firm is mentioned. Some also claim to measure the sentiment of the mention, usually using natural language processing tech.

All of these measures are indicators of online activity, and you can see the usefulness of them, as far as they go. They are, in my view, the equivalent of PR clippings services.

However, none of them measure whether the critical community, decision makers, are remotely influenced by online channels. It’s always necessary to ask: Influence on whom? Do any of these measures accurately assess the impact on real decision makers? In other words, do they measure the likely impact on behaviour of a buyer? Because if they don’t, if they measure a vague notion of industry activity or sentiment, then do they really reflect the ecosystem of influencers that impacts decisions?

More critically, can vendors construct marketing programmes around these measures to improve knowledge, lead generation and useful sales collateral? Because if they can’t, what are these measures useful for?

Tssk – more questions.That last one was rhetorical.

Next week’s post will probably pose more questions about how AR can use online channels to increase influence on their firms’ prospective customers.

Have online channels changed the nature of influence?

[This article was originally posted on the IIAR blog last month]

Determining the impact of the growth in online channels such as social media is one of the things that taxes most of us. I’m forever seeing new ‘influencer tracker’ services pop up, and in the world of analyst relations there’s continual discussion on whether and how to engage in online options like blogs, podcasts and social networking.

In response to the explosion of online influencer tracker services – there are over 100 nowadays, and counting – Nick Hayes and I wrote a paper* on how we think they are misleading marketers. The paper led to an invitation to post on the IIAR blog, to hopefully spark some discussion – thanks for the invite, Ludovic.

This first post focuses on whether influence as a concept has changed with the use of online channels. The second will look at how influence can be measured using online metrics. And the third will discuss the implications of online channels for AR and Influencer Relations professionals.

There’s an important context to any debate on influence, online or otherwise. It is that ecosystems of influencers are highly fragmented these days. Most decision makers are influenced by the traditional journalists and analysts, but also by consultants, academics, regulators, financiers, sourcing advisors, procurement professionals and other specialists, as well as peer end users.

Much of the influence exerted by this group has been enabled, in large part, by online channels. This has been an ongoing process for a decade. The web and search engines make it easier for anyone to reach the market, and easier for buyers to find what they’re looking for. Blogs and podcasts increase the reach of anyone inclined to use them. Social media is just the next step in this evolution – there’s no social media revolution going on.

But social media has provided a new channel for those people with the potential to influence, making communication between those people frictionless. To reach a group of like-minded adopters of a technology you used to have to organise a meeting in a mutually inconvenient location. Nowadays, you organise an unconference or participate in an online forum. It used to take months to organise an event, now it can take hours.

But has the nature of influence changed? Are decision makers influenced in different ways through online channels? You’d think so, given the hype, but as Nate Elliott at Forrester observed, “the huge majority of users influence each other face to face rather than through social online channels.”

It makes sense to understand the attributes of influence – the ability to discuss and persuade, knowledge and experience, willingness to express an opinion, the authority and gravitas with which to communicate that opinion, the opportunity to convey that opinion to the right audience at the right time. And so on.

Some of these attributes are facilitated by online channels, for sure. Others are removed from online impact completely. There’s no doubt that some of the smaller analyst firms, for example, are benefitting from their online presence, in terms of reaching their potential audience through blogging and other social media technologies. But these channels are not creating expertise or authority – simply the means to communicate them.

Can social media create a new kind of influence, by collating the collective wisdom of a connected crowd? After all, there is safety in numbers in doing what the crowd does. We used to have a version of that in the IT industry – no-one ever got fired for buying IBM. Imagine the power of that kind of statement, communicated instantly over the blogosphere. Or would it be immediately challenged and rejected by real users’ experience?

So, are analysts influencing via online channels? How is influence really conveyed by analysts to decision makers? Has it moved mainly to online or is it still by telephone enquiries and face-to-face advice?

*Free registration required, or email me at duncan.brown(at)influencer50.com.

FT’s Decision Dynamics – 3

This is the third and final part of a look at the FT’s Decision Dynamics survey (see here for part 1 and part 2. This part looks at the role of social media in decision making.

Frankly, I think there’s a lot of tosh and assertion on the importance of social media in influence. In short, while there’s been a rapid rise in the use of social media there’s little evidence that it’s impact decisions, particularly in the B2B world. We’re all blogging and Twittering and ‘LinkingIn’ to each other but not in the context of decision making. Anyway, back to the FT survey.

This first chart (below) illustrates that rapid rise in use of social media by enterprise decision makers. As the FT notes, “Work-related use of social bookmarking and community sites has exploded in the past year.” It has, but only to the levels seen to date for reading blogs and using professional networking sites.

(Source: Financial Times Annual Decision Dynamics Survey 2009)

The most telling chart, though, is this next one, which shows the usage of various types of social media in a work or leisure context. It’s interesting to see the use of social bookmarking (Dig, Delicious, etc) as a decision making aid. But webcasts and streaming video are also important and more established in the B2B marketing mix.

Most surprising, given the volume of noise on the subject, is the low penetration of Twitter.

(Source: Financial Times Annual Decision Dynamics Survey 2009)

My own take on this is that Twitter:

(a) Is useful if you know who you want to follow, but can’t help identify important people

(b) Is a great echo chamber of followers and followees, but has limited reach outside the Twitterverse, and

(c) Is really only used by people that don’t have a “traditional” job (autonomous, mobile, networked), which may include most of the readership of this blog and most analysts, but not your average CIO.

Influencer Marketing – it’s all about engagement

So you use one of the 70 online influence measurement firms to compose a list of your top influencers (= those people that influence decision makers). How do you make this information executable?

The whole point of understanding who your influencers are is that you engage with those influencers in order to influence them, making them more favourable (or less unfavourable) towards you.

The RoI is all in the engagement. You can use influencers to attract prospects to seminars. You can use influencers to create marketing content. You can embed pre-existing influencer-led content in your sales messages. You can tap influencers’ networks for lead generation. You can position influencers to overcome sales objections and close deals. And so on.

That’s what Influencer Marketing is all about.

Seth, clout and online influence

Seth summarises the primary lure of online measures of influence (which I’m equating with his term ‘clout’). If someone was able to parse the trillions of data points online… If someone could identify which people had influence… If we knew where the big idea would come from…

Aye, there’s the rub. The strength of online sources, the sheer volume of data, is its greatest weakness. The difficulty is searching the data for the needles in a haystack the size of a planet.

There’s no doubt that the promise of online measures of influence is hugely tempting. But my sense is that it is also unobtainable, at least in the short term.

This over-stated and under-delivered promise is worrying me increasingly, as it’s being pitched as a shortcut to understanding influence. At bst it shos the tip of the influence iceberg. The real danger is that it creates a skewed view of the totality of influential people, ignoring offline influencers.

Influencer50 will be issuing a white paper shortly discussing these issues in more detail – watch this space.

Twitter in “Pointless Babble” shock

The BBC reports that 40%  of Twitter content is “pointless babble.” Citing research by Pear Analytics, “40.5% (of Tweets) could be classified as pointless babble, 37.5% as conversational and 8.7% as having pass-along value. Self promotion and spam stood at 5.85% and 3.75% respectively.”

The shock is that only 40% is pointless babble. I’m surprised it’s that low. There’s no mention of advertising, though I assume that that’s the spam figure. I’m also surprised that spam accounts for only 3.75%, given the clamour that agencies have reported in using Twitter to spam target audiences. Still, early days yet…

We’ve noticed a rise in the use of blogging by influencers over the past four years, though it’s not nearly as high as you might think. Some markets (e.g. web development) feature blogging influencers considerably more than others (e.g. accounting software). But there is a general rise, to be sure.

We’ve recently added Twitter activity to our analysis – what’s immediately obvious is (a) there are still very few influencers using Twitter, and (b) those that do Twitter do so at the expenses of their blogging activity. So Twitter displaces blogging as the preferred mechanism of publishing content.

We’ll continue to watch the uptake of Twitter amongst influencers, though personally I’m skeptical. Seth fears losing six hours a day if he Twittered – enough said.

(Hat tip to Richard Holway for the heads-up on the BBC report)

New research into online influence

On the LinkedIn group for Influencer Marketing I recently posed a question on the relation (if any) between online and offline influence. I also asked if online indicators of influence could act as a proxy for offline influence.

One of the responses pointed me to a new survey being conducted by the Society of New Communication Research (SNCR) called ‘The New Symbiosis of Professional Networks.’ It’s being run in part by Don Bulmer, VP of Influencer Relations at SAP, and it aims to understand how decision makers are using social media and other (offline) networks to inform decisions.

The link to the survey is here. Please take 5 minutes to complete it – all participants will receive a copy of the report (due in the Autumn). The results should be interesting and relevant to everyone involved in influence.

Gladwell on online influence

Ian McKee at Vocanic gives a brief summary of a talk given by Malcolm Gladwell, in which the Tipping Point author discusses the relationship between connectors as influencers and social media users. Short answer: there isn’t one. Gladwell understands the difference between counting connections on Facebook or LinkedIn and the intimacy required to exert real influence.

We talked about this in the book. The majority of interactions between people remains based firmly in the real world, according to research by Walter Carl. There’s also no evidence that online activity is a reliable proxy for offline influence – there are plenty of people that exert influence and yet have very low online profiles (consultants, regulators, academics, etc).

So beware of quick and cheap analyses of influence based purely on online activity. There’s a danger in measuring what can easily be measured, rather than determining what needs to be measured and finding a way to measure it.

Why I don’t like the SIM Score

I’m prompted by Emily Riley at Forrester to discuss Razorfish’s Social Influence Marketing (SIM) Score, as published in the Fluent document.

The idea behind SIM, according to the Fluent report, is that firms employ “social media and social influencers to achieve the marketing and business needs of an organization.” No problem with that, though the focus of the report is anchored in consumerland.

A firm’s SIM score purports to “measure favorable impact of your brand.” Emily paraphrases the SIM methodology thus:

1. Calculate the number of positive, negative, and neutral conversations happening for your brand
2. Use the formula (Positive + Neutral – Negative)/ Total brand conversations
3. Calculate the number of positive, negative, and neutral conversations happening for your industry
4. Use the formula (Positive + Neutral – Negative)/ Total industry conversations
5. Divide lines 2/4 and you have your SIM score

Razorfish hopes that SIM will become as valuable as Net Promoter Score. I doubt it.

Firstly, NPS discounts passive (neutral) opinions: it only looks at positives and negative (subtract one from the other to get your NPS). SIM includes neutral scores and counts them, effectively, as positives. Thus it inflates scores. Intuition tells me that it would be better to exclude neutral opinions or, even better, count them as negative.

Why? Neutral opinions are as bad (from a marketing viewpoint) as negative ones. Lukewarm, ambivalent, disinterested – who’d want these sentiments recorded as positive? In fact it’s harder, arguably, to detect neutral opinions because they’re rarely declared. Negative opinions are often expressed (vociferously!) but few people make the effort to declare their indifference.

The second problem with SIM is its quantitative nature. It’s fine using quantitative measures for NPS because it surveys customer experience. SIM scans the web and notes all instances of chatter on your brand, whether the chatterers are customers or not, influencers or not. All of my experience in researching influence points to the use of numerical indicators as extremely poor in determining influence:

• Quantitative measures treat all comments equally, but influence is not spread evenly and equally. Some people’s opinions count for more than others. SIM doesn’t account for uneven distribution of influence.
• The number of times an individual declares an opinion is not proportional to their influence. Counting instances of sentiment inflates the importance of frequent contributors, rewarding ‘noisy’ people.
• Quantitative measures are too easy to fake. Social media users do this all the time, ‘sharing’ their list of friends with strangers to inflate their apparent popularity. It’s the same as exchanging links on web sites for ‘Google juice.’ It would be relatively straightforward to inflate positive or negative opinion of a brand in this manner.
• Numerical measures are prone to spikes of activity, and are therefore time-sensitive. Positive sentiment may rise after a public launch of a product. Negative opinions may peak after a lousy financial result. So the timing of SIM is critical, if it’s to be used as a benchmark for social media marketing. It also means that SIM can’t be used fairly as a comparative measure against other brands, whose cycle of product launches and financial results (and plenty of other factors) are out of step.

The third issue with SIM is that it doesn’t measure real experience. NPS surveys real customers and asks a simple question: “Would you refer this product/service to a friend or colleague?” It’s a qualified opinion, based on real experience. SIM measures all instances where a brand is mentioned, irrespective of whether the contributors are customers or not. Contributors may have no direct experience of your product, and they may be repeating hearsay or perceived wisdom. They may have an agenda to promote – they may, for example, be customers of an alternative brand or even employees of that competing brand. How can you be sure that these online conversations reflect a fair and true picture of attitudes?

Why has Razorfish implemented SIM in this way? My guess is that it measures what it can measure. It’s easy to measure social media because, by its nature, it’s public and accessible. It’s also online, which makes distance and physical proximity irrelevant.

It’s an order of magnitude harder to measure other aspects of influence, such as expertise and persuasiveness, or influence exchanges that occur face-to-face. But that doesn’t mean we shouldn’t try.

What’s the role of professional networking organisations?

I recently met up with CIO Connect, a networking organisation for (surprise) CIOs in the UK. Its value proposition is that it connects the most senior IT people in the land. It enables information sharing, the raising of individual profiles, professional development opportunities and other benefits.

I also attended a session run by the BCS on enhancing professionalism in the IT industry. The BCS is attempting to (and I’m paraphrasing liberally here) make IT practitioners as highly regarded as lawyers, accountants and other professionals (but not, one hopes, bankers…).

Both meetings left me wondering about the longevity of such professional networking organisations. Whether they are commercial businesses like CIO Connect or trade bodies like the BCS (a registered charity), I wonder at the degree of influence such organisations have, in the era of ad hoc social networking. If I can connect with a CIO directly, via LinkedIn for example, do I need CIO Connect? If I can prove my credentials because of the recommendations I have on my LinkedIn profile do I need CITP* accreditation from the BCS?

I think CIO Connect, TIF and such organisations can still add value, but increasingly this will be value to the vendor community, in brokering relationships with potential CIO-level customers. For example, CIO Connect now offers a vendor-sponsored CIOnet community in the UK, in addition to the core CIO-only service.

The BCS is struggling to introduce professionalism into businesses, yet other trades have managed this successfully. Trivial example: if you want a wood-burning stove installed in your house in the UK you should get a HETAS-accredited installer, otherwise you’ll fail building regulations (you read it here first). Yet getting my PC fixed is a lottery, with no accreditation scheme from the BCS or anyone else**. Are chimney sweeps really more of a profession than IT?

Some degree of regulatory compulsion is usually required to underwrite professionalism, but this buys trust from the public – essential in any professional body. The BCS is reluctant to go down the statutary route, and (I think) will suffer for it. At the moment it’s too easy to ignore the BCS.

How can networking organisations increase their influence and resist the diminishment of their proposition? Firstly they need to embrace the opposition – social networking. The BCS has various groups within LinkedIn that are to some degree active communities of networkers. Similarly, CIO Connect and other similar organisations have online capabilities to enable networking.

More importantly, they need to influence at a professional level. When was the last time a BCS representative appeared on the Today programme? Tony Collins, the editor at Computer Weekly, is a more regular guest commentator.

Influence often starts at home, so they need to get out more and start influencing their agenda. The secret to influencing is to deploy individuals capable of influencing. Organisations do not influence – it happens on a person-to-person basis.

This individual influencing approach is something that social media networking approaches like LinkedIn will find hard to deliver, because it’s one dimensional (ie online only). Individuals can influence online, but the most influential people influence in both online and real world situations.

And that’s the key to success, and longevity, for networking organisations.