I'm a special advisor to AMEC (the Association for the Measurement and Evaluation of Communication), wearing a CIPR hat as and when. I'm part of a working group assembling recommendations on the topic of influence for deliberation at the AMEC European Summit in Madrid this June. We have input from IPR, PRSA, Womma, SNCR, IAB and other groups, associations and institutes.
I took an action to create "something to shoot at", and I distributed the following over the weekend. On the basis that we're an open and transparent working group, I thought I'd post it here too. Do get in touch if you'd like to tell me what you think. Now's the time for dialogue – particularly if you can't attend the Madrid summit – ahead of the standards setting.
Date: 10th February 2013
To: The working group on influence
About this document
This document is "something to shoot at” in working towards the objectives of the AMEC / Conclave working group on influence.
We are trying to develop a standards approach to the terminology of and approach to influence flows – how influence goes around comes around – for the useful application by organizations seeking to encourage various stakeholders to think or behave as the organization would like.
[Update – Thanks to David Geddes for pointing out that the last paragraph may be interpretted as being supportive of an asymmetric model of PR ie. getting the message out at the expense of getting it in. It is not intended to. I always think of influence flowing every which way when invoking the word, although the ultimate objective of PR practice is helping the organization set and achieve its objectives.]
To put that objective into perspective, I believe the best way to exert useful influence remains to deliver great products and services so that your customers evangelize your brand to friends / family / colleagues / industry peers, and to be a well-run organization so that your employees and partners evangelize working with you. As social media facilitates what some call radical transparency, the 20th Century axiom ‘perception is reality’ is transformed in the 21st to ‘reality is perception’.
You have been influenced when you think in a way you wouldn’t otherwise have thought or do something you wouldn’t otherwise have done.
Influence is complex. In other words, changing your mind or actions is the result of many stimuli over time and entails conscious and subconscious processes. There is currently no scalable facility to ascertain or infer who or what caused someone to change their mind or behaviour.
Dictum 1 – Influence is a change in opinion or behaviour.
Complex systems aren’t easy to grasp. Many people appreciate that the weather is complex, that stock markets are complex, that city traffic is complex, but attributing relatively simple cause and effect in the business of influence appears too tempting for many. Oprah made him buy the book. The ad made her buy the sneakers. Her sister’s recommendation made her vacation in Italy.
While complexity science doesn’t rule out the instances in which a single stimulus suffices, it also recognises that this is the exception rather than the norm. In fact, she’s romanticised an Italian vacation for years, and for many reasons she herself can’t tease apart.
Reductionist approaches do not work here. Reductionism assumes that the underlying mechanisms of a system's behaviour can be understood by studying its parts independently. A complex system, however, is by definition more than the sum of its parts and its overall behaviour emerges through the interactions of constituent parts.
A complex system is one that by design or function or both is difficult to understand and verify. It’s a system in which there are multiple interactions between many different components. Complex systems constantly evolve and unfold over time. Complexity bridges the gap between the individual and the collective: from psychology to sociology, from organism to ecosystem, from genes to protein networks, from atoms to materials, from the PC to the World Wide Web, from individuals to society.
Dictum 2 – Influence is both input to and output of a complex system.
The word “influence” is used commonly in attributing an entity some contextual facility to get others to change their minds and actions. For example: The Beatles were an influential band; Chomsky is an influential intellectual; Alexander McQueen is/was influential in fashion.
We may listen out for Beatle-esque riffs, and look out for Chomsky citations and McQueen fashion memes. Perhaps it’s safe to say that the accolade has most often been awarded qualitatively rather than quantitatively. People do not generally count up the number of tracks that seem to have a Beatles-like sound for example. However, the Arts and Humanities Citation Index makes plain the incredible frequency with which Chomsky is cited.
Social media appears to leave traces similar to Chomsky’s citation count. The sentiment of both citations and social media references is usually taken to be positive, as supportive, and while the definition of influence above does not infer congruity, the organizational application requires it. This should therefore be made explicit in related standards.
Typically, both citations and social media references are interpreted as influence having happened and therefore the individual having had more influence than otherwise, and therefore having more influence. It is unclear under which circumstances this assumption may be valid, or indeed on what basis we might assume influence decays or grows with the passing of time. The zenith of Vanilla Ice’s influence on the music scene is past, but the full impact of the Reverend Thomas Bayes’ mathematics (in machine learning) has only played out more than two centuries after his death.
Common English language is therefore ambiguous. Influence is apparently both the ability one is attributed to change another’s opinion or behaviour, and the very changing of that opinion or behaviour. The first describes the source of or contributor to a change in the system, the latter describes the result. This ambiguity causes confusion.
I recommend that our standards define influence as a change in opinion or behaviour (per dictum 1 above), and that we substitute “potential influence” for the other use of the word. It is, after all, nothing more than a potential. The fluttering of the butterfly’s wings may have contributed to a storm this time last month, but that does not mean it will again. The driver’s driving style may have contributed to the traffic jam this time last week, but that does not mean it will again.
Our standards should be accompanied by a brief explanation of complexity. It is central to explaining our work and therefore to winning support for the standards. The simple insertion of the word ‘potential’ helps make this point.
We must scope what quantities, analyses and algorithms might help determine a proxy measure of potential influence, and how the corresponding context and conditions should be described.
Some critics of those services claiming to divine “influence scores” prefer to say that such services quantify little more than the propensity for an individual’s social media contributions to be shared. Some claim such services are confusing popularity for influence. An alternative label – social capital – has been suggested, but this phrase has been used for more than a century to describe the value of the network rather than that of an individual participant in a network, and we don’t want to introduce new ambiguities.
I think we must describe the characteristics and capabilities we consider elemental to any service that aspires to ‘score’ potential influence, a deficit in which forfeits the service from claiming the descriptor.
In compliance with the 7th Barcelona Principle (that transparency and replicability are paramount to sound measurement), I recommend that we insist upon methodological transparency. We cannot endorse black boxes. In fact I believe we have to condemn such opacity in this context explicitly.
Dictum 3 – Professionals don’t rely on opaque analyses.
Leonard Euler kicked off the field of network science in 1735 with the Seven Bridges of Königsberg problem featuring four nodes and seven edges. Today’s academics can simulate problems with many millions of nodes (typically individuals) and edges (relationships between the nodes).
I’m not going to undertake an extensive review of the state of network science here, but a few notable papers help us in our endeavours.
Influence maximization is the problem of finding a small subset of nodes (seed nodes) in a social network that could maximize the spread of influence. It is a discrete optimization problem in a social network that chooses an optimal initial seed set of given size to maximize influence under a certain information diffusion model.
Computer simulations using real life social network data and various simplifying assumptions show that selecting vertices (nodes) with maximum degrees (connections) as seeds results in larger influence spread than other heuristics, but is still not as large as the influence spread produced by other algorithms.
Importantly, it appears that influence ranking – the process of trying to score an individual’s network connectivity – is only good for selecting one seed.
Seed selection isn’t as easy as picking the most connected nodes. Not all connections are equal, and relatively few so-called friends are actually significant influencers of a given individual’s behaviour, while substantial heterogeneity across all community members exists. Descriptors from user profiles lack the power to determine who, per se, is influential, and friend counts and profile views also fall short of being able to identify influential site members.
From general observation, we can say that an organization might heighten its facilities to influence others if it methodically attunes itself to being influenced appropriately – after all the word is derived from the medieval Latin word influentia meaning ‘inflow’ (influere, from in- ‘into’ + fluere ‘to flow’). This is precisely why organizations conduct market research for example, compile daily news insights and business intelligence, and undertake social media monitoring.
Dictum 4 – To influence better, be influenced better.
To my knowledge, such reciprocity has not been modelled by academics. I believe this is because it is difficult to model the qualitative characteristics of the idea or message we’re seeking to spread. Yet that might in part be a function of what some consider a critical determinant of influence spread, namely the readiness of the community in question to be influenced. In extremis: “If society is ready to embrace a trend, almost anyone can start one – and it isn’t, then almost no one can.”
More on terminology
Every fast changing industry has its jargon. The lexicon emerges to aid efficient communication, but that efficiency is only achieved when everyone knows what the words and phrases really mean, and uses them consistently. Based on all our discussions to date, here are my recommendations.
Influencer – Anyone who contributes to someone else changing their opinion or behaviour. (I think this invalidates the draft “citizen influencer” definition. In recognition of complexity, I defer to a definition entailing a contribution rather than being the cause.)
Key influencer – Someone who, following statistical modelling and analysis, is considered key with some degree of confidence (alpha) to the efficacy of a program of influence.
Influential – A descriptor applied to an individual deemed to have been a key influencer and who might (but might not) remain one.
Types of influencer, key or otherwise:
Advocate – An individual who shows support for, pleads the case of or defends a brand, cause, product or service while remaining formally unaffiliated with it and unremunerated.
Ambassador – An individual remunerated by or otherwise allied with a brand; their actions are, in some manner, endorsed by the brand with an acknowledged and transparent affiliation that is mutually beneficial.
Professional / occupational – Individuals who by definition of their job function are in the position to influence others directly through authoritative or instructive statements.
Celebrity – An individual whose name recognition commands a great deal of public fascination (“celebrity status”) and has the ability to use their status to communicate with broad effect, either as advocate or ambassador.
Influencee – a person who changes their opinion or behaviour as the result of exposure to new information.
Note that this definition doesn’t convey the congruity of the influence effected. Every individual can be categorised as being an influencee (types 1-3) or not (type 0):
- Type 0 – no exposure to the message, no influence
- Type 1 – exposure to the message yet no influence
- Type 2 – exposure to the message and influenced as the originator intended
- Type 3 – exposure to the message and influenced contrary to the originator’s intention.
Influence modelling – statistical modelling and analysis to inform the design of a program of influence. (We should list some of the properties, inputs and constraints that might constitute an influence model, without implying assumed association or correlation of course.)
The terms "influencer marketing" and "influencer PR" are inappropriate in my opinion as they emphasise “influencer” over “influence”. If we wish to distinguish marketing and PR practice that utilises network analysis from practice that does not, perhaps we should say “network analysis marketing / PR” or “influence analysis marketing / PR”.
Of course this horse has already bolted – the term “influencer marketing” is already out there. But perhaps that’s appropriate as it’s typically invoked to endorse a methodology this document cautions against.
 P. Sheldrake. A Measure Of Influence. Communication World magazine, Jan/Feb 2013, IABC (See https://www.philipsheldrake.com/2013/01/a-measure-of-influence-iabc-communication-world-magazine/.)
 P. Sheldrake. The Business of Influence – Reframing Marketing and PR for the Digital Age. Wiley, 2011, ISBN 978-0-470-97862-7
 P. Sheldrake. The complexity of influence is a challenge – and an opportunity. The Guardian Media Network. 15th February 2012.
 G. Weng, U.S. Bhalla and R. Iyengar Complexity in Biological Signaling Systems Science 284:5411 (2/4/1999) 92-6. DOI: 10.1126/science.284.5411.92
 D. Rind. Complexity and Climate Science 284:5411 (2/4/1999) 105-7. DOI: 10.1126/science.284.5411.105
 W.B. Arthur. Compexity and the Economy Science 284:5411 (2/4/1999) 107-9. DOI: 10.1126/science.284.5411.107
 Professor Henrik Jeldtoft Jensen, Department of Mathematics, Imperial College. https://www2.imperial.ac.uk/~hjjens/
 Solis, B. Please Repeat: Influence is not popularity. A blog post. 11th August 2010. http://www.briansolis.com/2010/08/please-repeat-influence-is-not-popularity/
 Putnam, Robert D. Bowling Alone: The Collapse and Revival of American Community. New York: Simon & Schuster. 2000. (See http://bowlingalone.com/?page_id=13.)
 W. Chen, Y. Wang, and S. Yang. Efficient Influence Maximization in Social Networks. Proceedings of the 15th ACM SIGKDD Conference on Knowledge Discovery and Data Mining, pages 199-208, 2009.
 Ibid 14.
 D. Kempe, J. M. Kleinberg, and É. Tardos. Maximizing the spread of influence through a social network. In Proceedings of the 9th ACM SIGKDD Conference on Knowledge Discovery and Data Mining, pp 137–146, 2003.
 K. Jung, W. Heo, W. Chen. IRIE: Scalable and Robust Influence Maximization in Social Networks. Proceedings of the 12th IEEE International Conference on Data Mining (ICDM), pages 918-923, 2012.
 M. Trusov, A. Bodapati, R.E. Bucklin, Determining Influential Users in Internet Social Networks, Journal of Marketing Research, August 2010.
 C. Thompson, Is the Tipping Point Toast?, Fast Company, 1st February 2008. http://www.fastcompany.com/641124/tipping-point-toast