The position of analytics in trade-off decision-making

“We may regard the present state of the universe as the effect of its past and the cause of its future. An intellect which at a certain moment would know all forces that set nature in motion, and all positions of all items of which nature is composed, if this intellect were also vast enough to submit these data to analysis, it would embrace in a single formula the movements of the greatest bodies of the universe and those of the tiniest atom; for such an intellect nothing would be uncertain and the future just like the past would be present before its eyes.”

– Pierre Simon Laplace, A Philosophical Essay on Probabilities. Whether the Daemon is disproved, or supported by, by Chaos theory, Quantum mechanical irreversibility or ‘Free Will’ in some guise (which itself may be theorised to be an emergent property of the two listed theories) is a fascinating discussion though the answer may not address why we find the Daemon so appealing. Personally I recall having Laplacian thoughts as a ten-year-old merrily considering a grand equation. Perhaps I was influenced by the following exchange in the 1999 Sitcom Spaced: Brian : Chaos Theory! Tim : Eh? Brian : The predictability of random events. The notion that reality as we know it, past, present and future is actually a mathematically predictable preordained system. Daisy : So somewhere out there in the vastness of the unknown is an equation… for predicting the future! Brian : An equation so complex as to utterly defy any possibility of comprehension by even the most brilliant human mind, but an equation nonetheless. Tim : Oh my God! Brian : What? Tim : I’ve got some fucking Jaffa Cakes in my coat pocket!

“The results show that failure avoidance was negatively correlated with self-efficacy, goal commitment, and task performance. The relationship between failure avoidance and performance was mediated by relationships with self-efficacy and personal goals. Goal commitment moderated the relationship between personal goals and performance. The results of this study are discussed in terms of Locke’s motivational sequence, suggesting that failure avoidance motivation, although overlooked, has important consequences in goal-setting situations.”

Failure Avoidance Motivation in a Goal Setting Situation (Heimerdinger & Hinson’s 2008). This extract is taken from the abstract of the paper. Hearteningly there is a rich literature on the topic of failure as motivation. This essay is aimed at the negation of uncertainty through analytics and the associated dysfunction of organisations that over-index on this heuristic whereas my brief survey all of the psychological studies evidences the presence of the motivation and its influence on capacity rather than the second-order issue of emergent dysfunction within organisations.

“Fear of failure is examined from a need achievement perspective and in the context of research amongst high school and university students. Theory and data suggest that fear of failure can be separated into two camps: overstriving and self-protection. Although each has yields in terms of achievement or in terms of self-protection, they render the academic process an uncertain one for students marked by anxiety, low resilience, and vulnerability to learned helplessness.”

Fear of failure: Friend or foe? (Martin & Marsh 2006). Another excerpt from the abstract of a paper. There is something a little cheeky about this form of referencing as the presentation of academic rigour in combination with a thrifty scan of the first paragraph seems slightly duplicitous… don’t you think?

“This analysis restricts risk seeking in the domain of gains and risk aversion in the domain of losses to small probabilities, where overweighting is expected to hold. Indeed these are the typical conditions under which lottery tickets and insurance policies are sold. In prospect theory, the overweighting of small probabilities favors both gambling and insurance, while the S-shaped value function tends to inhibit both behaviors.”

Prospect Theory: An Anlysis of Decisons Under Risk (Kahneman; Tversky 1979). Though pithier descriptions of Loss aversion are available this references the paper that introduces the theory and is therefore deserving of a mention. Kahneman is of course the Nobel prizewinning Kahneman of Thinking Fast and Slow fame amongst much else. The Nobel prize is not awarded posthumously and Amos Tversky was excluded from the 2002 cash windfall by six years. Kahneman said of the situation “I feel it is a joint prize. We were twinned for more than a decade.”

Organisations are comprised of functional components with specific inputs, transformations, outputs and networked influences. The transactional and functional information comprises the operational model and could therefore be optimised or changed through rational, objective, study.

Pure rationality is, regrettably, fleetingly rare given the necessary human actors and perspectives. Negation of our natural fallibility is achieved through the deployment of thinking models, personal tendency, governance and so on. These controlling elements serve more masters than singular address of this issue but do play a central role.

The transformative development of organisations requires evaluation and attention. This activity is polluted by multiple forms of biases that ultimately conspire to create a distance between that which should be worked on and that which is. This is a form of suboptimal change and different elements of organisations are subjected to this distance more than others to the extent to which organisational components are variously sensitive and susceptible to bias.

This bias is fairly predictable given a decent understanding of our nature. Activities that are perceived as artistic and creative are highly prized. Marketing, product and graphic design attract the attention of those in search of self expression and joy. Sales and other means of direct revenue generation pull the spotlight towards themselves due to there obvious capacity to create personal wealth, mastery of conversation and opportunities to travel. Cyber-security is attractive to readers of spy novels and people who are enthused by developments in technology.

The effect of bias on research teams is an interesting and subtle area of dysfunction and it’s explanation is the central discussion of the article. Research teams are those elements of enterprise that supply truth and disambiguation to their audiences. The contribution of value to organisations is realised through two mechanisms. The first is through the advantages offered in decision making (efficiency, scope and results) and the second is through the accumulation of insight-as-an-asset or Intellectual property (IP hereafter). This essay focussed on the role of insights within decision making though IP accumulation adheres to similar principles but is perhaps deserving of its own, private, analysis.

Where and how do our biases manifest around research? The discipline has received a recent boost in its sex-appeal due to the apparent opportunities of artificial intelligence though we, in the field, are hamstrung by the difficulties of data engineering in what seems to be a roughly 3:1 investment of preparation:AI in the analytic space. This limitation is likely to be negated through the increased adoption of blob storage (which is to say centralised unstructured data) as a standard. Still though this slight buff to the glamour of the industry is likely here-to-stay and seems to me to be deserved.

Beyond the zeitgeist the charms of business intelligence surround the topic of uncertainty. Uncertainty is most plentiful and potent in the future. Well developed organisations are in a constant state of scanning their history to develop better models becoming, with enough investment, Laplacian Daemons. Mechanistic understandings of organisational and market behaviours to offer advantages to organisations. Contact centres are able to plan for demand within a cost-effective corridor of predicted volume. Bridgewater Associates offers a consistent return on its portfolio based on constant modelling.

It would be wrong to suggest that there is not advantage available to organisations through the study of the past though it does seem that these models cannot cope with the established and demonstrable noise created through emergent complexity. Somewhere between the mechanistic model and the chaotic reality lies the analytic satisficient (a portmanteau of satisfaction and sufficient coined by Herbert A. Simon in 1956) middle that is both true enough and cheap enough to be useful.

We often cling to the need for research because it alleviates the reputational risks of decision-making. It seems that this disposition, frequently found within the gas-lift-swivel-chairs of England, is a manifestation of irrational loss-aversion. Put more plainly, people (especially those with office jobs) seem to want not to lose more than they want to win and are prepared to emphasise research activities asymmetrically to this end rather than in strict pursuit of advantage.

For some instinctive evidencing of this tendency to value loss more than gain, we might imagine that the same proposition posed in language relating to either loss or gain may have differing uptake even though the game is essentially the same. The literature (most prominently D. Kahneman’s work) on loss aversion seems to lead to the same conclusion. Please consider the following manipulative statements, their positive or negative phrasing and how this polarity influences the relative persuasiveness.

• “Your fine is £100 if paid within the next 10 days it will be reduced to £5o.”
  • The language of loss is invoked as the offer will expire after the initial period. The opportunity not to lose is expiring and the recipient will ‘fail’ to take advantage more than they will exploit the opportunity.
  • A gain based formulation might be to say “Paying this £100 fine within the next ten days will save you £50.” This seems likely to be less compelling to most people even though the proposition is exactly the same.
• “Quitting smoking will allow you to keep playing football.”
  • Here the language is positive and arguably weak. It seems that some of the impotence of these gain-based articulations is related to what we feel that we deserve given our homeostatic nature.
  • Consider the stronger, loss-based, formulation: “If you don’t quit smoking you will be unable to play football”.
• “Stocks are limited, get yours before time runs out.”
  • Once again the language of loss is apparent here in the common marketing copy. The focus here is on the word “yours” which suggests that the target is already in possession of the item and need only perform the mere administrational effort of maintaining ownership.
  • The gain based articulation may be something “Stocks are limited, get one before time runs out.” his phrasing lacks the essential statement of pre-existing ownership and might, therefore, be considered to be a little weaker.

The power of this loss-emphasising-language serves as evidence of the intuitive sense of the potency of this polarity of argument. This tendency for us to avoid loss more than we seek to gain is present in everything we do and is especially visible within organisational decision-making. Though plenty of data exist to evidence this phenomenon it is instructive to acknowledge the presence of it within our intuitive responses to language as it is within these moments of reaction that dysfunctional decisions are commonly made.

The demonstrable loss-aversion bias has a powerful effect on many areas of organisational behaviour but particularly on the investment granted to business research. Optimal resourcing of this operation can be achieved through appropriate contextualisation of the enterprise within its larger aims. Too often though research is commissioned asymmetrically in service of preventing loss rather than to achieve advantage in its purest terms.

Given that people are likely to fear loss more than is strictly rational it follows that organisations may overemphasise loss-mitigation-based operations beyond what is optimal. Analytical functions within businesses are good candidates for this over-attention as the more effort that is poured into analysis the safer the decision that must be made is likely to be. Furthermore, it is hard to find the moment that the answer or analysis provided by analytical functions is accurate enough as the acceptance of the accuracy may well be determined, to some extent, by the organisation’s appetite for risk.

Unfortunately, this hyperextension of analytical functions in pursuit of ‘very-correct’ answers is doubly inefficient as gains in accuracy usually diminish over time and the misappropriation of the analytical resource limits the potential analytical gains that could be generated in other, less explored, areas.

Just as other attentional biases are mitigated through rationality; analytical functions must be controlled. In order to assist in this endeavour the following principles are humbly suggested:

• Data analysis should be proportional to an organisation’s aims.
  • The organisation must consider its positional advantage. Is there more to be gained or lost? Is the organisation ahead of the competition in an exhausted market with everything to lose or is it the entrant into an expanding landscape? Furthermore, what is the organisation’s capacity to lose and desire to win? Given these considerations (and more) an organisation might consider it wise to conduct analysis in accordance with these parameters.
  • In order that analytical activity can be commissioned within the appropriate boundaries, a high-functioning organisation may wish to track the purposes of its analytical undertaking and adjust the strategic resourcing in accordance with some custom-weighted framework.
  • Before moving on from this we should note that as ever, models are wrong. Purely rationing analytical endeavours in accordance with organisational aims can produce sub-optimal results given other variables. We might imagine that business advantage is less attainable through loss-prevention than gain-seeking and therefore an asymmetry in the emphasis placed upon the analytical activity, This, and other, compromises are important considerations within organisational design.
• The sufficiency of the analysis should be determined.
  • The commissioners of the analysis should work with the analysts to create a deliberate end-state so that it will be clear what level of accuracy is needed to be generated. This is not something that can be created without collaboration between these parties as the end-state must be satisfactory to the decision-makers and be analytically robust.
  • The format of the end-state conditions is worth dwelling on briefly. If the organisation is prepared to be wrong in one in ten similar decision-making situations the analysts may be inclined to aim for accuracy in their work that matches this statement. The way that organisations choose to think about what is good enough will shape the criteria itself and should, therefore, be a consideration.
• Decision parameters should be understood.
  • Just as the sufficiency of the analysis should be approached deliberately the exact parameters of the tabled decisions need to be explored and stated. Without a full understanding of the choices that are intended to be made a full analytical operation cannot be designed.
  • There are always limits within which organisational choices can be made. If an organisation is not capable of launching a differentiated campaign across its markets then the analysis should be similarly constrained within this condition.
  • Though it is clearly sensible to invest analytical effort in those areas that are within the scope of the decision-making activity this must be approached deliberately as the pull towards the unalloyed good of having more information is a powerful distortion.
  • This is not to say that there shouldn’t be some additional analytical stretch made beyond the pure parameters of the choice when such additional information may have other beneficiaries; just that this should be deliberate. It is often the case that analytical activity generates knowledge that can be gainfully deployed in other areas. Long term strategy,
• Modes of analysis should be deliberate.
  • When an organisation deploys analysis to increase certainty within a decision-making framework it should be understood that as the analysis progresses and the levels of certainty rise the mode of the analysis may change. It is quite common for a quick analysis and compilation of the known facts and parameters to be completed at an early moment of a large analytical project. From this point, deeper threads of understanding can be pulled at and the levels of certainty will rise. These distinct phases of an investigation are, necessarily, different and should be treated as such as there may be advantage available to organisations that prioritise certain elements of these analytical processes. Please see the diagram below for a worked example that demonstrates how this idea might usefully be deployed.

This essay demonstrates a key weakness in our ability to rationally commission analytical work as well as a few principles that may be gainfully deployed to strength our flawed reasoning. Though analytical endeavours provide powerful instruction in uncertain conditions the temptation to over-invest in a single question is alluring and may lead to inefficiency, or worse perhaps, missed opportunity.

The ultimate cure for the dysfunctions described here must be the construction and parameterisation of decision frameworks that can be optimised openly. The overindulgence of analysis in pursuit of unnecessary uncertainty can only occur in the dark and it is hoped that by allowing these endeavours to be treated like functional parts of a larger machine they can be exposed to beneficially supportive forces.