Blog

November 29, 2019

Written by

The 4 Principles of a Data Driven Organisation

Introduction

The world of commerce has evolved beyond recognition over the past 20 years; paradigm shifts that once occurred in multiples of years can now occur in months or weeks, the way customers interact with organisations has shifted the balance of power away from organisations leaving management teams baffled, and the price of technology has collapsed to a level where nano-commerce start-ups can successfully challenge larger capital rich blue-chips.

At Alliants we believe that many organisations have allowed themselves to become detached from their very reason for existing – you exist to generate a return for your shareholders and you are only able to do this if the value you extract from customers exceeds the money you invest to acquire and retain them. Once this approach is embraced we believe the winners will be those organisations that have laser guided analysis to optimise the net value created and ultimately deploy technology to automate the almost infinite operational adjustments to achieve ultimate optimisation.

To guide organisations on their journey we follow four principles to provide analysis for them to survive and thrive in this new era:

  1. Customer Accounting
  2. Cognitive Analytics
  3. Cohort Clustering
  4. Automated Decision Making

1. Customer Accounting

In a world where analysts and private equity backers put such a strong emphasis on performance of the most recent relatively miniscule time periods it is hard to elevate the debate to a more cerebral level. One principle though has remained a consistent thread through the many ages of commerce; long term value is created only if the return from each customer exceeds the investment you make to acquire them and retain them.

Before the internet gate crashed the party this was a relatively simple concept; you would invest money to build a store, you may have needed to tell people the store existed and as long as enough people used the store, and you made margin from each transaction to cover the costs, long-term value would be created. When the internet came along much changed; customers interacted through omni-channels, identification of their behaviour became easier and the opportunity to influence them democratised.

Today represents one of the most challenging commercial periods for generations. We have more data than we know how to use and the speed of change has led to a capability / experience paradox where a relatively small number of people “get it” and only a proportion of them have the experience to do something about it. It is telling therefore that a significant major concept has failed to keep pace with these changes; accounting.

Much effort by various bodies has gone into improving the data that organisations present externally so that this can be relied on for investment decisions consistently and understood. The approach they take has then cascaded into organisations as using this as your foundation is most likely to align your day to day management with external view of how you are performing. There is one major flaw; what has the EBITDA I make today got to do with the EBITDA I make tomorrow?

As we entered the internet age there was one group of businesses who led the field in understanding the behaviour of customers and it may come as some surprise – the mail order companies. For years they had obsessed with using data to optimise the space usage in their expensive catalogues and were fortunate enough to be able to identify all of their customers and their behaviour. This rapidly evolved into an early thinking in what we call customer accounting.

Even in an age where customers are more transient the majority of organisations follow a similar pattern; you pay to recruit a cohort of customers and it's only how this cohort behaves over time that will ultimately decide how profitable they have been. As a result, if you are concerned with the performance of any one year you need to be obsessed with the benefit in that year of customers acquired previously and the hopefully small cost to retain them, as well how much money you deploy to bring in new customers and of what quality.

Once an organisation makes the shift to customer accounting it changes the dialogue to one that is much more aligned to generating long-term shareholder value:

  • An obsession with the efficiency of marketing spend and the quality of customers acquired
  • A focus on maximising retention and frequency of customers
  • A culture that recognises what you do today can have a significant impact on what happens tomorrow
  • The framework that better aligns the long-term plan with very short-term targets

There is one obstacle to customer accounting and that is customer identification. Even in a world governed by GDPR the drive to identify customers will be hard to resist and as such solutions will continue to be developed. Most organisations can identify at least some of their customer and make some assumptions as this can often be enough to make better decisions; you often only need a few clues to solve the crime.

2. Cognitive Analytics

The world has moved on dramatically from an age where the CEO of a business would have to phone each of their stores at the end of the day to find out how they have done, and the stores themselves would find out by adding up all of the Z reads from the till. The explosion of data points in the past decade has analytical tools struggling to keep pace and, even when they can, organisations overwhelmed by the volume of data and potential insight.

We have a very simple definition of analysis at Alliants:

The organisation and examination of data that leads to optimal decision making

Underpinning this definition, we believe the objectives of analysis are to find out:

  • What has happened?
  • What has caused it to happen?
  • How long has it been happening for?
  • Is anything happening at a granular level that is being masked by aggregation?

We also believe that to achieve optimal decision making there are some principles that underpin any piece of analysis:

  • Is the analysis relevant?
  • Is the analysis accurate?
  • Is the analysis simple?
  • Is the analysis timely?

The above can’t be achieved without an appropriate data strategy and architecture but once these have been established the focus needs to turn to the analysis itself.

We are fortunate to be in an age where access to data has become much improved. facilitating its flow throughout an organisation. There are obvious downsides to this ranging from not being able to see the wood for the trees through to the other extreme of the often-talked-of analysis paralysis.

With 70% of the brain’s sensors being occupied by the visual memory system we believe the key to deliver our objectives and principles predominately lies in optimising analytics visually. There are many facets to this, but some include; minimising distraction and optimising focus, using cognitive principles to best demonstrate an observation, as well as using visual keys to link disparate themes more easily.

In addition to this we have also tried to encapsulate years of practical experience into building techniques into data analysis that better teases out relevant observations. This include:

  • Banding values together to create new attributes
  • Breaking values out into their constituent parts
  • Calculating the £ impact of all KPI variances
  • Using mix analysis where attributes have disparate values and where you can control the mix of those attributes
  • Use combinations of values to create new values as a shorthand
  • Extrapolating forward using scenarios based on past performance

The result is a suite of reports delivering easy to use, timely and relevant insight with the singular aim of optional decision making.

3. Cohort Clustering

The Oxford definition of analysis is:

[mass noun] detailed examination of the elements or structure of something: statistical analysis | [count noun] an analysis of popular culture.

The irony of this definition is that we believe it is closer to the truth than you may have initially thought. It is only by breaking larger data sets into smaller data sets, or cohorts, that you will be able to generate actionable insight.

These cohorts come in two forms; user generated, and machine generated.

User generated cohorts have always existed in organisations but were often limited by the data being collected; region, product group, time period etc. The increases in the level of data attributable to a transaction opens up a world of additional more complex cohorts that organisations are able define, especially in combination with customer accounting. Examples would be; grouping by acquisition source, by month of acquisition, by category of first purchase etc. A great start point for organisations is to take the attributes they posses and examine any deviation in KPI’s based upon these.

Machine generated cohorts take this principle further by devising clusters of similar cohorts but in doing so are likely to find combinations of attributes that a hundred analysts in a room may never find. In doing so they will enable the organisation to measure progress over time against a cohort that perhaps would never have been identified as being important before, but that contains the key to future success.

4. Automated Decision Making

The quantum of combinations of different attributes, situations and kpi’s mean that it is impossible for human resource to act upon this in a timely manner to optimise performance. Of course if you pursue the ultimate combination of attributes you end up at an individual.

There are therefore certain decisions that it would be better to have a machine make for you to achieve optimal results and these include; what price to sell out, what promotion to run, what content to display, and even what communication to make.

We believe that the handover of responsibility of the decision making above has to be accompanied by appropriate analytics so that you get the best of both worlds; decision making at a micro level by machine, control at a macro level by a human.

Conclusion

The world of commerce has undergone a generational change and, whilst there can often be sources of capital that sometimes are at odds with sound logic, at Alliants we believe that an organisation will ultimately be successful if it is in control of the balance between customer lifetime value and cost to acquire these customers.

To be in control an organisation needs to be able to make decision based on accurate, timely and insightful data and this needs to be presented when and where it is needed in a convenient form for optimal execution. The higher the level of inputs into decision making the more likely that these decisions should be made without human intervention, other than high quality monitoring of outcomes.

Ultimately every organisation should be aiming for a singular purpose; through all the decisions I could make how near can I get to choosing the ones that create optimal shareholder value. At Alliants we believe we are best placed to deliver this.