What is the Analytics Process?

Analytics – the path from data to strategy

When I was thinking about creating this blog I wanted to think of a set of guiding principles that would guide the topics that I explore, and I think the blog strap line “Turning data into knowledge, knowledge into insight, insight into strategy” goes some way to encapsulate the essence of those core principles, but it also implies there is some process for moving from data to strategy. How do we think analytically and what is the process that delivers solutions to our clients – what is the analytics process?

I want to take a detailed look at the Marketing analytics process in future posts, but before we get into the specifics of marketing analytics what is The analytics processmeant by the term Analysis?

‘Analysis’ has been defined in many different ways, here are some of those definitions;

  • “The process of separating something into its constituent elements for individual study”
  • “The detailed examination of a complex process, system, or situation to understand its components and inter-relationships”
  • “Looking for cause and effect”
  • “The examination and evaluation of relevant information to select the best course of action from among various alternatives”

“Analysis” for me is most definitely about breaking a situation or problem into its component parts so we can properly investigate and evaluate their relative significance. It’s also about evaluating the relationship each component of the problem has to the underlying root cause – analysis is the logical, systematic, and unbiased search for the truth, and can act as a valuable and powerful guide for business decisions and strategy.

Analysis follows a logical and systematic process from identifying and articulating the business problem right through to delivering and challenging the solution, listed below are what I believe are the six steps of marketing analysis;

1. Understand the business context

Before we get into the actual analysis it is worth remembering that the problem under consideration is not a purely analytical or theoretical exercise – we are doing this to understand and improve a business situation or problem.

Start by gathering background information about the current business situation and make sure you document specific business objectives that have been agreed with key decision makers. „ Get agreement upon criteria used to determine the success of this project from a business perspective.

  • Determine the real requirements – can they be quantified so you can demonstrate success?
  • Clarify assumptions in terms of the problem and its importance to the business
  • Verify constraints such as time and resources

2. Define the Problem

In order to perform any meaningful analysis which is going to be useful either operationally or strategically you really have to define the actual problem carefully. Get the problem definition wrong and any subsequent analysis is not going to help make the right decisions. Einstein is quoted as having said that if he had one hour to save the world he would spend fifty-five minutes defining the problem and only five minutes finding the solution!  Defining the problem is especially important when clients come to your with a tight brief, have they really understood the actual problem? Are they leading the analyst in the wrong direction?

In my experience it is worth rephrasing the question using different words to check that people still identify with the essence of the problem they are seeking to solve.  You also need to review and challenge each assumption and truism that has been used to frame the problem. Look for exceptions and outliers that don’t meet the common wisdom, do they lead you to see the problem in a different way. Another technique is to ask different people to define the problem and get some perspective. Conduct a longitudinal investigation by asking people who are at different levels of the business to define the problem, then get a latitudinal view by asking people at the same level to get some idea of the degree of consensus around the problem. Learn more on defining the problem.

3. Create a hypothesis

An hypothesis is a logical supposition or an educated guess. It provides a tentative explanation for the situation under investigation. If the problem has been well defined it should be straight forward to create one of more hypothesis to explore in more detail.  Use your experience and subject matter knowledge to guess ways in which you can use your data to solve the identified problem. Often you will have several candidate hypotheses and these provide a framework for you to structure your information collection process.

In order to remain objective don’t set out to prove or disprove your hypothesis. The aim of your investigation and analysis is to see if your data supports or rejects the hypothesis. If the data rejects your hypothesis then how do the findings frame a new hypothesis?

4. Collect your ‘Facts’

After you have framed your problem and created one or many hypotheses you need to pull together all the facts you have available.  You will have collected and reviewed much of the soft information in your problem definition and hypothesis generation stages, but now you need the hard facts – the numerical data!

All data consists of two critical components, the data itself and its associated meta-data – often referred to as ‘data about data’. Unless you understand what the data means its impossible to interpret that data correctly, and don’t forget all data by its very nature is historical – a record of some past event or action.

It is worth understanding that the context of the data can change over time.  It is not impossible, or for that matter uncommon, for the meaning of data to have been changed subtly over time as new meaning is assigned – even if its through historical coding inconsistencies.

Think about the type of data that might help add meaning to your analysis;

  • Existing raw data which is available
  • Data which can be derived from the raw data such as ratios or time series
  • Purchased data like geo-demographic or behavioural segments
  • Enrichment data like dates of birth, clean address information, or intention information

5. Analyse your ‘Facts’

There is a concept called the “knowledge continuum’ that has its roots in the 1800’s.  Raw facts (data) is at the bottom of the continuum and by understanding the meaning and context of our data (using meta data) we allow the analysis and synthesize of the data to provide more and more meaningful information which in turn can be used to gain a deeper insight to our particular business problem – ultimately driving strategy and action.

  • Use your data to test the big things first, look for evidence to support or challenge the perceived drivers of your business problem and your stated hypothesis.
  • Look for the root causes of issues
  • Let the data tell its story, but don’t be too focussed on proving or disproving your hypotheses or you may be in danger of trying to fit the evidence to the desired outcome.  Before you consider a ‘finding’ part of the story worth telling gauge its significance and importance
  • Look extra hard at the exceptions to the norm, do they tell you anything significant or do these exceptions only apply to certain categories of your sample (are they a specific demographic or segment or driven from a particular process?)
  • Experiment with different samples/segments look for differences or similarities how do they compare over time, by income, by profit, by proportion etc.
  • Cross tabulations or data cubes, graphs, distributions, histograms, and scatter plots can all help to visualise hidden meaning

6. Drawing conclusions – what does it all mean?

How exactly do we draw conclusions from all our analysis, my first response from posing that question was “experience and subject matter expertise”, but is that really the total story?

Research suggests that our interpretation of facts and evidence does depend on our own personal belief system. Our beliefs are largely based on the combined lifetime of the meanings we have assigned to events and experiences and how those experiences have been positively reinforced through successful outcomes – so experience and subject mater expertise does play a part in shaping how we interpret the world around us.

On closer examination our belief system plays a part but we extract meaning from analysis by understanding the context and relationships between our findings and some other fact – it’s this comparison and constant questioning that drives our deeper understanding and insight.

If you want to understand your findings or gain deeper insight then talk to people and share what you have learned, use others belief windows and generate lots of comparisons and questions, only then will you truly get the complete picture



April 2013

Shaun Williams

I am Head of Data & Insight for a major UK Charity having worked in the non-profit sector for nearly twenty years. I have a real interest in marketing strategy, innovation, data analysis, customer insight, and behavioural science.

More Posts - Website

Follow Me:

Leave a Reply