As you request and receive additional data and analyse this data within a framework, it is important to continually reconsider your hypothesis as it is likely to evolve over time as you address particular issues in more detail.
One of the reasons for segmenting your data is that statistical data relating to broad groups/issues may lead to wrong conclusions being drawn in relation to more specific groups/issues. To illustrate this point, imagine that the average salary in country ABC increased by 10% in a given year. Whilst this provides a good indicator of the state of the national economy, it is not necessarily representative of the whole population. It may be the case that the 10% growth was predominantly driven by an increase in the salaries of the wealthiest individuals; meanwhile the growth of the salaries of low and middle class representatives may have been marginal. If this were the case, the 10% average income growth would no longer indicate increasing prosperity and equality across the country. Segmenting your data to take into account different income brackets could help you to avoid drawing incorrect conclusions based on the broad “10% growth” statistic.
As there are various ways of segmenting particular data, you could simply tell the interviewer that you want to gain a better understanding of the main drivers of change relating to that data. The interviewer may then segment the data for you based on the criterion best applicable to the particular case. However, following your request, your interviewer may ask you to specify how you would like the data to be segmented, so consider the potential drivers of change before asking the question.
You should segment data while drilling down the branches of frameworks. The key thing to keep in mind while analysing a case is to be mutually exclusive and collectively exhaustive (MECE). Put simply, all the elements that could potentially cause/contribute to the problem you are faced with should be segmented into categories between which no overlap exists. This will ensure all the categories are “mutually exclusive”. You should subsequently extract as much information as possible from the data you have segmented. This will ensure that your analysis is “collectively exhaustive”, as all variants will have been considered.
For instance, if you were asked to brainstorm the factors contributing to the fall in the number of road accidents, you may first want to segment the different types of factors in the following manner: (1) Human factors (2) Technical factors.
As the two segments are unrelated (i.e. a particular factor will relate to either a technical factor or to a human factor, but not both), the analysis thus becomes mutually exclusive. You can then brainstorm factors relating to each of these mutually exclusive segments separately, ensuring that you exhaust all the possible factors relating to each (thus ensuring your answer is “collectively exhaustive”). For instance:
- Human factors: a driver’s age, sex, mental state, sobriety, quality of eyesight etc.
- Technical factors: the vehicle’s condition, its year of production, the condition of the roads etc.
Similarly, a firm might consider introducing new incentives aimed at motivating their employees to deliver better results. If you were asked to suggest a list of such strategies, you would start off by segmenting different types of potential incentives, for instance: (1) Financial incentives (2) Non-financial incentives.
Again, both the categories are mutually exclusive, as a particular strategy will not generally straddle both categories. We should next assess each of the two segments (financial/non-financial) individually, considering all the possible incentives that relate to each, thus ensuring the analysis is “collectively exhaustive”:
- Financial incentives: year-end bonuses, performance-related pay (PRP)/commission etc.
- Non-financial incentives: additional days off, promotion into new roles, the opportunity to work abroad, employee awards etc.
If you face a question relating to a decline in profitability and you discover a notable increase in fixed costs, you should ask the interviewer for more information on different components of the fixed costs, as it is highly unlikely that the increase can be equally attributed to all the different fixed costs. Fixed costs could be segmented in multiple ways, for instance based on geographical areas (e.g. fixed costs in Europe vs. Asia vs. Americas etc.). Similarly, if you discover that a growth in revenue was driven by an increase in the number of units sold, you may want to investigate further and ascertain whether this increase in sales is linked to geographical location, product type or customer bases etc.