Can I have some advice on how to make insights from graphs/tables?

Bain BCG Case Interview McKinsey
Recent activity on Oct 22, 2018
3 Answers
3.0 k Views
Anonymous A asked on Oct 21, 2018

I'm finding it difficult to be given a set of data like a balance sheet and then make recommendations of only that. Interpreting it isn't really a problem - if fixed costs make up a higher proportion of overral costs etc but I struggle with making insights from so few data points.

How can I practice this? I have in the mind the data dump presentation case in Bain's final round.


Overview of answers

  • Upvotes
  • Date ascending
  • Date descending
Best answer
replied on Oct 22, 2018
Collected McKinsey & BCG offers/ Ex-McKinsey consultant/Harvard/WBS/MSU

I use charts in my cases often. What's the biggest mistakes I see?

  1. Candidates forget what the client goal is
  2. Candidates to no connect the dots: do not use the information I gave them before
  3. Candidates try to find patterns in values.

While analysing a histogram of historical sales of car producer from 2010-2018 candidates normally answer:
"Sales were very flat over time, average probably 500,000 units a year, except for the big drop in 2014, maybe because of the financial crisis. Interestingly the share of washing machines is quite constant at 60% of units over time". Why is this not the best way to answer such question?

  • Statement of the facts. No insights - your client already knows that: candidate explains to people who are responsible for these numbers as their main job how these exact numbers look to an outsider.
  • Wild accusation (financial crisis - I am always amused, as I update my case charts by 1 year each 6 months, and candidates try to connect the data to historic events)
  • Lack of context, no idea why the candidate is doing what she/he's doing or where she/he's heading (assuming the chart is part of a case and not the exercise itself)

It's obvious that candidates have no clue what to do, but this is the worst way to buy time-don’t be silent. OK, since you asked, how SHOULD you do it? Key elements of any chart are: Title, axes descriptions (units!!), and legend. The values are far, far less important, in the end they only go into the equations. The approach I suggest is the following:

  1. Say title, axes descriptions with units, and legend out loud. Use medium to slow temp to buy thinking time.
  2. Find connections to all the facts and analysis you have done before. If necessary, state a client goal again.
  3. Choose relevant info according to client goals. There is a chance that a large chunk of data will be useless. Generating insights out of messy amounts of information is a core consulting skill. Thus, it is tested during the interview.
  4. Continue with the case to the next point with mentioning your next steps for this graph.

How would a model answer look like?
"On this chart, I see the sales number of cars from 2010-2018 for the client in thousand units. In order to now estimate the sales potential for the new "model T" as add on, I need to estimate the number of potential car sales to households in 2019. I would thus take the sales numbers from ...". Furthermore, I assume there is a link to a prior analysis I have done...-->

Repetition buys time and brings clarity (= shows your understanding to the interviewer, and thus your listening skills), and is not "wrong". Repeating the "why" shows you're in control and managing the situation, heading to solve the case. This is a trick consultants use all the time: Repeat facts or repeat the overall goal. :) And you still sound smart and have a bit more time to figure things out.

Hope this helps!

Kind regards,

Was this answer helpful?
replied on Oct 22, 2018
Ex-MBB, Experienced Hire; I will teach you not only the how, but also the why of case interviews

3 steps to reading a data table or chart:

(1) give me a high level understanding of what this table talks about (e.g., read the title, or the columns/rows);

(2) read the data, to understand what it says;

(3) draw an insight, and link it to your working hypothesis; this might be 5% of your work as a consultant, but 80% of the value you will bring

How to get better at developing insights? Well, practice helps obviously. Beyond this, don't forget what the question and what your current hypothesis are. An insight that has absolutely nothing to do with the current story may not be very useful: if I am tyring to understand what the reason is for a drop in profitability in my store, knowing that I sell cereal packages in 3 different colors may not be all that relevant -> I highly doubt this would be the 'insight' we are trying to get to.

One more thing to hopefully help you a little bit here: ask yourself what the story is, what we are doing to do with this data.

Was this answer helpful?
Anonymous replied on Oct 22, 2018

Also, look for structural breaks - why does one cost category massively change year-on-year? Why did growth stop? Why did revenue composition change drastically?

Large organisation tend to have a certain inertia - so unless you are looking at an ultra-high growth startup, you should not see wild swings in various revenue or cost composition. If there are, that's an indication that something is happening.

Was this answer helpful?
Egor gave the best answer


Collected McKinsey & BCG offers/ Ex-McKinsey consultant/Harvard/WBS/MSU
Q&A Upvotes
2 Reviews
How likely are you to recommend us to a friend or fellow student?
0 = Not likely
10 = Very likely