Want to get insights from this forum on some examples of the most quant work/modeling you've done/seen in consulting. Looking to transition into corporate strategy/strategy consulting post MBA and want to understand just how much quantitative work do you actually do on in engagements and how can I get into those lines of work.
While I love the overall strategic thinking and solving very unstructured problems, the level of fluffiness I've experienced through working on a few strategy engagement at a non MBB/non Monitor/Strategy& firm has frustrated me somewhat. I hope this is due to my specific firm and not just across industry (which is why I am not thinking of coming back). I don't have experience working in another consulting firm and unfortunately from speaking with friends working in other big 4 consulting firms, they seem to be doing similar things.
I particularly love the problem solving aspects of consulting that can be intellectually stimulating (may be a little to ideal here), hence not thinking of going into IB via MBA. But correct me if you think otherwise.
It depends on who is running the project, and what type of project it is. I suspect strategy projects will become increasingly quantitative. I'm more on the analytics side, but some example strategy projects I've worked on:
Pricing strategy for a F500 retailer:
Issue: retailer is losing ground to online competitors and needs to adjust its pricing strategy to remain competitive
Data: A few billion rows of transaction level data, customer attributes, and competitor pricing information
Technology: primarily Alteryx, Hadoop & Hive
Models: Modeling was relatively simple here. Assessed customer pricing against that of its primary competitor. Developed models to better understand customer behavior, market baskets frequently purchased, trends by region, customer profiles, etc.
Result: Developed pricing A/B testing plan to improve competitiveness and drive revenue growth. Working with the client to implement.
Marketing strategy for F500 retailer:
Issue: retailer has declining revenue, is reliant on coupons to drive revenue and traffic, and has confused the customer with too many marketing events.
Hundreds of millions of rows of transaction level data, customer attributes, and external customer demographic data sources.
Technology: primarily Tableau, SQL Server & R
Models: High dimensional data clustering on customer product purchases. Hierarchical regression analysis to create customized store baselines and estimate the impact of coupon exclusions on revenue. Analysis of marketing events to better understand the frequency and overlap of marketing events, with models to estimate revenue reduction due to confusion surrounding multiple marketing events.
Result: Reduction in the number of marketing events and reduction in the number of coupon exclusions reducing confusion for both customers and sales staff.
I also perform a significant amount of work in financial due diligence to better understand the drivers (demographic, physical, etc.) of location performance, white space expansion, pricing optimization, customer clustering, sales forecasting, etc.
Long story short the quantitative work is available if you want it (though I can't speak for MBB).
Have you ever seen anyone from Deloitte TC transfer over to S&O?
I'm haven't been with Deloitte for a few years, but I never heard of an instance of it happening. I'm sure it did, but I just wasn't aware of it.
It's essentially unheard of from what I understand. It's akin to someone going from Chase in Columbus, OH to JPM in NYC.
no. you train for and sit different qualifications (at least in the uk)
My good buddy transferred from Deloitte's DC-based Cyber Risk fed consulting arm to the NYC-based Cyber Risk division. He now focuses on banks etc. Understood it's not S&O, but if you're enough of a rock star, it's probably possible.
I'm just wondering what is the expectation for new joiners to your group. Do you expect new joiners to have some experience in using some combination of analytics tools and subsequently learn how to use tools which they are unfamiliar with on the job?
Does your group consists of mainly CS/Math/Statistics majors or Business majors? And what are your experiences working with CS or Business majors in your group?
Traditionally my group has been comprised of business, finance and accounting majors. The team is enabled by a combination of workflow based tools such as Alteryx and SQL. Moving forward we'll continue to recruit and staff traditional hires, but we'll incorporate more data scientists into the team. On the data science side we are looking to bring on folks with at least a masters in data science, comp sci, mathematics, statistics, operations research, engineering or economics with a preference for PhDs.
Folks joining from the business side are generally expected to get up to speed on tools like Alteryx and Tableau. People joining on the data science side will need to be able to pick up skills quickly as new technologies are released at a rapid pace. Most junior data scientists won't have exposure to much beyond R, Python, SQL, Linux command line and some basic hadoop/HiveQL, but the expectation would be that folks quickly get up to speed on some of the more recent technologies.
In terms of working with a mix of data scientists and business folks my preference would be to have both on the same team as they each bring valuable viewpoints. The business folks have a hard time understanding the analytics and can often interpret them incorrectly, and the data scientists often don't have the business acumen or presentation skills. I think it is imperative to have both in the same room to successfully drive insights for clients and provide an opportunity for cross functional learning.
They way I see it ,unstructured problems are problems that cannot be structured , not these that haven't been structured yet. How you can Quant something that cannot be structured? What DeloitteSandO posted is very nice and demanding but these are not unstructured problems, quite the contrary. You cannot big data your way out of Strategic planning, at least yet. Yes ofc you will get supported by quant but "fluffiness" is an organic part of this process
Agree, this is more operational strategy. Was usually just called operations and not strategy
Less structured problem with high-powered analytics; What lessons can we learn from the strategy of firms with sustained high performance (as measured by ROA) that can be applied to other firms?
This seems a simple enough problem. Find companies with high profitability, medium profitability, and low profitability and perform comparative studies to try and discern differences in strategy between them. However, it is very hard to select companies for study.
How long does a company have to be profitable for to have "sustained" high performance performance and not have its high performance simply be a function of natural variation? How does the size of a company impact its profitability? How should you control for company age? How should you control for industry? etc.
You have to leverage some pretty high powered analytics (markov chain monte carlo models, quantile regression, different filtering methods) along with a large database of corporate financials to separate high performing companies from those where high performance is simply an artifact of the natural variation of performance over time.
After you go through all that then you can perform your comparative analysis which may be a bit fluffier. The point is that if you aren't applying bringing structure and data to unstructured problems you are missing an opportunity.
True, didn't say otherwise.
Famous 80/20 tho implies that overdetailing a model won't really help you much as you can't model reality. It just makes you more prone to over rely on it.
Correaltion doesn't mean causality. So you may find correlations of size, management age, industries but the real factors may be something that you haven't accounted for. Maybe all these companies share a something tha you cannot predict with a model.
I know that's super fluffy but I'm trying to illustrate that SOME times the available data won't show you the solution. But I agree as I said. Not taking quant into consideration is a missed opportunity. Haven't worked at MBB but if they don't do it(at least as much as they should) it may be the case that they like to project a " magician" result
OW was historically a FS risk consulting firm. Definitively have some heavy quant work. Same goes for McKinsey's risk practice. Don't think Bain or BCG do much in that space.
Also retail and pricing projects in general can get quite data driven when you get billions of sales data points and need to try to find patterns / insights from that.
Lastly, you can also focus on DDs at e.g. LEK or Bain, if you are able to handle the workload :)
If you're interested in doing tons of quantitative analysis, focus on more data-rich industries or issues. Retail and travel come to mind here. In general, strategy work involves performing just enough analysis to validate a hypothesis. Heavy data crunching, while it can happen, is usually being done within the business or in specialized consulting practices. Increasingly, the consulting business is being disrupted by vertical-specialized analytics firms for whom this kind of work, like pricing & yield management and other fields driven by data science.
I don't work in consulting, but I do corporate strategy. The more quantitative element is actually around what I call "business stoichiometry" and being very, very crisp on what different inputs actually mean and comparing and transforming them properly and intelligently. In physics, you need to distinguish between momentum and force. The same is true for business metrics, but you don't always have the advantage of breaking things down to SI units. Finding the most salient relation between variables, and a way to convey this information effectively to stakeholders in ways that are easy to translate into action, is the area where value is created. We do perform clustering analyses and look at new ways to study problems and answer questions, but they're pretty bespoke solutions and the focus is less on mumbo-jumbo to dazzle a client and more on driving real results with the data that's actually available.
It varies significantly by the client type and nature of the engagement.
For example, if you are doing marketing optimization across products and geographies, it would get very data intensive as you need to derive bottom-up what the incremental revenues are across many dimensions and then provide different scenarios in comparison to what is expected top-down.
For spin-off/ sell-off projects, these are highly numerical and involve financial and valuation modeling - plenty of Excel here.
With regards to segmentation projects, these can be 100% data driven and the use of statistics software is not uncommon due to the sheer volume of data that needs to be cleansed, managed and analysed.
The more interesting question would be around the types of data that need to be analyzed for example: financial, non-financial, customer usage, product usage, channel usage, cost allocation & spend and productivity
Omnis repellat fugiat harum mollitia laudantium occaecati eos sapiente. Adipisci pariatur consequatur aut sit iste. Omnis quibusdam et voluptas possimus ullam eum exercitationem.
Est nostrum sint voluptatem et. In et hic consequatur nihil. Necessitatibus aliquam asperiores sequi aperiam eius pariatur inventore aut.
Iure eos aut exercitationem nam ducimus perspiciatis. Dolorum qui quidem consequatur omnis. Molestiae earum quis totam odit eum quo. Enim laudantium ab esse provident.
Voluptatum doloribus quas quo libero quasi id et. Quisquam dolorum consequuntur fugit. Nihil dolor qui vel maiores facere dolorem beatae voluptate.
See All Comments - 100% Free
WSO depends on everyone being able to pitch in when they know something. Unlock with your email and get bonus: 6 financial modeling lessons free ($199 value)
or Unlock with your social account...