Advanced Analytics, characterized by Data Science, Machine Learning, Artificial Intelligence, and Engineering – which includes big data & ML engineering, as well as specific activities in application engineering & cloud/infrastructure, is an area that offers analytics career opportunities for individuals with a diverse set of skills.
Successful companies (or a company’s internal analytics teams) in this space stand out for their ability to deliver significant & measurable business impact to their clients (again, internal or external). This is achieved by bringing diverse talent together, enabling them to collaborate in a great culture of disciplined innovation and exploration for value. Underneath all the roles, titles, terminology, tools, techniques, frameworks that are often talked about, are two foundational broad areas of capability:
The table below shows some of the specific capabilities within the technical & consulting streams. In Human Resources & Organizational Development parlance, these could be thought of as the analytics Career Streams / Role Families.
As we prepare to take a deep dive into one of these, a few points to note:
– These two are not mutually exclusive. Sometimes, individuals with top-notch technical capabilities are also very good at consulting and vice versa.
– Either one, or both, is required in different proportions along an advanced analytics program lifecycle that includes – discovering and defining business problems, translating those to analytic themes, designing & developing robust solutions, driving adoption, and measuring & articulating value.
– Individuals looking to build an advanced analytics career, either from scratch or by switching from another industry, have a higher chance of success if they are clear on where their capabilities are (technical or consulting) and which one truly excites them. Both being the same makes it easy!
As it is a significant exercise to talk about all the capabilities (or analytics career streams/role families) in a single blog, in this one, our focus is on the Analytics Consulting stream. We will talk about the Technology stream in upcoming blogs.
Analytics Consultants contribute hands-on in aspects of solution design & development, coordinate with the solution development team (such as data scientists/engineers), and have a relatively higher frequency of interaction with client stakeholders. They ensure that the overall goal of an advanced analytics program is achieved.
With balanced techno-functional expertise, they have a role to play in-the-markets (‘onsite’) – in upfront problem discovery & solution blueprinting phases, as well as at offshore/remote delivery locations – to curate insights and shape up the story as a solution takes shape. They also coordinate other parts of solution development. Their interactions are with client teams that roll up into the VP of Analytics/CAO/CDAO, or the office in charge of corporate strategy, or one of the business sponsors.
In comparison, Business/Domain Consulting roles bring deeper expertise around a particular industry or a function (and less on technical competencies), and are leveraged based on a specific need and mostly as in-market roles. Similarly, Technology Consulting roles focus more on enterprise data/technology roadmap development and program delivery, consulting the CIO/CTO organizations. In an advanced analytics organization, all 3 have a role to play, but in different proportions.
The diagram below depicts the competency dimensions & specific competencies required for Analytics Consultants
At Tiger Analytics, each one of the competencies has clear definitions, levels (say from 1-5) indicating gradual mastery, and training – online & classroom/workshops that enable cross and upskilling.
A brief description of the competencies, and how they are relevant in program settings are given below:
Technical Competency Dimension
– Extraction & preparation of data refers to the ability to talk to data through SQL (there are many dialects of SQL for different platforms, but a strong foundation in at least one is recommended)
– Ability to perform data analysis or build statistical/ML/AI models (using R/Python as the base)
– Ability to create visualizations (using Tableau/PowerBI or open-source libraries/ & packages suitable for R/Python).
In all of the above, there are two key components: one is to learn how to work with the tools, and the second is the logical skill component – knowing what to do with those tools. For example, clarity on “what’s the best way to aggregate this data for my analysis?” is needed even before you start writing an SQL “SELECT…..” statement.
Business <-> Analytics Translation
For Analytics Consultants, this is the most important competency dimension. Individuals in other streams will also benefit if they could afford to spend time and pick these up as they grow in their analytics careers. By the way, if you are wondering about <->, it refers to the ability to translate business problems to analytics themes, and analytics solutions back to business value. The specific competencies required within this dimension are:
Problem Discovery & Definition – ‘Our sales is going down’, for example, is not a problem statement that can be readily solved with a prescription algorithm that runs once in the morning-afternoon-and-night. At best, it’s taken as a symptom of an underlying problem before jumping to any kind of solution. Sometimes, there may not even be a problem. CXOs often ask their analytics teams to come up with strategies on ‘How can we be better than what we already are, through the best use of data: meaningfully (and ethically)?’ Structured approaches such as ‘design thinking’, ‘issue trees’, and ‘hypotheses formulation’ are important if the goal is to realize value by identifying and solving the right problems vs solving any problem.
Solution Blueprint Development – After isolating a problem, or a set of problems, how does one prepare a solution outline? Think of a building plan/blueprint. Even before the solution is developed, a blueprint gives a very good feel for the final outputs, and possibly, outcomes as well. There are structured ways of providing an outline of an analytics solution moving from hypotheses -> data sources -> specific data features & data granularity -> potential data enrichment -> exploratory insights -> model development & validation -> tracing potential impact of model metrics on business & financial metrics -> how outputs would be presented to business to ensure adoption -> initial deployment -> scaling across geos/business units. Communicating a plan for these without overwhelming the stakeholders is critical to avoid mismatch in expectations.
(Note: The unidirectional flow of steps depicted above is just for convenience. There is a lot of back-and-forth between steps helped by structured feedback, which needs to be well managed at run time to stay on course. I am resisting the use of the phrase ‘in an agile manner’ – since agile, when not applied well, degrades into an exercise in “rain dance” with the hope it will rain, as called out in this HBR article. I view that as an indication that agile approach/application in advanced analytics has to mature further.)
Storytelling with Data – Data, static charts, appealing & interactive visualizations, or even the models aren’t the story by themselves. The insights these artifacts contain, what they mean in a specific business decision context (and equally importantly, in the context of a business persona with immediate goals, future aspirations, constraints, past good/bad experiences), and how the insights can drive decisions that help realize value, told in a way that drives adoption and a bias for right action – that is storytelling! Again, easier said than done. Learning content from Jon Schwabish, and Cole Nussbaumer are immensely useful in knowing more about storytelling.
There is another reason why storytelling is important. Developing a robust solution takes time – even the fastest takes a few weeks, while businesses always want their solutions yesterday (been there, right!?). In such instances, storytelling is an effective way to keep your audience (stakeholders) riveted, without losing interest or getting impatient. Providing insights against key hypotheses and a glimpse of value early-on provides much-needed air-cover even as the solution is being developed by technical experts. Analytics consultants have a key role to play in this.
Business Value Articulation – Top-notch strategy consulting firms – even if they are not renowned for their analytics practice, are pioneers in this area. For example, you could communicate “a 10% improvement in forecast accuracy or reduction in MAPE, helped drive down days-on-hand inventory (say, by 3 days, from the current baseline), which translated to a USD 35MM in working capital savings”. Add to that how it compares with best-in-class & laggard peers/competitors in the industry in a single Powerpoint slide: it becomes easy to communicate value realized through advanced analytics even with executive sponsors when we translate from model metrics to business & financial metrics they understand well. (Of course, be sure to back it up detailed calculations & appropriate assumptions.)
Business/Domain Context – Of course, all of the above competencies have to be applied in the context of a specific industry or a business function. However, if you have a very good grounding on the above, even with basic exposure to an industry or a function, it is easier to build incremental understanding by participating in projects and applying the above consistently. (I have observed individuals with good knowledge about an industry, but lacking in the other competencies listed above, struggle to make an impact as analytics consultants).
Program Management: Third Competency Dimension
I will just include a brief description of this. Project management – deftly balancing scope, pacing, and cost; team management; client management; and quality management are the specific competencies within this dimension.
These may sound like stating the obvious, but I will give two examples to drive the need for depth on these. Client management is often referred to as the ability to push back. How can you push back (without repercussions) unless you build a good rapport with your client based on trust & being objective? Also, quality is not meeting SLAs – which is usually focused on timeliness, accuracy (or the absence of defects). Yes, these are important, but how do you know if you are adding value to your client business. Being green on SLAs is a necessary, but not sufficient condition to deliver client value.
Is each competency equally important to an Analytics Consultant?
With the analytics consulting competencies outlined, the next logical question that arises is: “are all the dimensions and competencies equally relevant to every analytics consultant, irrespective of their level of experience and responsibilities on a program?”
Analytics consultants are usually leveraged as:
– Individual contributors (referred to as analytics consultant), in which case the approximate time split between the 3-dimensions is typically 60-70% on technical, 20-30% business<->analytics translation, and about 10% time on program management competency dimension.
– A lead on a particular client program (referred as analytics program lead), in which case the time spend across the 3 dimensions are more like 30%, 40%, 30% across the 3 dimensions
– Or, to own and drive a portfolio of programs (called Delivery Partner*) for which the split becomes 10% on technical competency dimension and about 40-50% time on the other two dimensions. [*Client Partner is the equivalent in-market role, which also has revenue growth responsibilities]
If you are reading this blog while considering an advanced analytics career, it is very important to do a thorough evaluation and decide whether to aim for a technical or consulting role, based on:
– Your academic background, work experience so far (if any), and skills acquired through additional training
– The competencies & skills it takes to play the technical and consulting roles, and
– Which of the two areas truly excites you (this is often the toughest, and requires deep introspection)
A good self-assessment of these should help you make the best decision. Guidance in the form of articles – this one, hopefully; and interactions with current practitioners in this space could also help arrive at the best decision. If you are someone who has already gone through this journey, relate to the thoughts here or have additional/different thoughts to share, please do.