Analytics translators have the goal of integrating analytics capabilities in a company. They should identify value cases for the business, and they should help deploy data-driven applications to support or automate intelligent decision making.
The role is new and not yet well defined. Attention, it should be clear that this role or related roles will continuously evolve, and as well the role name might change.
This article tries to illuminate the vast possibilities within your analytics translator journey. The article is an excerpt from my lecture class, ‘The Analytics Translator.’
Please remember that an article or as well a MOOC can mostly teach facts. However, for the translator role, one has typically gone through different positions and situations in practice with constant feedback:
Each learning path is different. Thus the article will focus on a better understanding of the connected terminology between multiple subjects.
On Overall Skills: Business, IT, Analytics Literacy
The top domains for an analytics translater are fluent speaking and understanding of business know-how, IT constraints, and analytics understanding.
Especially understanding the domain know-how of your vertical (industry sector) is of high relevance. Independent of the industry sector, like retail, manufacturing, information technology, finance, or others, you have to understand the business’s logic, customers, and market.
Typically it is not possible to learn or to know all relevant topics. The most classical profile is a T-profile, which means one is an expert in one domain and understands other domains’ issues.
However, after conducting several interviews and rethinking critical skills, I would like to highlight the mandatory soft factors to become an excellent translater. Since every translator will have to moderate and translate between people, it requires multiple soft skills; these three seem to be most crucial (not only for this role): Empathy: to understand the needs, pains of all stakeholders, the clients, the different skill levels, and personal circumstances. Resilience: you will experience many situations that might be frustrating since, during the translation and mandatory change processes, you will experience a lot of resistance. Collaboration mindset: it is all about teamwork and engaging with many stakeholders.
The most successful analytics people are not motivated by reaching one particular goal. They’re motivated by pushing the limits into the distance over and over again. Leaving this role by heart, you will be continuously in a world of new topics and change.
Data Science vs. Analytics Translator
As an analytics translater, you are fluent in speaking and understanding business know-how, IT constraints, and analytics understanding.
Yes, it would be best if you had know-how in all three domains as well as a data scientist. However, you could argue that the analytics translator has the full business focus in each of these domains. The closer a data scientists move to a business department, the more his role will turn into an analytics translator job profile.
An analytics translator and data scientists are the perfect tandem to bridge the inner interception of math, business, and IT.
For an analytics translator’s role, the three top-level domains are now highlighted and decomposed in their generic groups. We (I conducted many interviews ) selected these 15 generic domain clusters for the lecture class carefully, always teased with a very high-level description. Each domain abstract comes with one piece of advice and a good reference to follow up when interested.
In my opinion, revisiting a top-level perspective is essentially independent of how deep your domain skill is.
Focus is to enable you to rethink and follow up on this domain when interested, always supported by reading advice for in-depth clarification.
On High-level management
Management characterizes the process of leading and directing an organization towards a goal through the allocation of resources (human, financial, material, tasks….).
As an analytics translator, you should continuously develop your management skills.
Always operate and manage in the broader context of your values (integrity), and these should be coherent to the organizational values, respectively.
Watch out & literature (The Five Most Important Questions You Will Ever Ask About Your Organization by Peter F. Drucker and Frances Hesselbein)
Either you are already an analytics translator, you are new in this role, or you are on your way to becoming a translator. You should regularly revisit the 5 important questions by Peter Drucker. Our advice: write down your answers in a narrative.
A narrative focusing on these five questions will explain the ‘why’ of a product, a service, or an entire company. A well-written narrative can inspire teams to work together or can articulate the purpose for yourself as well.
On Problem Solving Skills
Problems are everywhere, and these need to be solved. Part of your job and one key competence you have to develop is the art of problem-solving. Problems are everywhere which need to be solved. The path from framing a problem to solving it is complicated. Note that most problems you can not solve alone. As-a an analytics translator, you should internalize best practices on delivering results for a given situation.
According to the future jobs report world economic forum, problem-solving skills are supposed to be among the top future skill sets.
As a starting point for each problem-solving technique is the exact definition of the problem. Note that problem descriptions are often hypothesis; the precise definition might be iteratively re-phrased.
Watch out & literature (Bulletproof Problem Solving: The One Skill That Changes Everything, Charles Conn und Robert McLean)
You should especially take care of the disassemble process to decompose the problem into parts. This process is supported by multiple creative techniques (brainstorming, design-thinking, etc.) and requires close attention in the entire problem-solving process. Often this is related to tree-visualization or tree decomposition techniques. While doing the problem decomposition, you should always have these high-level questions in mind. What precisely is the problem, and for whom do we solve it? Which topics belong to the problem, and what is in and out? Which part should tackle first, which parts next? Can each hypothesis backed by analytics? How could you communicate a result, and what is necessary to convince stakeholders?
One of the most important steps is to understand the objective and the stakeholders behind it. Is it a company objective or a personal opinion, and do you have to sell the idea for an analytics solution to multiple stakeholders? Identifying the correct stakeholders can best be learned when looking at the Sales practice.
On Sales Practice
I needed a long time to understand or even accept the importance of sales and its dynamics.
Selling a product or service to an external customer or pushing a value case inside an organization often follows the same logical sequence.
Peter Drucker stated: ‘The purpose of a business is to create a customer.’
My opinion is: The purpose of an analytics translater is to create believers.
The sales process is more than the pitch itself. It is rigorous thinking and the process of closing a deal. The sales process is often identical in-side an organization to organize budget and stakeholders for an idea.
Watch out & literature (The New Strategic Selling: The Unique Sales System Proven Successful by the World’s Best Companies, Robert B. Miller , Stephen E. Heiman , et al.,)
One of the most straightforward questions to qualify an opportunity to answer is BANT rules (budget, authority, need, timeline)
For a successful push of an analytics project within or outside your organization, first, you have to answer the (BANT) questions: Who is the budget owner, and where does the money come from? Who has the authority, or who is your champion to help you push? How valuable is the current need? What is the timeline?
In the problem-solving section and the management literacy abstract, it is all about listening to your clients. I fully agree you will have many different stakeholders, sometimes with their ideas, pulling you into different directions.
Take care of a good prioritization of activities, and for this, you should apply elements from product management.
On Product Management
Product Management (PM) has the goal to create products/services customers love. Many questions to be answered within PM are tailored towards the (market) environment questions, the unique selling point, and prioritization.
As a Product Management, you should have a passion for your product, the same holds for an analytics translator and his activities, services, products.
What is feasible now, what in the future, and how to test your product as early as possible (minimum viable product) to get feedback.
Watch out & literature (Inspired: How To Create Products Customers Love by Marty Cagan)
One of the hardest things in product management is about saying no. One of the most dangerous phrases in product management is: We might need this!
One of the most dangerous phrases as an analytics translator: we might need this data
By all means, you have to listen to your clients and gather every information you can get. Still, you have to prioritize, prioritize, prioritize. Mainly, which data should you handle first? Boiling the entire big data lake will bring you in trouble and will burn resources.
The art of prioritization can only be solved when working hard on and with your product /data - eat your own dog food.
During each discussion with stakeholders and within each phase of your analytic journey, you will collect or generate new data.
On Data Management
Data is the heart of everything within a digital journey. The art of data gathering, preparing, and processing it is an art. In summary, it is all about data management.
Correct data handling is relevant to all stakeholders in the digital enterprise ecosystem, from business users to IT realization. Only when all agree on the importance of data, you will have success.
Watch out & reference (https://www.go-fair.org/fair-principles/)
Agreeing to the importance of data is one thing, following the underlying principles one other.
One of the top-level guiding frameworks is FAIR principles. It is about findable, accessible, interoperable, reusable. These top principles are simple to understand - still, you have to live-it!
Take special care that data is machine-readable, which is one top directive for analytics and automation. You will have a lot of fun with business excel nightmare :-)
Ok, data has to be accessible, and someone has to process it.
On Software Design
Software design is the process of envisioning and defining software solutions to a given problem.
As an analytics translator, it is beneficial to understand how software design works, why it takes so long, or how to ensure fast delivery cycles.
Priority is to learn and understand agile development and continuous delivery to ensure a fast market response.
Watch out & literature (https://agilemanifesto.org, Clean Code: A Handbook of Agile Software Craftsmanship, R. C. Martin”)
Not everybody analytics translator needs knowledge in a programming language. However, it helps to do small things alone and better understand software engineers’ mindset. Thus, one piece of advice is to learn some hacking skills, which means learning one specif programming language.
Buy all means, look out for the art of agile delivery. The agile working model is hype in business or management as well. Learn from well-aligned software teams. You will learn a lot about SCUM or KANBAN cycles!
During the entire development process, never forget that you have to deliver value for business.
On Proof-of-Values
A proof-of-concept is more tailored towards the proof of a technical realization. However, in the analytics space, we often have the situation that we can not be sure that data delivers new value - or we still need convincing arguments for the next more significant step.
Proof as fast as possible the value, especially the speed, is essential with constant feedback cycles.
Watch out & literature (CRISP-DM: https://en.wikipedia.org/wiki/Cross-industry_standard_process_for_data_mining)
The CRISP process is a proven method to apply the data-driven value cycle from business understanding to delivery, focusing on iterative delivery cycles.
Within the CRISP process, you will spend 70-80% of the time within the project’s first phases, within the link of the business problem statement and data understanding.
It is all about the semantics and meaning of the data. The art of selecting the correct data at the beginning of the process is vital for speed. As a translator, you should focus on this part for quick and proper delivery.
On Team Building
Only with a strong team you can deliver results; The grouping of the right skills paired with an inspiring goal is vital for success.
A golden rule in analytics is: innovation shows up when small teams work directly with the customer/business.
Of course, many soft factors for team building have to be developed over time like, trust, empowerment, and fun.
Watch out & literature (Organize for Complexity: How to Get Life Back Into Work to Build the High-Performance Organization, N. Pflaeging and P. Steinmann)
Analytics is a team sport since you can not solve it alone. Most of the problems are too complicated to solve.
There are often some lone fighters in the business, IT, or analytics who think they can solve everything by themself. Try to balance the level of knowledge.
Take care that the team is assembled for the required skills. Focus on diverse skills across the business, IT, and analytic know-how.
And take care that the team does not detach from the business. Losing the business focus might be the biggest challenge and is at the heart of your analytics translator job profile. Analytics problems to solve with prooven value is often more complex than the initial PowerPoint presentation. Most issues are complex, and you should design your team accordingly.
Project Management
You have a team and a task to be complete to reach a specific outcome, and someone has to be the bull terrier to achieve it: welcome in the challenging world of project management.
As an analytics translator, you will be in charge of managing something towards a deadline sooner or later. Within all tasks described in this lecture, the duty of project management is often crystal clear.
Still often challenging and contradicting to other skills within an analytics translator journey with less room for interpretation.
Watch out & literature (A Guide to the Project Management Body of Knowledge (PMBOK Guide), PMI.org)
As a project manager, you need for sure resilience, and hopefully, you don’t lose a positive attitude.
Project management is divided into phases and respective deliverables. Watch out to be very good at communicating the objectives, expectations, and often unpleasant situations.
The art of persistent claiming without burning your partner is the skill to learn from good project management practice. Learn the art of expectation management, which is often around clear communication of deadlines, tasks, and resource problems. A clear structure is essential; maybe you don’t need all aspects of the PMBOCK since we all like to work in an agile way. But deadlines are still crucial and many other things as well.
On Analytics
The term analytics often holds a duality in the meaning, the break down of a problem, and the hypothesis’s data proof. An analysis is a process of breaking a topic into smaller parts to gain a better understanding. It is the art of forming a hypothesis. The proving or reasoning of an idea on data falls into the field of statistics.
Both skills are just essentials for an analytics translator and bridge the business to the mathematical world.
Watch out & practice advise (https://www.kaggle.com)
The technical doing and working with data falls into the field of statistics. Statistics is the discipline that concerns the collection, organization, analysis, interpretation, and presentation of data.
The art of statistics can be learned and thought precisely via fantastic books, MOOCs, lectures.
Statistics are always present whenever you work with data or when pushing data applications towards business. Take care that you have a solid basis of statistics. Otherwise, you will end up in questions like:
Why should your mathematics be better than mine? Don’t fight on mathematical accuracy when you can not measure the business outcome.
We proved the value, maybe on a small example or if it was a one-off analytics task, all good.
Your focus should always be on integrating your solution within the enterprise architecture for continuous value delivery.
Here, often a skill gap arises.
On Enterprise Architecture
All analytics capabilities you build by human resources, by solution development, everything discussed in this lecture has to be embedded in the company’s overall foundational execution technology setup.
“The enterprise architecture is the organizing logic for business processes and IT infrastructure, reflecting the company’s operating model’s integration and standardization requirements.” (see book ref.)
Every adequate foundation for any analytics execution depends on tight alignment between business objectives and IT capabilities.
Watch out & literature (Enterprise Architecture As Strategy: Creating a Foundation for Business Execution by Jeanne W. Ross , Peter Weill)
Watch out for the underlying operating model, which defines the necessary business process integration and standardization for delivering goods and services to customers.
Only when you know which core processes are relevant for your organization, you will be able to embed analytics solutions.
Analytics and every data-driven endeavor require top-level management back-up, thus focus on topics aligned with your organization’s strategic path.
Given a good enterprise architecture design, you will find repeatable, stable processes, and thus your application will benefit from economies of scale.
On Process Modeling
Business processes are used to standardize the execution regardless of who is performing the activities. Good processes and their implementation can increase the market competitiveness.
It can support faster innovation cycles, a better quality of your end product, better customer service, and many other value levers of your business. Overloaded, complex, or nonstandardized processes will slow you down in each activity.
As an analytics translator involved in new applications, you should know the mechanics behind process design.
Watch out & literature (https://www.bpmn.org)
With new analytics applications, you should take special care of the future to be modeling. Automated analytics and new human to machine interfaces will give an entirely new process flow focusing on minimum touchpoints between the user and computer/machine.
The focus should always be usability and the human in the loop of an analytics application.
The to-be design of future processes might be very tricky since you have to design new fields and new technological possibilities. Take care not to over-engineer processes with limited content.
Note that every change in a process will affect humans’ work and be directly linked to change management.
On Decision Science
Decision science is the art to pragmatically formulate analytics results in an executable decision within a process.
Good decision making is an art, while we should always distinguish between decisions taken by humans and automated decisions by computers. We all have to make decisions, mostly on incomplete information continuously. This science is linked to behavior science/economics. Many decisions within a company can be automated, which is then linked more to the art of data science.
In both cases or hybrid cases, delivering valuable decisions with data support means deciding under uncertainty.
It describes the edge between theoretical analytics and applied analytics.
Watch out & literature (Thinking, Fast and Slow, D. Kahneman. Storytelling with Data: A Data Visualization Guide for Professionals, C.N. Knaflic )
How does data-driven flow back in companies’ decisions, how are they integrated into the business process and the overall enterprise framework?
Even with the best derived analytical solution, you will fight over and over against systematic distortions towards opinions, gut-feelings, wrongly selected data sets.
The topic of bias is a challenge for human-made decisions, and machine inferred decisions, respectively.
A human bias is a systematic pattern of deviation from the norm in judgment where people tend to replace a complicated question with one which is easy to answer.
Statistical bias is disproportionate weight in favor of or against a business hypothesis. This is mostly caused by a data set analyzed, which is not representative of the problem.
An analytics translator develops a sense of the difference between human bias and statistical bias.
On Artificial Intelligence
Every decision based on data will phase sooner or later a challenge of higher automation and the need towards intelligent execution.
The ‘I’ in every solution is towards the intelligent automation of everything.
Artificial Intelligence is doubtless one of the biggest trends these days, and new applications and ideas pop up everywhere. Ensure that you understand the direction and feasibility of the ongoing development.
Watch out & literature (Life 3.0: Being Human in the Age of Artificial Intelligence, Max Tegmark)
The analytics translator’s biggest challenge is balancing hope, fears, wishes, and myth in this topic.
Watch out for the complexity of this topic, and try to embed services that can be consumed rather than develop new AI applications.
By all means, take care of the human-centric design. The user has to accept the solution and should be integrated with the highest ethical standards.
On Causality
Understanding causality is the holy grail of decision making and analytics. It describes the relationship between a cause and its effect. There are often complex dependencies between actions one can take, and the final seen result in complex environments.
As an analytics translator, you will often face the challenge of understanding the critical influencing factors of a decision; usually, there are many hidden factors and biases.
Correlation does not imply causation, and developing a sense for the dependencies will lead to better decisions.
Watch out & literature (The Book of Why: The New Science of Cause and Effect by J. Pearl and D. Mackenzie)
Bringing things into a causal chain can not be proven on historical data.
The right balance between defining a valid hypothesis to solve a business problem and setting up and performing the initial ‘test’ within the complex reality is the thin line between talking vs. performing.
The causal theory is the ultimate push and call to action for doing something.
In summary, doing it and measuring it over endless planning. producing results vs. watching results running an A/B test or just speaking about it practicing a new topic or just consuming information
You have to move something and measure the result to derive causality actively.
The same holds for your analytics translator journey. You have to do the job, maybe with missing know-how and lack of information.
Doing is better than reading Cheers :-)