Daddy Awesome's Writings

All the things I am writing good or bad

Project maintained by daddyawesome Hosted on GitHub Pages —

Getting Grounded on Analytics

Notes on Getting Grounded on Analytics

Back

Week 1 Assessment

The following 10 competencies apply to teams comprising not only of data engineers and data scientists but also to a new emerging segment of Analytics-enabled professionals including data stewards, functional analysts, and analytics managers.

The first 4 competencies comprise the Business and Organization skills:

The next 5 competencies comprise the Technical skills:

The last competency is on Workplace skills:

Each of these competencies has a 3-level proficiency expectation as part of a toolkit.

● Level 1: Entry Level

● Level 2: Intermediate Level

● Level 3: Expert Level

Business and Organization Skills

Technical Skills

Workplace Skills

21st Century Skills

Many actually call these skills soft skills but there’s really nothing “soft” about these skills.

These skills are harder to develop as they take the longest time and is further influenced by the culture and the environment that an individual is exposed to.

WEEK 3

**The DELTA+Model **

These seven elements are:
D for integrated, high-quality, and easily accessible data
E for managing Analytics resources in a coordinated fashion across the enterprise
L for strong, committed leadership that understands the importance of analytics and constantly advocates their use in decision and actions
T for selecting the right, strategic, organizational targets that will be the core of an Analytics roadmap
A for nurturing high-performing analytics professionals

The continued growth of big data and the introduction of new Analytics techniques like machine learning provided the + in the DELTA+ model:

T for the technologies that will support Analytics across the organization
A for the various analytical techniques ranging from simple descriptive statistics to machine learning Let’s look at each of these elements.

As we go through them, do a self-assessment of where you think your organization is.

D is for Data.

It is no secret that many organizations face data quality issues.

For meaningful Analytics to happen, organizations must ensure that high quality data is organized and accessible by the right people.

Make an assessment about Data in your organization:

Level 1: Inconsistent, poor-quality, and unstandardized data; difficult to do substantial analysis; no groups with strong data orientation

Level 2: Standardize and structured data, mostly in functional or process silos; senior executives do not discuss data management

Level 3: Key data domains identified, and central data repositories created

Level 4: Integrated, accurate, common data in central repositories; data still mainly an IT matter, little unique data

Level 5: Relentless search for new data and metrics leveraging structured and unstructured data (e.g., text, video); data viewed as a strategic asset

E is for Enterprise.

Analytical organizations advocate a single and consistent perspective for Analytics across the enterprise. This is accomplished by setting an Analytics strategy and building a roadmap to implement that strategy.

Make an assessment about your organization as an analytical Enterprise:

Level 1: No enterprise perspective on data or analytics; poorly integrated systems

Level 2: Islands of data, technology, and expertise deliver local value

Level 3: Process or business unit focus for analytics; infrastructure for analytics beginning to coalesce

Level 4: Key data, technology, and analytics professionals managed from an enterprise perspective

Level 5: Key analytical resources focused on enterprise priorities and differentiation

L is for Leadership.

Analytical organizations have leaders who fully embrace

Analytics and lead company culture towards data-driven decision-making.

Beyond the C-level, all levels of leadership within the enterprise should support Analytics.

Make an assessment about Leaders in your organization:

Level 1: Little awareness of or interest in analytics

Level 2: Local leaders emerge but have little connection

Level 3: Senior leaders recognize importance of analytics and developing analytical capabilities

Level 4: Senior leaders develop analytical plans and build analytical capabilities

Level 5: Strong leaders behave analytically and show passion for analytical competition

T is for Targets.

Analytics efforts must be aligned with specific, strategic targets that are also aligned with the objectives of the organization.

At the highest maturity level, these targets become embedded in the strategic planning process and are considered business initiatives and not just Analytics initiatives.

Make an assessment about the Targets in your organization:

Level 1: No targeting of opportunities

Level 2: Multiple disconnected targets, typically not of strategic importance

Level 3: Analytical efforts coalesce behind a small set of important targets

Level 4: Analytics is centered on a few key business domains with explicit and ambitious outcomes

Level 5: Analytics is integral to the company’s distinctive capability and strategy

A is for Analytics Professionals.

Organizations require analytical talent that covers a range of skills and roles as we have learned in this course.

Once the right people are in place, keeping them motivated with creative and challenging projects is crucial.

Make an assessment about Analytics Professionals in your organization:

Level 1: Few skills that are attached to specific functions

Level 2: Unconnected pockets of analytics professionals; unmanaged mix of skills

Level 3: Analytics professionals recognized as key talent and focused on important business areas

Level 4: Highly capable analytics professionals explicitly recruited, developed, deployed, and engaged

Level 5: World-class professional analytics professionals; cultivation of analytical amateurs across the enterprise

T is for Technology.

As the technology for Analytics rapidly evolves, an organization’s ability to deploy and manage the underlying infrastructure, tools, and technologies become increasingly important.

Make an assessment about Technology in your organization:

Level 1: Desktop technology, standard office packages, poorly integrated systems

Level 2: Individual analytical initiatives, statistical packages, descriptive analytics, database querying, tabulations

Level 3: Enterprise analytical plan, tool and platforms; predictive analytical packages

Level 4: Enterprise analytical plan and processes, cloud-based big data

Level 5: Sophisticated, enterprise-wide big data and analytics infrastructure, cognitive technologies, prescriptive analytics

A is for Analytical Techniques.

With rapidly evolving technology comes a higher level of sophistication from the various analytical techniques that organizations may use in their decision-making process.

This could range from simple descriptive statistics to machine learning.

Make an assessment about Analytical Techniques in your organization:

Level 1: Mostly ad-hoc, simple math, extrapolation, trending

Level 2: Basic statistics, segmentation, database querying, tabulations of key metrics are leveraged to gain insights

Level 3: Simple predictive analytics, classification and clustering, dynamic forecasts

Level 4: Advanced predictive methods deployed to discover insights, advanced optimization, sentiment analytics, text and image analytics

Level 5: Neural nets and deep learning, genetic algorithms, advanced machine learning

The Organizational Maturity Model

Organizations mature their analytical capabilities as they develop in the seven areas of DELTA+.

As introduced in “Competing on Analytics” and developed in “Analytics at Work,” this maturity model helps organizations measure their growth across these areas.

At Level 1, the organization is Analytically Impaired. The organization lacks one or several of the prerequisites for serious analytical work, such as data, analytical skills, or senior management interest.

At Level 2, the organization has Localized Analytics. There are pockets of analytical activity within the organization, but they are not coordinated or focused on strategic targets.

At Level 3, the organization has strong Analytical Aspirations. The organization envisions a more analytical future, has established analytical capabilities, and has a few significant initiatives under way, but progress is slow—often because some critical factors have been too difficult to implement.

At Level 4, the organization becomes an Analytical Organization. The organization has the needed human and technological resources, applies analytics regularly, and realizes benefits across the business. But its strategic focus is not grounded in analytics, and it hasn’t turned analytics to competitive advantage.

At Level 5, the organization is an Analytical Competitor. The organization routinely uses analytics as a distinctive business capability.

It takes an enterprise-wide approach, has committed and involved leadership, and has achieved large-scale results.

So how does your organization fare?

Where do you see strengths in your organization and where do you also see areas for improvement?









Week 4: An Organizational Strategy Road Map for Analytics

What is Startegy

Let’s start with what a strategy is. A strategy:

What we want to highlight in this definition is alignment.

Anything that anyone does in an organization should be aligned to the overall mission and vision of the organization.

If a project or an initiative is not aligned with the organization’s strategy, then there’s really no point in doing that project in the first place. This holds true for Analytics initiatives.

One of the reasons we are not seeing more and more organizations in the Philippines embrace Analytics is that the leadership team is not seeing the value of Analytics projects being initiated within a unit or department, which is usually IT.

Analytics projects are usually started within a unit without aligning the intended output to the organization’s overall objectives.

Thus, when presented to the leadership team, the potential value is lost. Let’s see how we can fix this.

Act 1: A Company’s Vision and Goals

Let’s look into how an Analytics strategy roadmap can be developed for a retailer. Recall that we have the following roles in Analytics: ● Data Steward ● Data Engineer ● Data Scientist ● Functional Analyst ● Analytics Manager Let’s add a few more roles that we can find in most organizations that will be supporting our Analytics team: ● Leadership ● Department Head ● IT Engineer We will show in this example how these roles are involved in the strategy roadmap.

Through hard work and overall business savviness, Aling Maria’s sari-sari store has transformed into a large retail company focused solely on Filipino-made products. Her company’s vision: a Filipino brand for every Filipino product at every Filipino’s home.

As the CEO, she and her LEADERSHIP team sets annual objectives for the company.

This year, she wants to, among others, increase their market share from 15% to 20%.

This is the goal that she cascaded to the entire company at the start of their fiscal year. The various departments took this to hear and developed their own strategies that will contribute to the larger goal of increased market share.

In their annual planning, as a case in point, The DEPARTMENT HEAD of the Marketing Department, said, “In order for us to increase our market share, we need to attract new customers.”

She then asked several ideas from her team who are experts in their own right in marketing.

She got the usual suggestions of increasing marketing campaigns, going into social media, getting celebrities for endorsements.

She agreed, in principle, but she wanted to be more focused and more targeted.

One FUNCTIONAL ANALYST suggested, “How about looking into past sales records and see who are our loyal customers? Perhaps we can deduce who to target based on this information.”

The department head, having read somewhere about data being the new oil and knowing that the leadership team is in support of innovative ideas in the company, said, “Let’s try that. And, oh, by the way, I think our IT department is trying out Analytics and has a team of people doing some proof of concepts. If we need to, we can probably borrow some of their people.”

The functional analyst goes back to her desk and starts looking into the customer database (which she has access to as part of her job).

As she scrolls past, because of her expertise in marketing and in retail, she began to have this nagging feeling that there are certain things about their customers that could tell them who to target.

Looking at the data, she has this hunch that gender, generation, location, and income level could affect a customer’s decision to buy.

She would like to investigate further but she doesn’t know how to proceed.

She remembers what her department head said and went to IT.

She explained her dilemma to the IT department head who quickly (and excitedly) introduced her to their lead DATA SCIENTIST.

The data scientist also got excited as she felt that, finally, she can do something that is aligned to their organization’s agenda.

She talked to the functional analyst and explained to her that what she currently has – that nagging feeling, that hunch – is actually a hypothesis.

And that is how Analytics projects are started: with a hunch that needs to be proven by data.

Act 2: A Proof of Concept

As part of the research method, the data scientist further explained that they need more data points to test their hypothesis.

She would definitely need access to data.

With the functional analyst, the data scientist went to the DATA STEWARD of the customer and sales data.

As the data keeper, the data steward asked both on what data they would need and why.

She scrutinized their request and questioned the need for sensitive, personally-identifiable information, that is, data that could point to a single known person.

Based on her evaluation, she determined that such information as name, TIN, and actual birth month and date will not matter in the project.

The data scientist agreed, and they were then given the needed data.

Using her skills in statistics and Analytics methods and algorithms, the data scientist went on and trained and tested her model to determine whether gender, generation, location, and income level do affect a customer’s decision to buy.

And, if they do affect decision making, to what degree do they affect that decision.

After several iterations, the data scientist was able to come up with a formula to score a potential customer’s decision to buy.

In her formula, she determined that gender, generation, and income level contributed equally to the decision-making process of customers, but that location doesn’t matter at all.

Furthermore, the formula suggests that millennials would be their most likely customers.

Having only basic domain experience in marketing and retail, the data scientist conferred with the functional analyst.

Upon seeing the result, the functional analyst got excited as her hunch was correct.

Furthermore, based on her domain expertise, she did feel that millennials would be their most likely customers.

At least now, she has data to back her intuition.

The functional analyst went on and setup a meeting with her department head to present their findings.

With the help of the data scientist, she created a presentation with easy-to-understand visualization that is focused on their message, “We need to target millennials”.

The presentation included a high-level view of the research method that was done but it was not technical or even mathematical as these will not have any use for the department head.

At the end of the meeting, the department head congratulated the functional analyst and the data scientist for the presentation.

She was very impressed, and she also felt that what they presented made sense based also on what she has observed.

She also just needed data to back what she observed.

And now that she has, she invited the functional analyst and the data scientist to a meeting with Aling Maria herself.

On the day of the meeting, quite nervous but also confident of the output of their small project, the team presented their findings.

They started with echoing their company’s goal which is to increase market share.

They then presented their research study on which customer segment to attract, discussed their findings, and, towards the end, connected back their project to the company’s overall goal.

Quiet but listening intently the whole time, Aling Maria just smiled . . . and gave two thumbs up!

She gave the marketing department the green light to proceed and even directed them to see what other data they should consider to achieve their goals.

Act 3: Operationalizing Analytics

With the green light to proceed, the marketing department went ahead and implemented the “customer attraction” algorithm of the data scientist.

However, in response to Aling Maria’s ask to look into other data that they should consider, the department head felt that this now needs to turn into a real project.

Using the successful proof-of- concept as a business case, she was given the go ahead to proceed.

To manage a project this big, the department head employs an ANALYTICS MANAGER.

With her project management skills, the analytics manager assembled a team to plan for the project.

She knew that she needed the functional analyst and the data scientist to be part of the team.

But as they now are going to operationalize the proof-of-concept and as they also now need to look into other pieces of information, they need more people in the team.

In their first meeting, the functional analyst, again, gave a hunch – a hypothesis – that product brand and quality matters to buying decisions.

Furthermore, she felt that they need to look into social media and/or customer surveys to get feedback about the products that they sell.

As they needed new data, they met with the data steward who told them what data they can have – including data about their supplier and products which is in a totally separate database system – but they don’t presently collect customer feedback or social media information.

Having some experience managing Analytics projects before, the analytics manager determined that they would also need an IT ENGINEER and a DATA ENGINEER in the team.

The IT ENGINEER would have to develop an application that would get customer feedback.

The DATA ENGINEER would need to bring together data from social media, the customer database, the supplier and products database, and the sales database into a single repository for the data scientist to work on her new algorithm.

With a carefully laid out project plan, the analytics manager monitors the entire project, providing regular updates to the leadership team and addressing challenges in a timely manner.

After data has been consolidated and transformed to pieces of information, the data scientist gained new insights on how product brand and quality, and customer feedback affect customer buying decisions.

With these insights, the functional analyst came up with imperatives for the leadership team not only on customer segments that they need to target but also on the products and brands that they should be selling.

With a story on how data was transformed to information to insights and to imperatives, the entire analytics project team confidently presented their findings to Aling Maria who, again, at the end just smiled and said with confidence, “Our market leadership is assured thanks to data.”

## An Analytics Strategy that Works

What would help make an Analytics project successful? Well, just look back to what you learned in the previous modules.

It’s the DELTA+ model.

  1. You need consolidated and integrated data of the highest quality and accessible to the right people.
  2. You need an enterprise-wide culture that is open to innovation.
  3. You need leaders who are willing to take a chance in what could be potentially new for them.
  4. You need targets that are aligned from the top of the organization down to all the departments and units.
  5. You need the right analytics professionals at the right roles.
  6. You need just the right technology for your project.
  7. You need to know the right analytical techniques to apply to your project.

So, how do you start?

In most cases, you have to do a proof-of-concept first to prove to your leadership that there’s value in Analytics.

In doing a proof-of-concept, select a project for a specific department first making sure that your project’s goal is still aligned with the overall goal of the organization.

Have targets that could demonstrate value, for example, return on investment, or productivity gains, or cost savings.

Do you need already all the analytics roles to begin with? No.

From the use case, the proof-of- concept was just done by a functional analyst and a data scientist. You need a functional analyst with really strong domain expertise to start off with a hypothesis and a really good data scientist to test and prove (or disprove) that hypothesis.

With them and a supportive leadership team, you would have started to build a successful analytics strategy roadmap for your organization.