5 Email writing tips for analysts

“In labouring to be concise, I become obscure”

from Horace (poet)

 

The key task of an analyst is explain complex concepts simply. Below are 5 tips that help me write effective email messages.

Email writing tips 

1. Start your email message with a topic sentence
2. Use proper names instead of pronouns
3. Provide examples if asking to decide between one option versus another option 
4. Keep it short  
5. Use bullet points when writing options 

1. Start your email message with a topic sentence

Most executives and decision makers view their emails on a smart phone, thus showing a preview of the  first sentence of every email. That's a perfect window to grab their with the following:

• if you need something, make your first sentence a question. 
• If you have something to show then state in your first sentence that "the attached file provides..."

2.  Use proper names instead of pronouns.

I suggest you forget the words he, she, we, and it (especially "we" and "it"). Replace them with proper names, objects, and concepts.  This way, the reader isn't confused about who or what you are talking about. 

3. Provide examples if asking to decide between one option versus another option

Just like ordering food from a fast food restuarant, it's nice to see options. For example, if you're asking the reader to decide if a report is to display a physician's name by [first name] [last name] or [last name], [first name]. Why not take the extra minute and provide real names from your organization. 

4. Keep it short

Busy executives and decision makers receive a lot of emails. There isn't time to read novels. State what you want in your topic sentence, provide a couple of supporting bullet points or paragraphs, and get out. If your reader needs more info, let them ask. 

5. Use bullet points when writing options

A great way to break up the monotony of an email is to use bullet points. I suggest using a maximum of three. There is something powerful in the number three. 

As I read the poet Horace's quote above, I believe that  a well written email has less "I" words and more words that describe the issue or concept.

What did I miss?

What does an analyst do? Question, Investigate, and Communicate

“The Golden Rule of Habit Change: You can't extinguish a bad habit, you can only change it.”

Charles Duhigg, The Power of Habit: Why We Do What We Do in Life and Business

I recently read the following article, Aviate, Navigate, Communicate: Business Crisis Management From a Pilots Perspective. It struck an accord because in another life time, I was a young Student Naval Flight Officer in the U.S. Navy. The phrase, Aviate, Navigate, Communicate, is a habit drummed into aviators to utilize when an emergency happens in the air. It represents the basic activities needed to keep one alive while troubleshooting an issue. It’s a habit that requires actions and no thoughts to follow.

“Aviate, Navigate, Communicate” led me to devise “Question, Investigate, Communicate” for my profession in analytics.

Question

At the very minimum, an analyst needs to understand the business problem. Some call it requirements gathering, writing specifications, or just plainly “what’s the question?”

I like to ask the following questions:

• What’s the issue/problem?
• Who else needs to see this info?
• Where do you plan to use this information? Meeting room? Smart phone?
• What does the final output look like?
• When is it needed?
• How do you (business user) plan to use this info?

If the above questions look similar a journalist’s “the 5 W’s” then you’d be right. I often imagine that I’m a journalist when I’m first presented with an issue or request. I treat each request with a healthy dose of skepticism.

Investigate

Now that I know what the core problem is, I need to develop options or possible solutions.

Investigation can take on several forms such as:

• Data extraction from databases to produce a data report 
• ad-hoc querying against corporate data-sets to satisfy a “what if” question 
• Descriptive statistics applied to a data-set to understand 
• Data Visualizations in the form of charts and maps 
• A data model created to describe a population with embedded business logic 
• Facilitation or Negotiation with varying parties to arrive at a set of options 

I like this quote from Rahm Emanuel, “Don’t come in here and dump a problem. I have a whole desk full of those. Bring a set of solutions.” That is analysis. It attempts to distill a problem to core components and offers various thoughts on how to describe or solve the problem

Communicate

No matter how brilliant a piece of analysis is, without communicating the results to the right person at the right time then analysis isn’t complete. Communication is actually the cornerstone to great analysis. Think about it, the task of analysis is to dissect key facts from vast amounts of information and transform it into useful “bit-size and flavorful” bites.

Communicating results can take on many forms:

• report/dashboard
• excel model
• e-mail/memo
• presentation in a meeting
• phone call/Face-to-face interaction

 No matter the form, one must be mindful to how much or how little information to give to the business user.

When the stress level is high and time is short, remember to “Question, Analyze, and Communicate”

Trust me, data cubes are great

"Come with me if you want to live" 
 Arnold Schwarzenegger in Terminator 2: Judgment Day 


Two thoughts come to mind as I remember this quote:

1. Linda Hamilton's character must take a leap of faith and quickly
2. The Terminator has a plan that will weather the upcoming change, i.e. Terminator T-1000's relentless pursuit.

In the case of Healthcare analystics, perhaps the above quote should read something like  "Trust me, this new technology (EHR, data cubes, etc) will work if you intend to survive healthcare reform". 

Trust and new technology...Can that those words be used in the same sentence? I'm optimistic so yes. 

I'm faced with transitioning to data cubes. I'm told I'll have little need for my old skills. The new technology will be more efficient and robust. Over the years, I've become quite proficient with Microsoft Excel/Access and SQL. I feel comfortable. How can I trust the results of a data cube when time is of essence?

Two actions are needed to answer this question:

• Design a cube with a solid specification
• Test the cube with a robust Q/A checklist. This ensures the specification was followed and business logic developed appropriately. 

Given where my mind is today, I want to focus on the Q/A checklist. I googled "cube testing" and most links (this one or this one) are from a developer's perspective. If I relied on these links, it's as if my mechanic believes my car is fixed, but I need to be assured in terms I can understand. 


I developed the list below with the help of Google searches, colleagues, and mostly from frustrating experiences. 

1.    Ensure requested cube attributes are present 

2.    Ensure requested cube attributes have appropriate display names

3.    Challenge the logic behind each attribute. 
     If you are transitioning from a legacy system to a data cube, this is a vital step.

4.    Ensure cube attributes list properties either from the underlying database or custom list.

5.    Ensure relationships between fact table and dimension tables actually work as advertised

6.    Place extreme conditions in to see cube to see how it reacts 

7.    Look for known trends in the legacy system and see if they are present in the data cube

8.    Does the data cube answer specified business questions?

9.    Can data cube based reports replace your legacy reports?

BI tools for data analysis

I recently performed a Google search on selecting Business Intelligence (BI) software tools. The first prominent link led me to read the white paper, How to Choose the right Business Intelligence Technology, I like how the article lays out the various styles of analysis. It goes on to describe the Microsoft tools needed to implement the each style. 

The article got me thinking about my quest to define the role of the analyst. Reading the blog "Reporting vs. Analysis: What’s the Difference?", help formalize my thoughts. Check out Brent Dykes's key thoughts on the differences between reporting and analysis:

Data Reporting is...

• Organizing data into informational summaries 
• following a push approach, where reports are pushed to users who are then expected to extract meaningful insights and take appropriate actions for themselves (i.e., self-serve) 
• Outputs: canned reports, dashboards, and alerts (reports sent to users based on a triggering event) 
• providing no or limited context about what’s happening in the data. In some cases, the end users already possess the necessary context to understand and interpret the data correctly. 
• not going to answer the “so what?” question on its own

Data analysis is...

• Exploring data and reports in order to extract meaningful insights 
• Following a pull approach, where particular data is pulled by an analyst in order to answer specific business questions. 
• Outputs: Ad hoc responses (reports) and Analysis presentations (a comprehensive, deep-dive analysis) 
• Providing context which is critical to good analysis. In order to tell a meaningful story with the data to drive specific actions, context becomes an essential component of the story line. 
• Emphasizing data points that are significant, unique, or special - and explain why they are important to the business 

Upon retrospect, I would add one more style of BI reporting to their list of 5 styles.I call it:

Business Analysis
This style is performed by a business/clinical/health data/data analyst with the purpose of answering a business question that leads to action. The analysis must sniff out the business question through questioning and discovering the appropriate business context. The end user will receive a written analysis with supporting charts and tables. The analyst will need tools that allow to him/her sift through sometimes millions of rows and find a meaningful summaries of this data.

Tool Options to support Business Analysis

MS Excel and Power Pivot for MS Excel

The most versatile tool in Microsoft's arsenal continues to be MS Excel. Excel can consume SQL data extracts, data cubes, and external data from vendors. Utilize various built-in functions and add-ins to perform analysis on the data points. With the Power Pivot add-in installed, Excel merge disparate data sources with relative ease. You can even create your own data cube of the source data.

MS Word
Yes, Word is a tool. One needs to express the end results of analysis in a coherent and organized fashion. Word provides to means to accomplish this.

MS Outlook
I know that it is strange to list Outlook as a BI tool. Here is my thought, rarely is an analysis so compelling that it leads to immediate action. Part of the analsyst's job is to follow-up with the business user. If no action is taken then the analysis was neither useful or needed. Utilizing Tasks and Calendar events to track follow-up conversations is vital to a successful analysis.

Describe impact on an organization
I've heard "we have lots of data but no information". In order to answer a business question, one must transform data into a meaningful story. Given the right data and the appropriate tools, an analyst can create a compelling story that can be understood by many.

I find myself travelling down this path toward insights and actionable data. After walking down the reporting path for many years, this will be a major paradigm shift.