In-Depth

7 Tips for Working with Data Analysts

Seven tips to help your data analyst give you the best insights from your data.

• By Ted Cuzzillo
• 05/04/2011

Data analysts tend to feel like odd ducks. Crucial as they can be to business success, many say that clients just don't know how to work with them for the best results -- failing to explain context, for example, or micromanaging, or just weighing them down in any number of ways.

It takes two to analyze -- and just a few simple things can make the difference between disappointment and golden eggs.

Though it's hard to generalize about this emerging group, I've noticed several trends: First, many data analysts have had little or no formal training for the job. They may be "good at math," they may even know statistics inside and out, and they're all skilled with one or more tools. Many, though, don't know how to tell a story -- with all its power to give context and set facts in memory.

Second, the data analyst's closest relative in business may be the journalist. The best analysts spin stories from their observations of data, they ask a lot of questions -- some of them inconvenient. They know a little about a lot, they follow trails from one clue or question to another, and they need elbow room.

"There's a distorted idea of what data analysts do," said advanced-analytics consultant Theresa Doyon, Ph.D. Her doctorate's in social psychology, and she trained later to use SAS and other tools. She's seen job listings that ask for high-end data scientists or a degree in computer science for database marketing analysis, which actually requires far lower qualifications than that.

True, analysts are a kind of quant. They're also a kind of artist, which is usually not the kind of beast that finds an easy nest in most businesses.

Here are seven tips to help you get the best results from your data analyst.

Tip #1: To launch a project, include in the first meeting someone who knows the data and will consume the analysis. Sit down with the analyst, explain the context, and go into depth with the questions.

What's your intent? How did you form the questions? The better the analyst understands your situation, the better he or she can provide actionable results.

"I want to work with the people who live and breathe this data, I want to talk to the audience," said Joe Mako, an analyst at Cincinnati-based S2.

Tip #2: Ask honest questions and don't micromanage.

"There's always someone who seems to think they know what you ought to be doing and how you ought to be doing it," said Doyon. One client, for example, asked her to segment his customer base and was unhappy when she found different segments than he'd decided on beforehand. "What was in [his] head didn't exist," she recalled.

Given enough depth and room to roam, a good analyst can find the data's story and spin it for insight and memorability. "People are so used to the idea that analysts are cold and analytical," said Doyon. They don't recognize one mark of superior analysts: they're storytellers. "Data is their paint."

Tip #3: Explain the data's lineage.

What's been done to the data? Is it raw, or has it been transformed in some way, and how? Explain business rules. One analyst, who asked for anonymity, recalls discovering an apparently deliberate misspelling of Connecticut in a large set of transactions. Why? In another project, one of a retail store's cash registers showed zero sales on every third week. Often only the client knows the significance. Such puzzles should be explained as much as possible at the start.

Tip #4: Allow for failure.

Analysts try one approach then another -- and sometimes they miss. If failure's not allowed, they play it safe and paths go unexplored. Traditional business culture has little appetite for "test and learn" and usually opts for "Fail and get out!"

Tip #5: Respect their time.

Because analysts' work is often not well understood within organizations, many seem to assume it's free. Save everyone's time by learning basic terminology before you meet to hear results. Little else kills a meeting more than letting one of 11 people in the room stall for 10 minutes to understand a term such as "KPI."

Tip #6: Change priorities judiciously.

Analysis, like most projects, runs better when there's a long roadmap laid out at the start.

Tip #7: Learn to perform data analysis yourself.

You can learn how analysts think. You might even start performing some analysis of your own with the new generation of easy-to-use desktop tools.

Bonus Tip: If somehow you've started off on the wrong foot with a data analyst, there's always Ben and Jerry's ice cream -- or beer.

Ted Cuzzillo is a journalist and industry analyst focused on analysts' tools and needs as well as the environments in which they work. You can contact him directly at [email protected]. If you're a data analyst, he'd appreciate your participation in his survey; you'll receive a free preview of his report when it's complete.