On Episode 12 of Excel TV I said a little about a trainee who started crying as I was teaching her the details of her new job.crying-kiwa007

This story comes up in conversations around the challenges of working with data. This particular challenge was a complete data stewardship bull ride, and it’s great if you like riding bulls.

NOTE: I purposefully resisted calling it a nightmare because it’s important to distinguish between

  1. Something that no one likes: nightmares
  2. Something terrifying, with real risk, that some people love: bullriding

Working with data is more like bullriding.


She, Rhonda, was taking over a role that involved exam completions and professional certifications. She would be the person who would determine who completed required coursework in the previous month, then officially notify each person’s employer and send out a gold pin for each new designee.



There was a report that was developed for this purpose. It was designed to list new designees, their completion dates, their employers, and where to mail the pins. BOOM! DONE!

But no. The report wasn’t good at handling odd situations.

RULE: From the date that a candidate orders their first course, they have 3 years to finish.

ODD SITUATION 1: Tony started his coursework under Employer A in 2010 and finished while working for Employer B in 2011. The report treated Tony like 2 different Tonys who were both incomplete.

ODD SITUATION 2: Courtney started in 2003 and quit. She started all over again in 2007 and completed in 2009. The report saw her as a 6-yr foot-dragger and therefore ineligible.

These are serious issues in data stewardship: knowing where your reports fail, where they’re strong, and if there are satisfactory workarounds to add to the process.

Loose Definitions & Extenuating Circumstances

3 years to complete the curriculum, you say?

Hmmm … what about

  • 3 years + 1 day? Sure
  • 3 years + 6 months? Probably not
  • 3 years + 2 months and that included 6 months where the candidate did nothing but rehab after a car accident? Probably
  • 3 years + 3 months due to having to retake an exam? Maybe, but probably not
  • 4 years? No

These situations constantly had the process owner reviewing situations, making decisions and overriding the report.

How Bad Was The Report, Really?

Data is never ever perfect. But when the stakes are high, you have less room for error. We were looking at roughly 5 people out of 45 people each month having something wrong. That’s an 11.11% error rate. That’s too high because people would call and complain. Each complaint had to be researched, and if the complaint was legit, we’d send out that gold pin overnight, and we’d cover the cost. A monthly thrilling bull ride degraded to a daily annoyance.

But why not develop a brand new report? Because it’s an expensive and time-consuming process. Consider that this was only a monthly process, and there were so many scenarios that had to be built into the report. It could easily be 6 to 12 months of having the outside consultant developing, testing and tweaking a brand new report that impacted less than 1% of the entire customer base.

So, no. Developing a new report wasn’t an option. Neither was it an option to use the existing report.


rodeo bullLet’s be clear. I’m not making fun of anyone. This was some hard stuff, and not everyone wants to find themselves on the back of a 1700-lb bull. The sight and smell of the real live beast is when the tears started.

Rhonda didn’t want this role; data wasn’t her thing, but she’d been made the chosen one. So, those weren’t wimp tears, they were terror tears.

Years before turning this role over to Rhonda, we’d decided that the existing report was useless. So, the process always started with a raw 60,000-row data dump. Anyone who’d ever taken one of the courses was in that spreadsheet. We’d use Excel to whittle down to the 45-ish newest designees.

And it was difficult because we had to think through all the criteria that would have someone on the final list or off the final list


  • Completed all 4 courses within 3 years or, a reasonable time afterward.
  • Started 10 years ago. Started again 2 years ago and completed.
  • Michael Carter and Michael Carter, after verifying they’re father and son.

  • Completed unreasonably beyond the 3-yr mark.
  • Took all 4 courses but failed one of the final exams, and they haven’t even ordered the retake exam.
  • Completed at the early part of this month and will be in next month’s process.
  • Monroe Abramson because we received email saying that he resigned soon after completing the program.
I was sitting behind Rhonda as we went through the process. looking over her shoulder at her computer screen. I would occasionally ask if she understood where we were in the process, and she’d cautiously say, “no.” Fair enough. So, I would explain, remembering that it took me several months to internalize that process. This was her first time.

Her supervisor came up behind us and asked, “so, how’s it goin’?” And that’s when Rhonda turned around and I saw the tears rolling. OMG! I had no clue.

The supervisor was shocked, and said, “please excuse us, Oz. I think she’d like to talk with me.”


There are people who understand data, and people who don’t. There are people who love analyzing data–as long as it’s clean. And there are folks who’ll work through the mess. Rhonda, absolutely did not want that role, starting with messy data and slowly removing people until she had a perfect final list. Without a reliable report, we had to think of things like:

You don’t hear an analyst whenever you’re interviewing them. You see them. You tell by the way their face lights up.

Rick Grantham, Excel TV

Why do we take the original list and get rid of the orders that were from more than 4 years ago?


  1. It minimizes the distractions of people who started a long time ago and started again.
  2. At the end we’ll get to visually inspect for people who took a little longer than the 3 years; make a judgement call about completions that are on the cusp of “reasonable”, and alert others that they need to order retake exams.
  3. We reduce our workspace clutter by sifting through just the data that’s only 4 years old, it knocks our list down from 60K rows to maybe 7K.

The new designees want to be part of the monthly ceremony and show off their new gold pins in front of their friends, families, colleagues and haters. They want the career boosts that are associated. And when someone is left off of that list, here comes another phone call that starts with, “YOU PEOPLE … !”


So, all of this goes way beyond Excel and fancy new Office 365 features. It goes beyond programming in R, writing SQL queries, and getting unpaid interns to do grunt work. I agree with Rick, during the Excel TV conversation when we discuss picking analysts by those who light up around conversations about data. They can be taught the tools and methods if they don’t have them.

We’ve got to identify and nurture the people who want to be up on the back of a bull and have some fundamental wherewithal for being successful.

bucking bull

Cry, Don’t Worry About It courtesy of kiwa007
bull photo credit: Base Camp Baker via photopin cc
bull and rider photo credit: dview.us via photopin cc