“You answered the question correctly, but you lost the sale.”
I’m 26 years old and I’ve been working at Tableau for approximately 6 months. I’m on the phone with Marc Rueter who is dialing in from an undisclosed location (i.e. across the hall) in the Lakeview building, Fremont, Washington, the 4th floor of which houses the entirety of Tableau Software, Inc. presently in the summer of 2011.
Those of us who worked at Tableau at this time, particularly the technical experts, consider Marc Rueter to be something of a god. He designed the Product Consultant role that was my entry-level job at the firm, and he defined the strategy behind all of Tableau’s training, pre-sales consulting, services delivery, and more as the company emerged as a startup.
Marc interviewed me, hired me, mentored me, and now was proctoring my Gold Certification, the highest level of internal product certification Tableau offers. I had invested weeks into preparing for the technical questions he would ask, which started from a common list of prepared challenges but quickly spun into off-the-rails and improvised challenges designed to hurt your confidence and test your ability to adapt when dealing with a particularly difficult customer.
Sometimes the questions bordered on the existential – after asking you about a particularly difficult Table Calculation, Marc might ask, “What else can I do with this?” at which point you were expected to demonstrate some new, simple, beautiful data insight, previously unprepared for. You were supposed to speak the language of data.
Through working with Marc, through getting to know my more-experienced colleagues (the Wilson Pos and Jesse Gebhardts of the world, among others), and through being thrust constantly into situations I wasn’t prepared for, I became one of the leading Tableau users in the world. Here are the lessons I learned.
I interpreted Marc’s questions and reactions to my responses with the attention to detail of Stanley Tucci’s Nigel interpreting Meryl Streep’s Miranda in the film The Devil Wears Prada, in which every smile, lip purse, and glance is a reaction to be recognized and responded to in accordance with protocol. Looking back, it was as much of a psychological exam as it was a technical one.
In one portion of the certification process, I was asked about how Tableau compares to Cognos. “We’re currently using Cognos for B.I. Why should we switch to Tableau?” I had been coached about customers like these – I.T. departments who had made substantial investments in the Big B.I. evil empires. Remember that at this time Tableau had around $30M in annual revenue and had finally talked itself into being recognized as a market challenger in Gartner and other analyst reports. We were looking for passionate customers, but we didn’t want to ruffle too many feathers.
I launched into a talk-track about how Tableau can often sit beside traditional B.I. standards rather than replace them, doing things that Cognos and others can’t do like provide fast, interactive visuals while not disrupting the operational reporting those platforms are good at. I felt pleased with my answer.
“Do you want to ask me if I like Cognos?”
I knew where this was going, but I bit anyway. “How do you like Cognos?”
“We hate it.”
Listen. Don’t assume you’re smarter than your customer. Try to speak to the things that they care about. And if they hate Cognos, feel free to trash it.
Dustin Smith was asked to build a control chart with certain data points highlighted based on numbers of standard deviations from the mean. When he hesitated, Marc said, “I’m going to go get a margarita and I’ll be back in an hour. If you’ve finished by then, I’ll pass you.”
In my exam, I was asked to build a Pareto chart. This was an easy one. I began, “there’s two ways, a quick way, and a…” Marc interrupted me.
“Stop giving me options. You’re a technical salesperson. Tell me how to do it.”
But when I did tell him how to do it, there was always a wrinkle. There are multiple ways to solve most data problems, and Marc prides himself on knowing all of them.
Today, I am Marc to a small portion of the company. I defend the vision and purpose of Tableau. I train and mentor new consultants. And I proctor Gold Certifications.
One of my favorite things to ask when I perform the Gold exam is “How do I know when it’s time to expand my Tableau Server?”
This question can go a number of different directions. It can be a discussion about hardware, or about the number of users you can support with Tableau. It can also be about monitoring usage statistics and the performance of different visualizations. It can lead to conversations about distributed architecture or network latency or even data strategy. The people who perform best in this part of the test are the ones who clarify.
But a decent percentage of the people who sit for my exam immediately start talking about one of the aforementioned topics. Usually I take that as a signal to tell them that my question was really about one of the other topics. The Gold exam is a training exercise as well as a test. You can never be fully prepared for a test that, by definition, is intended to see how you perform when you aren’t prepared.
And sometimes, I use Marc’s line. “You answered correctly. But you lost the deal.”
Show the simplest solution
Marc’s favorite question was this one: using Tableau Superstore data set, I want to see a list of customers who have purchased both Furniture and Office Supplies.
This is trickier than it sounds. If you are a Tableau user, go ahead and try it right now. The issue is when you filter to multiple members of a single dimension in Tableau, the WHERE clause Tableau generates is “OR” not “AND.” AND is tricky.
There are a few ways to solve it. One involves using sets to select lists of customers who meet both criteria, then showing the intersection between them (Tableau didn’t have combined sets when I took the exam in 2011, which made this option slightly more difficult). Most people arrive at this answer pretty quickly. It’s easy to implement, but it has one flaw. If you were taking the test with Marc, he’d likely sit and watch you do it, and then ask, “what’s wrong with this approach?”
The problem is that the answer you get, while correct, is static. This is a common challenge with data – how do I keep from having to refresh this report every time my data is updated? By selecting a list of customers who have purchased certain items today, we exclude returning customers who might in the future purchase items that fall into those same categories. We also eliminate the ability to look at new customers. A refresh isn’t difficult, but that’s not the point.
By the time I took this exam there had been 10 or 12 other people who had taken it, and the question was a bit of a legend. I knew the second response – by creating a calculated filter within a specific dimension, you filter the contents of that dimension but nothing else. It’s a little-used feature in Tableau because it’s hidden in a dialog most users ignore. The calculated filter includes some syntax that may be difficult to understand for newer users, but it was easy enough to memorize. It reads like this:
MAX(IF Category=’Furniture’ THEN 1 END) + MAX(IF Category=’Office Supplies’ THEN 1 END) = 2
(For style points, a testee could use the IIF(test, true, false, result) syntax rather than the standard VBA of IF THEN ELSE END, both of which seem to exist in Tableau mostly to cause arguments about which one is best. For the record, they’re the same).
Know the best solution
In my exam, I was ready for Marc. I showed the calculated logic first. I could almost hear him smile to himself through the phone. This motherfucker thinks he’s smart.
“But I don’t want to have to write a calculation to do this.”
Ok, I said, and I showed him the combined set logic.
“But that way isn’t dynamic.”
I paused. This was the moment Marc was waiting for. This is the entire point of Tableau’s certification program. If you are going to sell Tableau, if you’re going to consult with users about how to solve their business problems, if you’re going to present yourself as a technical expert, you aren’t going to be prepared for every question. There isn’t a script for every data challenge ever written.
“Charles, I know you know how to answer it the way it’s written in the prep materials. That’s why I’m asking you to do it a different way. No calcs. But dynamic.”
An expert Tableau user doesn’t use a script. He or she listens to questions, thinks about them, and then recommends a solution.
Data analysis is a language. It is a language that encourages dialog, listening, and interpretation. Andy Cotgreave comments on this in the podcast we recorded together. If there was a formula for how to solve every data problem, solutions would lack the beauty and engagement that forces people to think about what they’re communicating. This principle applies to the mathematical and logical formulae that underlay presentations of data, not just the visual results.
- Listen – Try to understand your customer. Don’t assume you are smarter than them.
- Have confidence – Be assured that you are an expert. You may not be smarter than your customer, but you know more about Tableau than they do.
- Clarify – If you’re unsure about what you’re being asked, clarify until you are sure.
- Show the simplest solution – This helps the customer understand that the answer to their question isn’t a holy grail or a skill that will require months of training. It presents Tableau as a set of tools, not an answer to their one question.
- Know the best solution – Sometimes the simplest way isn’t the best. Be prepared to handle the follow-up question.
*One caveat: “Show” doesn’t always mean “demonstrate.” Sometimes it means “send documentation.” Other times it just means “Say ‘Yes.’”
Marc likes to challenge people. He likes to prove that he is smarter than you. He likes that he’s seen as an expert. And he deservedly takes credit for building and defining Tableau’s presales organization.
But he’s not a dick. He always shows you the right answer afterward.
Until next time, may all of your simplest solutions also be the best.