Hello and welcome to the latest issue of Data Patterns. If this is your first one, you’ll find all previous issues in the archive.
As I mentioned on a post on LinkedIn, I’m kicking off a series of interviews (feel free to call it a podcast) with business operators, executives and data professionals on the topic of being data driven.
By the way if you’re interested in participating or know someone who’d be great for me to interview, just reply to this email and I’ll take it from there.
But why? Hasn’t this topic been already run into the ground?
Yes we’ve talked about it ad-nauseam but it always leads to more confusion. Before you know it, the conversation goes from “data driven” to “data informed” and the discussion starts going in circles.
I want to go straight to the source.
By talking to industry professionals, especially business operators (CEOs, CFOs, COOs, EVPs, etc.) who make the day to day decisions and long term planning, I hope to uncover useful patterns I can share.
To that end, I present you with the first interview in the series with Chitrang Davè, a seasoned data leader in the life sciences and health care fields.
Below are some of the most interesting tidbits from the interview as transcribed by Otter (so forgive the raw looking format).
We get into:
What does it mean to be a data driven operator?
Examples of how data was used to improve operational performance
Helping sales reps meet their quota while showing them exactly how it impacts the bottom line
Improving the quality of medical devices and reducing scrap through image processing and CV
How data is used to drive operations at the executive level
Challenges to becoming even more data driven
I’m still workin on getting the recording up, but wanted to share some of what we talked about right away.
Let’s get into it!
What does it mean to be a data driven operator?
Chitrang: It means means different things to different people. At the core, most businesses and most business operators [already] operate that way. If you talk to any good operator, you get a pretty good sense of this. They’re intimately aware of the drivers, they know what numbers matter, they know what the size of the market looks like, what their revenue targets look like, what their internal drivers look like. So in general, these operators, the very good ones are really data driven.
We all know that business operators use numbers (aka data) to make decisions; numbers like (revenue, run rate, profit margin, etc.) Many of us in the industry are also familiar with the quarterly review and planning process.
Here’s Chitrang again:
When, when it comes to how we enable some of this, from where we sit as data practitioners, I think that's where the challenges are is is finding and understanding getting into the minds of these operators understanding what those are, and finding the corresponding sources that we can derive from and the transformation that we need to run and really putting that in a consumable digestible format, outside of their just their heads and in, you know, making it more widely available.
The same can be applied to individual departments. You can assume executives know what they’re doing and what numbers they want to see and work to enable that. My professional experience bears this same exact resemblance.
This is good but I think we can do even better, but that’s a discussion for another day.
Back to the interview.
In your experience, have you seen any examples of how analytics helps drive a process like Amazon’s Weekly Business Review?
Chitrang: As you know, I've been involved in, preparing for these reviews. And, and in automating where possible. So there was there was a time when it would take a team of analysts a couple of weeks to prepare for a quarterly review, because they're collecting information from a dozen sources and putting it in PowerPoint decks, etc.
And, and we are at a point where all of that can can really be automated and review can be live with live data. So you you're pulling up live scorecards and live dashboards and and focusing your effort on the system telling you what are the exceptions are. So this is, you know, obviously majority varies by business and by function. But, but that's where we want to be headed to have those life. reviews in life, scorecards.
I remember these reviews very well. When I was a business analyst, I was also involved in “pulling numbers,” creating charts and dashboards in order to help executives craft a narrative for the quarterly board meeting.
These usually involved things like current performance and future plans. It was the most stressful time of the year for me because the metrics and charts would sometimes change quarter to quarter, so I had to write SQL from scratch each time.
This is the main reason I’m working on SOMA. I want to remove as much of this burden from the hands of analysts as possible so they can focus on things that will really impact the business.
What are some specific examples of using data to improve operations?
Example 1: Sales rep visiting a hospital to sell a medical device
We discussed how a sales rep can meet their quota by determining which clients they might need to visit in order to get the highest likelihood of a sale. This is often a simple process of segmentation, one of the earliest uses of analytics in business, but still quite valid.
Chitrang: Yeah, so let's dive into some of the specifics. I think the, the one thing I was thinking about was as a, let's say, I'm a rep in sales in field, I'm a field rep in a sales organization, I'm visiting customers, and I care about what are my incentives, and I care about what drives those, those, my comp and my benefits, so I am looking at what my targets are, you know, what my year to date, quarter to date numbers look like?
So those are some of the things that drive me I know what I'm being measured on. And, you know, I'm going to pay attention to those. But from the from a business, from a sales leadership business perspective. They want to drive my focus my effort on things that are most valuable to the overall business. And sometimes they may not align sometimes not.
Chitrang is raising an important point here. If we want the entire organization to be data driven, then we must define the right metrics and set the right incentives at an individual level so that they lever up to the key output metrics operators care about.
This way, everyone can see exactly how their performance impacts the organization’s objectives.
Yes, I know, I care about where I am percent to plan for my quarter. But tell me how I can close that, how can I can meet that [quota] and how I can beat that [quota]. So in that case, in this example, it's specifically about all the hospitals in my territory, these are the three I need to go check on because the data is telling me something that is amiss
I asked here about specific metrics the rep might use to find which hospitals to visit.
Without getting too specific, I already touched on kind of what I was referring to is, you know, what do we know about the caseload or or case mix at a particular hospital because I see the because CMS or the Medicare makes that available publicly so we can go out and look at what the case mix looks like and does that match with what our business does with this hospital.
Example 2: Medical devices manufacturing.
Another example we talked about was in the manufacturing of medical devices. The key performance indicator here is high quality / low scrap rate.
If a device doesn’t meet certain quality standards it’s scrapped. So where analytics comes in is in figuring out as early as possible if what you see in step x will eventually lead to scrap three steps later.
Chitrang: So so it's really important to know where we are trending as early as we can and, and identify the where stuff fail. So in this in this example, knowing that having on step two, for example, something that is in or out of threshold, will then three steps later, resulting in strapping the eyes and wrap is much more important.
As you transform raw materials into parts by milling, drilling, welding, sanding, cutting, etc. it’s important to capture key quality metrics at every step.
For example looking at parts under a microscope by a human to verify welds, which could later be automated by using hi-res image processing through deep learning.
So quality checks, just automating those and, and some of these are tests you can run, but then there are others where you actually have humans, looking under a microscope to see whether the welds look, right, or the placement of the stuff on the board looks right, these are all in the past human stuff. But now we can just take high speed, high resolution images, throw it against the model, and within a fraction of a second get a yay or nay or a probability score that then you can only you can you can set a threshold or do a human.
What are some of the challenges to being more data driven?
Chitrang: There are so many, and starting with the data, you know, the biggest thing I keep hearing is, I don't have good quality data, or I don't, or I spend too much time getting to a data that I can then use preparing the data. Again, comparing cleaning, cleansing, transforming this that takes too much time. And, and Central or, like the ones I have led sometimes are not able to meet demand. And so how do we enable businesses to do these things more effectively, as a challenge for leaders like mine? Like, like I have to deal with?
Data quality is still one of the biggest challenges in driving adoption of metric review processes going from a quarterly cadence to a monthly or even weekly cadence.
This is another problem that gets solved once you adopt a standard like SOMA for your metrics. SOMA dictates exactly which key business activities need to be instrumented in order to calculate each metric.
That’s it for now. Until next time.