Visual Introduction to Machine Learning

I came across a fantastic Visual Introduction to Machine Learning at  Not only is this a great primer on ML, but the visual presentation is one of the best I’ve ever seen.  Kudos to r2d3!



Healthcare Predictions, 2016 Edition

It’s the end of the year and everyone is starting their predictions for the new year.  Here are my top 3 predictions for Healthcare.

1. Telehealth Dominates

Telehealth had a great year in 2015.  From an upstart technology in 2014, 2015 showed real usage and real numbers.  There are lots of pilot projects that have shown that you can receive better and cheaper care, especially in rural settings.

The break out for 2016 will be larger health systems moving from thinking about it to doing something about it.  Current challenges around communication infrastructure will be solved by large players like Verizon, making the reality of full stream video a reality, no matter where you live.

2. Analytics Breaks the Top 3

It’s well known in the software and hardware world that if you’re not in the top 3 problems that the C-Suite is focused on, you’re probably not going to sell much.  I think that 2016 is going to be the year that analytics becomes the #3 priority for Health System CIO’s (behind security and IoT).

Up to this point, there has been a lot of talk about how analytics can solve problems, ranging from Readmissions to Improved Patient Flow, but there haven’t been enough success stories to show a big financial benefit.   This is the year that plucky upstarts in Health IT storm the castle and show Healthcare what the rest of the world has already learned about unlocking big data.

3. Device Hacking Becomes All Too Real

There has been plenty of press over the last couple of years about how insecure biotech devices are.  Whether it’s pacemakers or blood pressure monitors, the hospital setting is ripe for full scale attacks.   Up to this point, most security issues have been proof of concept.  As health care related data becomes more valuable for criminals they will continue to seek attack vectors that are easy to penetrate.  I think that 2016 is the year that we see the first real attack on a health system that stems from insecure, networked devices in the ambulatory environment.

We’ll recap at the end of 2016 and see how close we hit the mark.

Minimum Explainable Product

I was having coffee with my good friend Jonathon Morgan, CEO of Popily the other day and got talking about fund raising.  I had mentioned to Jonathon that the marketing on the Popily had changed quite a bit since the last time we had talked in August, and was in fact, very similar to someone he had identified as an early competitor over the summer. It ends up that Jonathon’s pitch was getting hung up on how he pitched the service.

Earlier in the year, Jonathon and team had settled on marketing Popily as a data discovery tool to complement your expensive BI tools.  You’d use Tableau to find the answer to the question you already knew to ask, something the Data Scientist did.  You would use Popily to help you figure out what questions you could ask about your data.

Over the course of an hour, we ended up talking a lot about a concept that I call ‘Minimum Explainable Product’. Much like an elevator pitch, the MEP is that bare minimum that it takes to explain what your product or service does.  Unlike an elevator pitch, however, it’s purely focused on features, and doesn’t include value.

My product at LeadingReach has a lot of features, from workflow to referral management to patient engagement, but when I’m talking to doctors, my MEP is very simple, “We replace the fax machine for referrals and we keep patients from calling your office for driving directions and appointment confirmations.”  That’s it.  It’s 5% of the product but summarizes the two biggest features.

Much like Minimum Viable Product (MVP) and Minimum Sellable Product (MSP), this approach works to really reduce what you do to its simplest forms.

For Popily, explaining how their product reduces the need to have a data scientist was too complicated and complicated VC discussions.  Switching to “beautiful charts from your spreadsheets and data” is easy for everyone to understand. After all, anyone that uses excel understands the value of beautiful charts.


What Keeps Healthcare CIOs up at Night

Yesterday, I attended the December Austin HIMSS meeting featuring a distinguished panel of four area Healthcare CIOs: John Mason of Hill Country Memorial (formerly HCA – St. Davids, Mike Minx from Seton Healthcare (Ascension), Matt Chambers from Baylor, Scott & White and Bill Philips from University Health Systems, San Antonio.

This was a great CIO panel and really highlighted some of the concerns and thoughts of healthcare technology leaders.  I thought it was worth recapping some of the more interesting discussions.

Security, Security, Security

Far and away the most talked about topic at the meeting was security.  Every CIO expressed various levels of concern with regard to hospital and healthcare security in general. Five years ago, HIPAA and data breaches were the number one concern of CIO’s, but today’s challenges are morphing along with technology and a connected world.

I found most interesting a comment by Mike Minx regarding small scale devices.  Mike’s chief concern was how do you secure medical devices, such as pacemakers or insulin pumps. It’s not new that these types of devices have proven to be easily hackable.  In a hospital settings, Mike and other participants wondered how CIOs can manage the security complexity of 1000’s of devices, connected or otherwise, inside the hospital setting.


Matt Chambers talked about the difficulty in finding qualified technical personnel.  Healthcare isn’t a “glamorous or sexy industry” and “doesn’t pay as well as technology giants like Google or Apple that we compete with for headcount.”  Matt said that the Dallas IT sector has over 40,000 open IT positions at any given time.

Mike Minx mentioned that Seton has started an apprentice program to hire two year graduates and train them in specific technology disciplines. Bill Philips mentioned that they are partnering with San Antonio to open technology high schools to help with the problem.

If It’s Connected, It’s My Problem

The panel ended with discussion around the changing role of the Healthcare CIO. Each participant talked about how more connected devices, and more devices in general have morphed their role from a straight IT leader, into that of the digital strategist.


Overall, a great panel discussion.  CIO’s in healthcare face the same challenges that I have seen in other industries, security always tops the list.  An increasingly connected hospital AND patients is generating new challenges and opportunities for everyone.  I’m excited to see how things change at next years session.

How Much to Charge for your Product

I’ve mentioned in past posts the challenge of determining product value and pricing.  I recently came across an article on Saastr (Next Time – Ask for $1,000,000. A Year. I’m Not Kidding) and it got me thinking again, not only about how you price a product, but different ways to continuously validate pricing.

Jason Lemkin, the author, starts out with a pretty simple premise:

The first is to challenge them to double their highest ACV ever on the next similar prospect.  If your highest paying customer pays you $30,000 a year … even if all the rest pay you only $3,600 a year (a not uncommon scenario) … then why can’t you get $60,000 a year from the next Big Prospect that come in the door?

He then goes on to describe how he encourages all start up founders and sales people to push the envelope for pricing.  I don’t disagree with anything he mentions in this article except for one thing:

Now, if your biggest customer pays you $60,000 today … asking for $500,000 a year, or even $1m, may sound nuts.

But turns out it isn’t.

Do it when a prospect asks you for something that pushes your solution.  When it’s clear they want to buy, but they want the product to go further than it does today.

“Well, we don’t do that today.  But for $700,000 a year I can commit to having that by Q2’16.”

As a Product Management executive, I absolutely hate the execution of this tactic.  In theory, this is a great way to a) determine what features your product needs for this class/type of customer and b) help determine the price point for it.  In reality, however, what usually happens is you train your sales force to sell any feature request for any price point.

The most dangerous part of this tactic is when you have 10 or more customers that have been promised 10 or more different features and already paid for them.  I have both worked for and witnessed many a startup fail because of this.

As a Product Management executive I proactively coach our sales staff to bring me in early to conversations where they think we can get higher price points for legitimate features.  I encourage them to continually prod prospects, but I am adamant that I am always the one that maintains and more importantly commits to the delivery of features and time frames.

I’m always looking for increased price points, but mitigate the danger to your roadmap and your customer base by making sure everyone understands that it’s not a feature free for all.

Teaching Your Kids to Code

LifeHacker has a story today about an interactive programming tool developed by  It’s a really great idea and a visual way to teach kids how to code.

In a genius move, licensed MineCraft and Star Wars as the two sandbox environments that your kids learn in.  I remember doing some similar 30 years ago in Turtle.

Highly recommended.

Confusing Price with Pricing

I recently came across a really good article by Ronen Nir, Why Startups Shouldn’t confuse “Price” with “Pricing” (and how to tell them apart), in it, Ronen does a great job of laying out the difference between  Price versus Pricing.

Price represents the value that the product or service brings to the customer while pricing is the method by which the company determines the price.

While I dislike Ronen referring to this as Price instead of just sticking with product value, the definition is spot on and a good launch point for understanding how you set product pricing.

I mentor a lot of very early stage startups, usually those that are a single found with an idea, or a pre-alpha product.  Pricing (or product value) v Price can be a confusing concept to impart on someone who doesn’t have a Product Management or Sales background and more importantly, has not actually gone through the process of price discovery.

I have found that the best way to determine price/product value is to spend a lot of time with customers or prospects and document their current environment as much as you can, and then attribute cost structures to it.

As an example, when determining the price for my current product line at LeadingReach, I literally sat at doctors offices for hours at a time with a stop watch and documented every single system they interacted with and how much time (to the second) they spent interacting with it.  Included in this “Time and Motion” study was the time they moved between systems (walking to the printer for a demographic sheet, walking to the fax machine, etc).

After I had collected all of this data, I built a flow chart of what the typical process looked like and color coded each system.  Additional research gave me a good indication of price per minute for various systems.  Adding all of that up, I could very confidently address how much money was spent on this process today.

If my product was 40% more efficient at this process, I now know the price (product value), or “the value that the product or service brings to the customer”.  From here I  worked with our sales team to determine our pricing strategy based on a multitude of factors (Saas v On-Premise, Size of Clinic).

Price (product value) is not a difficult metric to establish, the bigger problem is that most people don’t have experience in determining it.  Time and Motion studies are a great way to determine not only process but other features that your product should include.